diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,38998 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9993073193257906,
+  "eval_steps": 500,
+  "global_step": 6495,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.7414,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.629043195856986e-06,
+      "loss": 0.6127,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1669348782094284e-06,
+      "loss": 0.6269,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.258086391713972e-06,
+      "loss": 0.668,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.104449259132793e-06,
+      "loss": 0.5709,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.795978074066414e-06,
+      "loss": 0.5347,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.380657356191162e-06,
+      "loss": 0.5793,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.887129587570957e-06,
+      "loss": 0.5413,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333869756418857e-06,
+      "loss": 0.5794,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.73349245498978e-06,
+      "loss": 0.5366,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.094995158510905e-06,
+      "loss": 0.5449,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.4250212699234e-06,
+      "loss": 0.5436,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.72861586265778e-06,
+      "loss": 0.5511,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0009700552048147e-05,
+      "loss": 0.5267,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0271384137342222e-05,
+      "loss": 0.5591,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0516172783427943e-05,
+      "loss": 0.5587,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0746116371107466e-05,
+      "loss": 0.5906,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0962912952275841e-05,
+      "loss": 0.5172,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1167984925712543e-05,
+      "loss": 0.5964,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1362535650846765e-05,
+      "loss": 0.556,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.154759223440059e-05,
+      "loss": 0.558,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1724038354367891e-05,
+      "loss": 0.5231,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1892639781609205e-05,
+      "loss": 0.5243,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2054064465780386e-05,
+      "loss": 0.5643,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2208898518265586e-05,
+      "loss": 0.5798,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2357659058514765e-05,
+      "loss": 0.5157,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2500804634628285e-05,
+      "loss": 0.5638,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2638743747905133e-05,
+      "loss": 0.559,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2771841880842693e-05,
+      "loss": 0.5117,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2900427333199209e-05,
+      "loss": 0.5411,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3024796100762397e-05,
+      "loss": 0.5403,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.314521597928493e-05,
+      "loss": 0.534,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3261930036720335e-05,
+      "loss": 0.5453,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3375159566964452e-05,
+      "loss": 0.5023,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3485106615323954e-05,
+      "loss": 0.5701,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3591956148132828e-05,
+      "loss": 0.5227,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3695877925041064e-05,
+      "loss": 0.5402,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3797028121569526e-05,
+      "loss": 0.5549,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3895550740867208e-05,
+      "loss": 0.5338,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3991578846703751e-05,
+      "loss": 0.5595,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4085235644191126e-05,
+      "loss": 0.5634,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4176635430257577e-05,
+      "loss": 0.5429,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4265884432268126e-05,
+      "loss": 0.5567,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4353081550224877e-05,
+      "loss": 0.5341,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.443831901555165e-05,
+      "loss": 0.5465,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4521682977466191e-05,
+      "loss": 0.5487,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.460325402628616e-05,
+      "loss": 0.5191,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4683107661637372e-05,
+      "loss": 0.5485,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4761314712382324e-05,
+      "loss": 0.5462,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4837941714122573e-05,
+      "loss": 0.5508,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4913051249316895e-05,
+      "loss": 0.5508,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4986702254371752e-05,
+      "loss": 0.5285,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.505895029747943e-05,
+      "loss": 0.5524,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5129847830485272e-05,
+      "loss": 0.5748,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5199444417643699e-05,
+      "loss": 0.5256,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5267786943762122e-05,
+      "loss": 0.5548,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5334919803921972e-05,
+      "loss": 0.5152,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5400885076699676e-05,
+      "loss": 0.5235,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.546572268258014e-05,
+      "loss": 0.5543,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.552947052905619e-05,
+      "loss": 0.5552,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5592164643734683e-05,
+      "loss": 0.5867,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5653839296619384e-05,
+      "loss": 0.5586,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.571452711261002e-05,
+      "loss": 0.5198,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5774259175141914e-05,
+      "loss": 0.5515,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.583306512179057e-05,
+      "loss": 0.5585,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5890973232577318e-05,
+      "loss": 0.5637,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5948010511634617e-05,
+      "loss": 0.5595,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.600420276282144e-05,
+      "loss": 0.53,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6059574659818635e-05,
+      "loss": 0.5199,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6114149811180942e-05,
+      "loss": 0.5646,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.616795082077488e-05,
+      "loss": 0.5285,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6220999343989814e-05,
+      "loss": 0.5141,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6273316140071974e-05,
+      "loss": 0.5378,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.632492112089805e-05,
+      "loss": 0.5332,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6375833396475015e-05,
+      "loss": 0.5384,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6426071317426514e-05,
+      "loss": 0.5382,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.647565251470207e-05,
+      "loss": 0.5583,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6524593936724194e-05,
+      "loss": 0.5608,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6572911884169272e-05,
+      "loss": 0.5147,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6620622042560734e-05,
+      "loss": 0.4975,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6667739512837714e-05,
+      "loss": 0.5382,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6714278840048114e-05,
+      "loss": 0.5506,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.676025404030262e-05,
+      "loss": 0.5459,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.680567862611456e-05,
+      "loss": 0.5484,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.685056563024026e-05,
+      "loss": 0.5492,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6894927628125112e-05,
+      "loss": 0.5357,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6938776759052122e-05,
+      "loss": 0.5336,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6982124746081863e-05,
+      "loss": 0.5592,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.702498291486582e-05,
+      "loss": 0.5068,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7067362211408637e-05,
+      "loss": 0.567,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.710927321884894e-05,
+      "loss": 0.5325,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7150726173323178e-05,
+      "loss": 0.5324,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.719173097897183e-05,
+      "loss": 0.519,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7232297222143145e-05,
+      "loss": 0.4775,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7272434184845334e-05,
+      "loss": 0.524,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7312150857494356e-05,
+      "loss": 0.546,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.735145595100121e-05,
+      "loss": 0.5072,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7390357908239312e-05,
+      "loss": 0.5349,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7428864914929763e-05,
+      "loss": 0.5509,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.746698490997956e-05,
+      "loss": 0.5344,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.750472559530546e-05,
+      "loss": 0.5365,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.754209444517388e-05,
+      "loss": 0.5241,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7579098715085187e-05,
+      "loss": 0.5399,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7615745450228736e-05,
+      "loss": 0.5328,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7652041493533384e-05,
+      "loss": 0.5362,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7687993493336415e-05,
+      "loss": 0.5452,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7723607910692433e-05,
+      "loss": 0.5678,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7758891026342256e-05,
+      "loss": 0.5423,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7793848947360683e-05,
+      "loss": 0.5451,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7828487613500683e-05,
+      "loss": 0.5609,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7862812803250492e-05,
+      "loss": 0.5076,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7896830139619104e-05,
+      "loss": 0.5363,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7930545095664606e-05,
+      "loss": 0.5401,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7963962999778956e-05,
+      "loss": 0.5547,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7997089040742e-05,
+      "loss": 0.534,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8029928272556664e-05,
+      "loss": 0.5572,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8062485619076636e-05,
+      "loss": 0.5501,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8094765878437127e-05,
+      "loss": 0.514,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.812677372729863e-05,
+      "loss": 0.5623,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.815851372491318e-05,
+      "loss": 0.5578,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.818999031702181e-05,
+      "loss": 0.5416,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8221207839591668e-05,
+      "loss": 0.5506,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8252170522400552e-05,
+      "loss": 0.5694,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.828288249247637e-05,
+      "loss": 0.5329,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8313347777398384e-05,
+      "loss": 0.5267,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8343570308467003e-05,
+      "loss": 0.5433,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8373553923748272e-05,
+      "loss": 0.5353,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.84033023709989e-05,
+      "loss": 0.5412,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8432819310477555e-05,
+      "loss": 0.5642,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.846210831764756e-05,
+      "loss": 0.5058,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8491172885776048e-05,
+      "loss": 0.5529,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8520016428434306e-05,
+      "loss": 0.5669,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8548642281903704e-05,
+      "loss": 0.5544,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.85770537074916e-05,
+      "loss": 0.5362,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.860525389376108e-05,
+      "loss": 0.5348,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8633245958678425e-05,
+      "loss": 0.5391,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.866103295168196e-05,
+      "loss": 0.5746,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.868861785567562e-05,
+      "loss": 0.537,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8716003588950567e-05,
+      "loss": 0.5581,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8743193007037927e-05,
+      "loss": 0.5801,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.877018890449559e-05,
+      "loss": 0.527,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8796994016631868e-05,
+      "loss": 0.5206,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8823611021168685e-05,
+      "loss": 0.5466,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.88500425398468e-05,
+      "loss": 0.5149,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8876291139975484e-05,
+      "loss": 0.5264,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8902359335928962e-05,
+      "loss": 0.5139,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.892824959059175e-05,
+      "loss": 0.5266,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8953964316755035e-05,
+      "loss": 0.5279,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8979505878465988e-05,
+      "loss": 0.5317,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9004876592332e-05,
+      "loss": 0.5013,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9030078728781557e-05,
+      "loss": 0.5373,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.90551145132835e-05,
+      "loss": 0.5483,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9079986127526325e-05,
+      "loss": 0.548,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9104695710559053e-05,
+      "loss": 0.5916,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.912924535989519e-05,
+      "loss": 0.5659,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.915363713258118e-05,
+      "loss": 0.6128,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.917787304623072e-05,
+      "loss": 0.5286,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9201955080026257e-05,
+      "loss": 0.5458,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9225885175688857e-05,
+      "loss": 0.5554,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9249665238417722e-05,
+      "loss": 0.542,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9273297137800367e-05,
+      "loss": 0.5382,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.92967827086947e-05,
+      "loss": 0.5272,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.932012375208388e-05,
+      "loss": 0.5589,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9343322035905098e-05,
+      "loss": 0.5547,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9366379295853125e-05,
+      "loss": 0.5769,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.938929723615961e-05,
+      "loss": 0.5529,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9412077530348928e-05,
+      "loss": 0.5116,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9434721821971546e-05,
+      "loss": 0.5872,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.945723172531556e-05,
+      "loss": 0.5302,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9479608826097248e-05,
+      "loss": 0.5514,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.95018546821314e-05,
+      "loss": 0.511,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9523970823982097e-05,
+      "loss": 0.5396,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9545958755594613e-05,
+      "loss": 0.5844,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9567819954909106e-05,
+      "loss": 0.5852,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.958955587445675e-05,
+      "loss": 0.5667,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9611167941938848e-05,
+      "loss": 0.5469,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9632657560789565e-05,
+      "loss": 0.5768,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9654026110722805e-05,
+      "loss": 0.5151,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9675274948263752e-05,
+      "loss": 0.5007,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696405407265622e-05,
+      "loss": 0.5236,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97174187994121e-05,
+      "loss": 0.5554,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9738316414705926e-05,
+      "loss": 0.5222,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975909952194411e-05,
+      "loss": 0.5574,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9779769369180166e-05,
+      "loss": 0.4974,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800327184173855e-05,
+      "loss": 0.544,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820774174828814e-05,
+      "loss": 0.517,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841111529618374e-05,
+      "loss": 0.5318,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9861340418000133e-05,
+      "loss": 0.5431,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.988146199081945e-05,
+      "loss": 0.5601,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9901477380702322e-05,
+      "loss": 0.5703,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.992138770243796e-05,
+      "loss": 0.5488,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9941194053351344e-05,
+      "loss": 0.5416,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9960897513666135e-05,
+      "loss": 0.5429,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9980499146858193e-05,
+      "loss": 0.5408,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.5539,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.5554,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9996825396825398e-05,
+      "loss": 0.4997,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9993650793650798e-05,
+      "loss": 0.5338,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999047619047619e-05,
+      "loss": 0.5472,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987301587301587e-05,
+      "loss": 0.5067,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984126984126987e-05,
+      "loss": 0.5689,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9980952380952383e-05,
+      "loss": 0.5153,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997777777777778e-05,
+      "loss": 0.6012,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9974603174603176e-05,
+      "loss": 0.5278,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971428571428572e-05,
+      "loss": 0.5182,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.996825396825397e-05,
+      "loss": 0.5269,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9965079365079368e-05,
+      "loss": 0.5467,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9961904761904764e-05,
+      "loss": 0.524,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.995873015873016e-05,
+      "loss": 0.5324,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9955555555555557e-05,
+      "loss": 0.4895,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9952380952380953e-05,
+      "loss": 0.5364,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9949206349206353e-05,
+      "loss": 0.5519,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.994603174603175e-05,
+      "loss": 0.5415,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9942857142857142e-05,
+      "loss": 0.5337,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9939682539682542e-05,
+      "loss": 0.5556,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9936507936507938e-05,
+      "loss": 0.6063,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9933333333333334e-05,
+      "loss": 0.5157,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.993015873015873e-05,
+      "loss": 0.5204,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9926984126984127e-05,
+      "loss": 0.528,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9923809523809527e-05,
+      "loss": 0.5219,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9920634920634923e-05,
+      "loss": 0.544,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991746031746032e-05,
+      "loss": 0.5272,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9914285714285716e-05,
+      "loss": 0.5063,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9911111111111112e-05,
+      "loss": 0.5613,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.990793650793651e-05,
+      "loss": 0.546,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9904761904761908e-05,
+      "loss": 0.5462,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9901587301587304e-05,
+      "loss": 0.5452,
+      "step": 227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.98984126984127e-05,
+      "loss": 0.4992,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9895238095238097e-05,
+      "loss": 0.5108,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9892063492063493e-05,
+      "loss": 0.5734,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.988888888888889e-05,
+      "loss": 0.493,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.988571428571429e-05,
+      "loss": 0.5421,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9882539682539682e-05,
+      "loss": 0.5482,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9879365079365082e-05,
+      "loss": 0.5095,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987619047619048e-05,
+      "loss": 0.5654,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9873015873015875e-05,
+      "loss": 0.5393,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986984126984127e-05,
+      "loss": 0.5666,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9866666666666667e-05,
+      "loss": 0.5534,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9863492063492064e-05,
+      "loss": 0.551,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9860317460317463e-05,
+      "loss": 0.5345,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.985714285714286e-05,
+      "loss": 0.5273,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9853968253968256e-05,
+      "loss": 0.5212,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9850793650793652e-05,
+      "loss": 0.4981,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984761904761905e-05,
+      "loss": 0.5565,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9844444444444445e-05,
+      "loss": 0.5804,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9841269841269845e-05,
+      "loss": 0.5933,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.983809523809524e-05,
+      "loss": 0.5563,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9834920634920637e-05,
+      "loss": 0.5157,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9831746031746034e-05,
+      "loss": 0.5094,
+      "step": 249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.982857142857143e-05,
+      "loss": 0.5107,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9825396825396826e-05,
+      "loss": 0.5585,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9822222222222226e-05,
+      "loss": 0.5473,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.981904761904762e-05,
+      "loss": 0.4924,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.981587301587302e-05,
+      "loss": 0.516,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9812698412698415e-05,
+      "loss": 0.5358,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.980952380952381e-05,
+      "loss": 0.5432,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9806349206349208e-05,
+      "loss": 0.5335,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9803174603174604e-05,
+      "loss": 0.5201,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.98e-05,
+      "loss": 0.5294,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.97968253968254e-05,
+      "loss": 0.5429,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9793650793650796e-05,
+      "loss": 0.546,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9790476190476193e-05,
+      "loss": 0.5582,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.978730158730159e-05,
+      "loss": 0.5142,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9784126984126985e-05,
+      "loss": 0.5399,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.978095238095238e-05,
+      "loss": 0.5356,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.977777777777778e-05,
+      "loss": 0.5215,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9774603174603174e-05,
+      "loss": 0.5344,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9771428571428574e-05,
+      "loss": 0.5139,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.976825396825397e-05,
+      "loss": 0.5532,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9765079365079366e-05,
+      "loss": 0.5438,
+      "step": 270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9761904761904763e-05,
+      "loss": 0.5257,
+      "step": 271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.975873015873016e-05,
+      "loss": 0.4871,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9755555555555555e-05,
+      "loss": 0.5193,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9752380952380955e-05,
+      "loss": 0.4841,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.974920634920635e-05,
+      "loss": 0.5619,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9746031746031748e-05,
+      "loss": 0.5152,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9742857142857144e-05,
+      "loss": 0.5146,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.973968253968254e-05,
+      "loss": 0.5119,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9736507936507937e-05,
+      "loss": 0.5428,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 0.516,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9730158730158733e-05,
+      "loss": 0.5346,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.972698412698413e-05,
+      "loss": 0.5148,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9723809523809525e-05,
+      "loss": 0.5542,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.972063492063492e-05,
+      "loss": 0.5408,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9717460317460318e-05,
+      "loss": 0.5215,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9714285714285718e-05,
+      "loss": 0.5159,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.971111111111111e-05,
+      "loss": 0.5307,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.970793650793651e-05,
+      "loss": 0.4957,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9704761904761907e-05,
+      "loss": 0.5413,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9701587301587303e-05,
+      "loss": 0.5391,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9698412698412703e-05,
+      "loss": 0.5334,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9695238095238096e-05,
+      "loss": 0.5464,
+      "step": 292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9692063492063492e-05,
+      "loss": 0.531,
+      "step": 293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.968888888888889e-05,
+      "loss": 0.5302,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9685714285714288e-05,
+      "loss": 0.5297,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9682539682539684e-05,
+      "loss": 0.5275,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.967936507936508e-05,
+      "loss": 0.48,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9676190476190477e-05,
+      "loss": 0.4982,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9673015873015873e-05,
+      "loss": 0.5203,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9669841269841273e-05,
+      "loss": 0.5151,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9666666666666666e-05,
+      "loss": 0.5238,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9663492063492066e-05,
+      "loss": 0.5496,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9660317460317462e-05,
+      "loss": 0.5449,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9657142857142858e-05,
+      "loss": 0.5339,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9653968253968258e-05,
+      "loss": 0.5647,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.965079365079365e-05,
+      "loss": 0.5031,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9647619047619047e-05,
+      "loss": 0.5748,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9644444444444447e-05,
+      "loss": 0.5469,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9641269841269843e-05,
+      "loss": 0.5217,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.963809523809524e-05,
+      "loss": 0.5158,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9634920634920636e-05,
+      "loss": 0.5298,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9631746031746032e-05,
+      "loss": 0.5465,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9628571428571432e-05,
+      "loss": 0.575,
+      "step": 313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9625396825396828e-05,
+      "loss": 0.5184,
+      "step": 314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9622222222222224e-05,
+      "loss": 0.5613,
+      "step": 315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961904761904762e-05,
+      "loss": 0.5391,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9615873015873017e-05,
+      "loss": 0.4767,
+      "step": 317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9612698412698413e-05,
+      "loss": 0.5258,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9609523809523813e-05,
+      "loss": 0.4927,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.960634920634921e-05,
+      "loss": 0.5606,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9603174603174602e-05,
+      "loss": 0.5263,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.5247,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.95968253968254e-05,
+      "loss": 0.5114,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9593650793650795e-05,
+      "loss": 0.5338,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9590476190476194e-05,
+      "loss": 0.5304,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9587301587301587e-05,
+      "loss": 0.5118,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9584126984126987e-05,
+      "loss": 0.4834,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9580952380952383e-05,
+      "loss": 0.5001,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.957777777777778e-05,
+      "loss": 0.4847,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9574603174603176e-05,
+      "loss": 0.503,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9571428571428572e-05,
+      "loss": 0.5311,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.956825396825397e-05,
+      "loss": 0.5324,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.956507936507937e-05,
+      "loss": 0.5104,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9561904761904765e-05,
+      "loss": 0.5197,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9558730158730158e-05,
+      "loss": 0.5362,
+      "step": 335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.5287,
+      "step": 336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9552380952380954e-05,
+      "loss": 0.5479,
+      "step": 337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.954920634920635e-05,
+      "loss": 0.5605,
+      "step": 338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.954603174603175e-05,
+      "loss": 0.5025,
+      "step": 339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9542857142857143e-05,
+      "loss": 0.5172,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9539682539682542e-05,
+      "loss": 0.5304,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.953650793650794e-05,
+      "loss": 0.5073,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9533333333333335e-05,
+      "loss": 0.5145,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.953015873015873e-05,
+      "loss": 0.5039,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9526984126984128e-05,
+      "loss": 0.5324,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9523809523809524e-05,
+      "loss": 0.4918,
+      "step": 346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9520634920634924e-05,
+      "loss": 0.5242,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.951746031746032e-05,
+      "loss": 0.5163,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9514285714285716e-05,
+      "loss": 0.5579,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9511111111111113e-05,
+      "loss": 0.5364,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.950793650793651e-05,
+      "loss": 0.5156,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9504761904761905e-05,
+      "loss": 0.546,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9501587301587305e-05,
+      "loss": 0.575,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.94984126984127e-05,
+      "loss": 0.5084,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9495238095238098e-05,
+      "loss": 0.4977,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9492063492063494e-05,
+      "loss": 0.5214,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.948888888888889e-05,
+      "loss": 0.518,
+      "step": 357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9485714285714286e-05,
+      "loss": 0.5127,
+      "step": 358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9482539682539686e-05,
+      "loss": 0.5238,
+      "step": 359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.947936507936508e-05,
+      "loss": 0.5024,
+      "step": 360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.947619047619048e-05,
+      "loss": 0.513,
+      "step": 361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9473015873015875e-05,
+      "loss": 0.5415,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.946984126984127e-05,
+      "loss": 0.5188,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 0.5337,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9463492063492064e-05,
+      "loss": 0.5902,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.946031746031746e-05,
+      "loss": 0.5524,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.945714285714286e-05,
+      "loss": 0.5206,
+      "step": 367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9453968253968256e-05,
+      "loss": 0.5213,
+      "step": 368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9450793650793653e-05,
+      "loss": 0.5582,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.944761904761905e-05,
+      "loss": 0.5082,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.5086,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9441269841269842e-05,
+      "loss": 0.5165,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.943809523809524e-05,
+      "loss": 0.4984,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9434920634920634e-05,
+      "loss": 0.519,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9431746031746034e-05,
+      "loss": 0.5301,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.5608,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9425396825396827e-05,
+      "loss": 0.518,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9422222222222223e-05,
+      "loss": 0.5409,
+      "step": 378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.941904761904762e-05,
+      "loss": 0.5624,
+      "step": 379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9415873015873016e-05,
+      "loss": 0.5277,
+      "step": 380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9412698412698415e-05,
+      "loss": 0.5055,
+      "step": 381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.940952380952381e-05,
+      "loss": 0.5245,
+      "step": 382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9406349206349208e-05,
+      "loss": 0.5658,
+      "step": 383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9403174603174604e-05,
+      "loss": 0.5122,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.94e-05,
+      "loss": 0.5459,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9396825396825397e-05,
+      "loss": 0.4801,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9393650793650797e-05,
+      "loss": 0.5114,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9390476190476193e-05,
+      "loss": 0.5053,
+      "step": 388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.938730158730159e-05,
+      "loss": 0.542,
+      "step": 389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9384126984126986e-05,
+      "loss": 0.4946,
+      "step": 390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9380952380952382e-05,
+      "loss": 0.5365,
+      "step": 391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9377777777777778e-05,
+      "loss": 0.5043,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9374603174603178e-05,
+      "loss": 0.5385,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.937142857142857e-05,
+      "loss": 0.5282,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.936825396825397e-05,
+      "loss": 0.5016,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9365079365079367e-05,
+      "loss": 0.4994,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9361904761904763e-05,
+      "loss": 0.5268,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9358730158730163e-05,
+      "loss": 0.525,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9355555555555556e-05,
+      "loss": 0.4909,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9352380952380952e-05,
+      "loss": 0.513,
+      "step": 400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9349206349206352e-05,
+      "loss": 0.5438,
+      "step": 401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9346031746031748e-05,
+      "loss": 0.4991,
+      "step": 402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9342857142857144e-05,
+      "loss": 0.4826,
+      "step": 403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.933968253968254e-05,
+      "loss": 0.5132,
+      "step": 404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9336507936507937e-05,
+      "loss": 0.5366,
+      "step": 405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 0.5092,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9330158730158733e-05,
+      "loss": 0.5148,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.932698412698413e-05,
+      "loss": 0.5334,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9323809523809526e-05,
+      "loss": 0.5231,
+      "step": 409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9320634920634922e-05,
+      "loss": 0.5056,
+      "step": 410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.931746031746032e-05,
+      "loss": 0.5305,
+      "step": 411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9314285714285718e-05,
+      "loss": 0.5445,
+      "step": 412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.931111111111111e-05,
+      "loss": 0.5064,
+      "step": 413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9307936507936507e-05,
+      "loss": 0.4977,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9304761904761907e-05,
+      "loss": 0.5346,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9301587301587303e-05,
+      "loss": 0.5374,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.92984126984127e-05,
+      "loss": 0.5294,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9295238095238096e-05,
+      "loss": 0.4894,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9292063492063492e-05,
+      "loss": 0.5621,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.928888888888889e-05,
+      "loss": 0.5166,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.4837,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9282539682539685e-05,
+      "loss": 0.4728,
+      "step": 422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.927936507936508e-05,
+      "loss": 0.5523,
+      "step": 423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9276190476190477e-05,
+      "loss": 0.5397,
+      "step": 424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9273015873015874e-05,
+      "loss": 0.5261,
+      "step": 425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9269841269841273e-05,
+      "loss": 0.5445,
+      "step": 426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.926666666666667e-05,
+      "loss": 0.5221,
+      "step": 427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9263492063492063e-05,
+      "loss": 0.5204,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9260317460317462e-05,
+      "loss": 0.5247,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.925714285714286e-05,
+      "loss": 0.5184,
+      "step": 430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9253968253968255e-05,
+      "loss": 0.531,
+      "step": 431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9250793650793655e-05,
+      "loss": 0.5304,
+      "step": 432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9247619047619048e-05,
+      "loss": 0.5127,
+      "step": 433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9244444444444444e-05,
+      "loss": 0.5407,
+      "step": 434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9241269841269844e-05,
+      "loss": 0.5178,
+      "step": 435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.923809523809524e-05,
+      "loss": 0.5136,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9234920634920636e-05,
+      "loss": 0.54,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9231746031746033e-05,
+      "loss": 0.5227,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.922857142857143e-05,
+      "loss": 0.5511,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.922539682539683e-05,
+      "loss": 0.4637,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9222222222222225e-05,
+      "loss": 0.4944,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.921904761904762e-05,
+      "loss": 0.5277,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9215873015873018e-05,
+      "loss": 0.4844,
+      "step": 443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9212698412698414e-05,
+      "loss": 0.5336,
+      "step": 444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.920952380952381e-05,
+      "loss": 0.5026,
+      "step": 445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.920634920634921e-05,
+      "loss": 0.4892,
+      "step": 446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9203174603174606e-05,
+      "loss": 0.5463,
+      "step": 447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.4914,
+      "step": 448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.91968253968254e-05,
+      "loss": 0.5117,
+      "step": 449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9193650793650795e-05,
+      "loss": 0.5312,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.919047619047619e-05,
+      "loss": 0.5232,
+      "step": 451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9187301587301588e-05,
+      "loss": 0.5125,
+      "step": 452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9184126984126984e-05,
+      "loss": 0.5335,
+      "step": 453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9180952380952384e-05,
+      "loss": 0.5187,
+      "step": 454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.917777777777778e-05,
+      "loss": 0.5594,
+      "step": 455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9174603174603176e-05,
+      "loss": 0.4928,
+      "step": 456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9171428571428573e-05,
+      "loss": 0.5276,
+      "step": 457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.916825396825397e-05,
+      "loss": 0.5028,
+      "step": 458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9165079365079365e-05,
+      "loss": 0.4582,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9161904761904765e-05,
+      "loss": 0.4911,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.915873015873016e-05,
+      "loss": 0.5356,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9155555555555558e-05,
+      "loss": 0.5042,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9152380952380954e-05,
+      "loss": 0.4762,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.914920634920635e-05,
+      "loss": 0.5361,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9146031746031747e-05,
+      "loss": 0.5045,
+      "step": 465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9142857142857146e-05,
+      "loss": 0.5189,
+      "step": 466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913968253968254e-05,
+      "loss": 0.5659,
+      "step": 467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913650793650794e-05,
+      "loss": 0.5353,
+      "step": 468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9133333333333335e-05,
+      "loss": 0.5151,
+      "step": 469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913015873015873e-05,
+      "loss": 0.5851,
+      "step": 470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9126984126984128e-05,
+      "loss": 0.538,
+      "step": 471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9123809523809524e-05,
+      "loss": 0.5318,
+      "step": 472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.912063492063492e-05,
+      "loss": 0.4996,
+      "step": 473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.911746031746032e-05,
+      "loss": 0.5311,
+      "step": 474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9114285714285717e-05,
+      "loss": 0.4776,
+      "step": 475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 0.5526,
+      "step": 476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.910793650793651e-05,
+      "loss": 0.5336,
+      "step": 477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9104761904761906e-05,
+      "loss": 0.5304,
+      "step": 478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9101587301587302e-05,
+      "loss": 0.5126,
+      "step": 479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.90984126984127e-05,
+      "loss": 0.5203,
+      "step": 480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9095238095238098e-05,
+      "loss": 0.5021,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9092063492063494e-05,
+      "loss": 0.541,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.908888888888889e-05,
+      "loss": 0.5663,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9085714285714287e-05,
+      "loss": 0.5103,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9082539682539683e-05,
+      "loss": 0.4825,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9079365079365083e-05,
+      "loss": 0.5252,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9076190476190476e-05,
+      "loss": 0.5282,
+      "step": 487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9073015873015876e-05,
+      "loss": 0.5106,
+      "step": 488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9069841269841272e-05,
+      "loss": 0.531,
+      "step": 489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9066666666666668e-05,
+      "loss": 0.4818,
+      "step": 490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9063492063492065e-05,
+      "loss": 0.5112,
+      "step": 491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.906031746031746e-05,
+      "loss": 0.5438,
+      "step": 492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9057142857142857e-05,
+      "loss": 0.5329,
+      "step": 493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9053968253968257e-05,
+      "loss": 0.5,
+      "step": 494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9050793650793653e-05,
+      "loss": 0.5238,
+      "step": 495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 0.5209,
+      "step": 496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9044444444444446e-05,
+      "loss": 0.4783,
+      "step": 497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9041269841269842e-05,
+      "loss": 0.5309,
+      "step": 498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.903809523809524e-05,
+      "loss": 0.503,
+      "step": 499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9034920634920638e-05,
+      "loss": 0.504,
+      "step": 500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.903174603174603e-05,
+      "loss": 0.5322,
+      "step": 501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.902857142857143e-05,
+      "loss": 0.5353,
+      "step": 502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9025396825396827e-05,
+      "loss": 0.5352,
+      "step": 503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9022222222222223e-05,
+      "loss": 0.5113,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.901904761904762e-05,
+      "loss": 0.5116,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9015873015873016e-05,
+      "loss": 0.5251,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9012698412698412e-05,
+      "loss": 0.551,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9009523809523812e-05,
+      "loss": 0.5309,
+      "step": 508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.900634920634921e-05,
+      "loss": 0.5341,
+      "step": 509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9003174603174605e-05,
+      "loss": 0.5064,
+      "step": 510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9e-05,
+      "loss": 0.5096,
+      "step": 511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8996825396825397e-05,
+      "loss": 0.5088,
+      "step": 512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8993650793650794e-05,
+      "loss": 0.527,
+      "step": 513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8990476190476193e-05,
+      "loss": 0.518,
+      "step": 514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.898730158730159e-05,
+      "loss": 0.5409,
+      "step": 515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8984126984126986e-05,
+      "loss": 0.5071,
+      "step": 516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8980952380952382e-05,
+      "loss": 0.4926,
+      "step": 517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.897777777777778e-05,
+      "loss": 0.5361,
+      "step": 518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.897460317460318e-05,
+      "loss": 0.5381,
+      "step": 519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8971428571428575e-05,
+      "loss": 0.5209,
+      "step": 520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8968253968253968e-05,
+      "loss": 0.5245,
+      "step": 521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8965079365079367e-05,
+      "loss": 0.5051,
+      "step": 522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8961904761904764e-05,
+      "loss": 0.5322,
+      "step": 523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.895873015873016e-05,
+      "loss": 0.5344,
+      "step": 524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.895555555555556e-05,
+      "loss": 0.5204,
+      "step": 525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8952380952380953e-05,
+      "loss": 0.512,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.894920634920635e-05,
+      "loss": 0.5135,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.894603174603175e-05,
+      "loss": 0.5582,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8942857142857145e-05,
+      "loss": 0.4995,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.893968253968254e-05,
+      "loss": 0.5296,
+      "step": 530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8936507936507938e-05,
+      "loss": 0.5152,
+      "step": 531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 0.4929,
+      "step": 532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8930158730158734e-05,
+      "loss": 0.4763,
+      "step": 533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.892698412698413e-05,
+      "loss": 0.495,
+      "step": 534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8923809523809523e-05,
+      "loss": 0.5362,
+      "step": 535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8920634920634923e-05,
+      "loss": 0.5358,
+      "step": 536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.891746031746032e-05,
+      "loss": 0.5059,
+      "step": 537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8914285714285715e-05,
+      "loss": 0.536,
+      "step": 538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8911111111111115e-05,
+      "loss": 0.5146,
+      "step": 539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8907936507936508e-05,
+      "loss": 0.5055,
+      "step": 540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8904761904761904e-05,
+      "loss": 0.4993,
+      "step": 541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8901587301587304e-05,
+      "loss": 0.4766,
+      "step": 542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.88984126984127e-05,
+      "loss": 0.5119,
+      "step": 543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8895238095238096e-05,
+      "loss": 0.535,
+      "step": 544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8892063492063493e-05,
+      "loss": 0.4862,
+      "step": 545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 0.5331,
+      "step": 546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.888571428571429e-05,
+      "loss": 0.5591,
+      "step": 547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8882539682539685e-05,
+      "loss": 0.5133,
+      "step": 548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.887936507936508e-05,
+      "loss": 0.522,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8876190476190478e-05,
+      "loss": 0.4776,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8873015873015874e-05,
+      "loss": 0.5564,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.886984126984127e-05,
+      "loss": 0.5349,
+      "step": 552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.886666666666667e-05,
+      "loss": 0.5368,
+      "step": 553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8863492063492066e-05,
+      "loss": 0.5063,
+      "step": 554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.886031746031746e-05,
+      "loss": 0.4973,
+      "step": 555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 0.4923,
+      "step": 556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8853968253968255e-05,
+      "loss": 0.5314,
+      "step": 557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8850793650793652e-05,
+      "loss": 0.519,
+      "step": 558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.884761904761905e-05,
+      "loss": 0.5529,
+      "step": 559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8844444444444444e-05,
+      "loss": 0.531,
+      "step": 560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8841269841269844e-05,
+      "loss": 0.4955,
+      "step": 561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.883809523809524e-05,
+      "loss": 0.4942,
+      "step": 562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8834920634920637e-05,
+      "loss": 0.5048,
+      "step": 563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8831746031746033e-05,
+      "loss": 0.5308,
+      "step": 564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.882857142857143e-05,
+      "loss": 0.5222,
+      "step": 565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8825396825396826e-05,
+      "loss": 0.526,
+      "step": 566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8822222222222225e-05,
+      "loss": 0.5556,
+      "step": 567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.881904761904762e-05,
+      "loss": 0.542,
+      "step": 568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8815873015873018e-05,
+      "loss": 0.5601,
+      "step": 569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8812698412698414e-05,
+      "loss": 0.4933,
+      "step": 570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.880952380952381e-05,
+      "loss": 0.5086,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8806349206349207e-05,
+      "loss": 0.5303,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8803174603174607e-05,
+      "loss": 0.5378,
+      "step": 573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.88e-05,
+      "loss": 0.4785,
+      "step": 574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.87968253968254e-05,
+      "loss": 0.5491,
+      "step": 575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8793650793650796e-05,
+      "loss": 0.5315,
+      "step": 576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8790476190476192e-05,
+      "loss": 0.5341,
+      "step": 577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8787301587301588e-05,
+      "loss": 0.5354,
+      "step": 578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8784126984126985e-05,
+      "loss": 0.5434,
+      "step": 579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.878095238095238e-05,
+      "loss": 0.5014,
+      "step": 580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.877777777777778e-05,
+      "loss": 0.5615,
+      "step": 581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8774603174603177e-05,
+      "loss": 0.5544,
+      "step": 582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8771428571428573e-05,
+      "loss": 0.5496,
+      "step": 583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.876825396825397e-05,
+      "loss": 0.5292,
+      "step": 584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8765079365079366e-05,
+      "loss": 0.5209,
+      "step": 585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8761904761904762e-05,
+      "loss": 0.4902,
+      "step": 586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8758730158730162e-05,
+      "loss": 0.509,
+      "step": 587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8755555555555558e-05,
+      "loss": 0.5053,
+      "step": 588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8752380952380955e-05,
+      "loss": 0.5148,
+      "step": 589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.874920634920635e-05,
+      "loss": 0.5454,
+      "step": 590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8746031746031747e-05,
+      "loss": 0.5385,
+      "step": 591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8742857142857143e-05,
+      "loss": 0.4959,
+      "step": 592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8739682539682543e-05,
+      "loss": 0.5561,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8736507936507936e-05,
+      "loss": 0.516,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8733333333333336e-05,
+      "loss": 0.4984,
+      "step": 595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8730158730158732e-05,
+      "loss": 0.4863,
+      "step": 596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.872698412698413e-05,
+      "loss": 0.5322,
+      "step": 597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8723809523809525e-05,
+      "loss": 0.5327,
+      "step": 598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.872063492063492e-05,
+      "loss": 0.4711,
+      "step": 599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8717460317460317e-05,
+      "loss": 0.495,
+      "step": 600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8714285714285717e-05,
+      "loss": 0.475,
+      "step": 601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8711111111111113e-05,
+      "loss": 0.5049,
+      "step": 602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.870793650793651e-05,
+      "loss": 0.4952,
+      "step": 603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8704761904761906e-05,
+      "loss": 0.5313,
+      "step": 604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8701587301587302e-05,
+      "loss": 0.5388,
+      "step": 605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.86984126984127e-05,
+      "loss": 0.4895,
+      "step": 606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.86952380952381e-05,
+      "loss": 0.5534,
+      "step": 607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8692063492063495e-05,
+      "loss": 0.4735,
+      "step": 608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.868888888888889e-05,
+      "loss": 0.5011,
+      "step": 609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8685714285714287e-05,
+      "loss": 0.4966,
+      "step": 610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8682539682539684e-05,
+      "loss": 0.5156,
+      "step": 611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.867936507936508e-05,
+      "loss": 0.5005,
+      "step": 612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8676190476190476e-05,
+      "loss": 0.4922,
+      "step": 613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8673015873015873e-05,
+      "loss": 0.5149,
+      "step": 614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8669841269841272e-05,
+      "loss": 0.5011,
+      "step": 615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.5601,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8663492063492065e-05,
+      "loss": 0.5153,
+      "step": 617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866031746031746e-05,
+      "loss": 0.4914,
+      "step": 618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8657142857142858e-05,
+      "loss": 0.505,
+      "step": 619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8653968253968254e-05,
+      "loss": 0.4955,
+      "step": 620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8650793650793654e-05,
+      "loss": 0.5126,
+      "step": 621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864761904761905e-05,
+      "loss": 0.5298,
+      "step": 622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8644444444444446e-05,
+      "loss": 0.5224,
+      "step": 623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8641269841269843e-05,
+      "loss": 0.5412,
+      "step": 624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863809523809524e-05,
+      "loss": 0.5097,
+      "step": 625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8634920634920635e-05,
+      "loss": 0.5399,
+      "step": 626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8631746031746035e-05,
+      "loss": 0.5442,
+      "step": 627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8628571428571428e-05,
+      "loss": 0.5023,
+      "step": 628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8625396825396828e-05,
+      "loss": 0.512,
+      "step": 629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8622222222222224e-05,
+      "loss": 0.5095,
+      "step": 630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861904761904762e-05,
+      "loss": 0.4959,
+      "step": 631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861587301587302e-05,
+      "loss": 0.5213,
+      "step": 632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8612698412698413e-05,
+      "loss": 0.5256,
+      "step": 633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860952380952381e-05,
+      "loss": 0.4982,
+      "step": 634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860634920634921e-05,
+      "loss": 0.5387,
+      "step": 635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8603174603174605e-05,
+      "loss": 0.5066,
+      "step": 636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.86e-05,
+      "loss": 0.5208,
+      "step": 637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8596825396825398e-05,
+      "loss": 0.504,
+      "step": 638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8593650793650794e-05,
+      "loss": 0.5305,
+      "step": 639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.859047619047619e-05,
+      "loss": 0.5027,
+      "step": 640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.858730158730159e-05,
+      "loss": 0.5074,
+      "step": 641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8584126984126986e-05,
+      "loss": 0.5434,
+      "step": 642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8580952380952383e-05,
+      "loss": 0.5272,
+      "step": 643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.857777777777778e-05,
+      "loss": 0.516,
+      "step": 644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8574603174603175e-05,
+      "loss": 0.5218,
+      "step": 645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.5205,
+      "step": 646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.856825396825397e-05,
+      "loss": 0.5632,
+      "step": 647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8565079365079364e-05,
+      "loss": 0.5013,
+      "step": 648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8561904761904764e-05,
+      "loss": 0.5398,
+      "step": 649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.855873015873016e-05,
+      "loss": 0.5109,
+      "step": 650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8555555555555557e-05,
+      "loss": 0.5294,
+      "step": 651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8552380952380953e-05,
+      "loss": 0.5539,
+      "step": 652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.854920634920635e-05,
+      "loss": 0.4908,
+      "step": 653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.854603174603175e-05,
+      "loss": 0.5235,
+      "step": 654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8542857142857145e-05,
+      "loss": 0.5288,
+      "step": 655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853968253968254e-05,
+      "loss": 0.5276,
+      "step": 656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8536507936507938e-05,
+      "loss": 0.5309,
+      "step": 657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8533333333333334e-05,
+      "loss": 0.5065,
+      "step": 658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853015873015873e-05,
+      "loss": 0.4717,
+      "step": 659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852698412698413e-05,
+      "loss": 0.4873,
+      "step": 660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8523809523809527e-05,
+      "loss": 0.5081,
+      "step": 661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.852063492063492e-05,
+      "loss": 0.5262,
+      "step": 662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.851746031746032e-05,
+      "loss": 0.5321,
+      "step": 663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8514285714285716e-05,
+      "loss": 0.5379,
+      "step": 664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8511111111111112e-05,
+      "loss": 0.5275,
+      "step": 665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.850793650793651e-05,
+      "loss": 0.5428,
+      "step": 666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8504761904761905e-05,
+      "loss": 0.4967,
+      "step": 667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8501587301587304e-05,
+      "loss": 0.5017,
+      "step": 668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.84984126984127e-05,
+      "loss": 0.5435,
+      "step": 669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8495238095238097e-05,
+      "loss": 0.4598,
+      "step": 670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8492063492063493e-05,
+      "loss": 0.4958,
+      "step": 671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.848888888888889e-05,
+      "loss": 0.5246,
+      "step": 672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8485714285714286e-05,
+      "loss": 0.5166,
+      "step": 673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8482539682539686e-05,
+      "loss": 0.514,
+      "step": 674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8479365079365082e-05,
+      "loss": 0.5004,
+      "step": 675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8476190476190478e-05,
+      "loss": 0.4927,
+      "step": 676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8473015873015875e-05,
+      "loss": 0.5121,
+      "step": 677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.846984126984127e-05,
+      "loss": 0.5585,
+      "step": 678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8466666666666667e-05,
+      "loss": 0.4714,
+      "step": 679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8463492063492067e-05,
+      "loss": 0.5172,
+      "step": 680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8460317460317463e-05,
+      "loss": 0.4936,
+      "step": 681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.845714285714286e-05,
+      "loss": 0.5152,
+      "step": 682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8453968253968256e-05,
+      "loss": 0.4992,
+      "step": 683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8450793650793652e-05,
+      "loss": 0.5324,
+      "step": 684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.844761904761905e-05,
+      "loss": 0.4965,
+      "step": 685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8444444444444448e-05,
+      "loss": 0.5514,
+      "step": 686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.844126984126984e-05,
+      "loss": 0.5136,
+      "step": 687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.843809523809524e-05,
+      "loss": 0.4932,
+      "step": 688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8434920634920637e-05,
+      "loss": 0.5318,
+      "step": 689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8431746031746033e-05,
+      "loss": 0.4744,
+      "step": 690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.842857142857143e-05,
+      "loss": 0.4881,
+      "step": 691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8425396825396826e-05,
+      "loss": 0.5504,
+      "step": 692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8422222222222222e-05,
+      "loss": 0.5306,
+      "step": 693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8419047619047622e-05,
+      "loss": 0.4945,
+      "step": 694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.841587301587302e-05,
+      "loss": 0.5241,
+      "step": 695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8412698412698415e-05,
+      "loss": 0.4724,
+      "step": 696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.840952380952381e-05,
+      "loss": 0.5047,
+      "step": 697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8406349206349207e-05,
+      "loss": 0.5112,
+      "step": 698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8403174603174604e-05,
+      "loss": 0.5067,
+      "step": 699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.4908,
+      "step": 700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8396825396825396e-05,
+      "loss": 0.493,
+      "step": 701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8393650793650796e-05,
+      "loss": 0.4875,
+      "step": 702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8390476190476192e-05,
+      "loss": 0.518,
+      "step": 703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.838730158730159e-05,
+      "loss": 0.5164,
+      "step": 704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8384126984126985e-05,
+      "loss": 0.5056,
+      "step": 705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.838095238095238e-05,
+      "loss": 0.5151,
+      "step": 706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8377777777777778e-05,
+      "loss": 0.525,
+      "step": 707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8374603174603177e-05,
+      "loss": 0.4812,
+      "step": 708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8371428571428574e-05,
+      "loss": 0.4945,
+      "step": 709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.836825396825397e-05,
+      "loss": 0.5424,
+      "step": 710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8365079365079366e-05,
+      "loss": 0.4919,
+      "step": 711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8361904761904763e-05,
+      "loss": 0.5372,
+      "step": 712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.835873015873016e-05,
+      "loss": 0.544,
+      "step": 713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.835555555555556e-05,
+      "loss": 0.5129,
+      "step": 714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8352380952380955e-05,
+      "loss": 0.5346,
+      "step": 715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.834920634920635e-05,
+      "loss": 0.5212,
+      "step": 716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8346031746031748e-05,
+      "loss": 0.5418,
+      "step": 717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8342857142857144e-05,
+      "loss": 0.506,
+      "step": 718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.833968253968254e-05,
+      "loss": 0.4954,
+      "step": 719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.833650793650794e-05,
+      "loss": 0.5796,
+      "step": 720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.4834,
+      "step": 721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8330158730158733e-05,
+      "loss": 0.5198,
+      "step": 722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.832698412698413e-05,
+      "loss": 0.5413,
+      "step": 723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8323809523809525e-05,
+      "loss": 0.5001,
+      "step": 724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8320634920634925e-05,
+      "loss": 0.4828,
+      "step": 725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8317460317460318e-05,
+      "loss": 0.5163,
+      "step": 726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8314285714285714e-05,
+      "loss": 0.5116,
+      "step": 727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8311111111111114e-05,
+      "loss": 0.507,
+      "step": 728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.830793650793651e-05,
+      "loss": 0.5662,
+      "step": 729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8304761904761906e-05,
+      "loss": 0.5229,
+      "step": 730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8301587301587303e-05,
+      "loss": 0.5276,
+      "step": 731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.82984126984127e-05,
+      "loss": 0.4801,
+      "step": 732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8295238095238095e-05,
+      "loss": 0.554,
+      "step": 733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8292063492063495e-05,
+      "loss": 0.4921,
+      "step": 734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8288888888888888e-05,
+      "loss": 0.5044,
+      "step": 735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 0.497,
+      "step": 736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8282539682539684e-05,
+      "loss": 0.4907,
+      "step": 737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.827936507936508e-05,
+      "loss": 0.5095,
+      "step": 738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.827619047619048e-05,
+      "loss": 0.505,
+      "step": 739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8273015873015873e-05,
+      "loss": 0.5526,
+      "step": 740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.826984126984127e-05,
+      "loss": 0.5409,
+      "step": 741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.826666666666667e-05,
+      "loss": 0.5003,
+      "step": 742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8263492063492065e-05,
+      "loss": 0.5156,
+      "step": 743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8260317460317462e-05,
+      "loss": 0.5183,
+      "step": 744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8257142857142858e-05,
+      "loss": 0.474,
+      "step": 745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8253968253968254e-05,
+      "loss": 0.5185,
+      "step": 746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.825079365079365e-05,
+      "loss": 0.5287,
+      "step": 747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.824761904761905e-05,
+      "loss": 0.4949,
+      "step": 748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8244444444444447e-05,
+      "loss": 0.472,
+      "step": 749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8241269841269843e-05,
+      "loss": 0.4945,
+      "step": 750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.823809523809524e-05,
+      "loss": 0.4792,
+      "step": 751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8234920634920636e-05,
+      "loss": 0.4926,
+      "step": 752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8231746031746035e-05,
+      "loss": 0.515,
+      "step": 753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.822857142857143e-05,
+      "loss": 0.5193,
+      "step": 754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8225396825396825e-05,
+      "loss": 0.5088,
+      "step": 755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 0.5299,
+      "step": 756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.821904761904762e-05,
+      "loss": 0.5364,
+      "step": 757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8215873015873017e-05,
+      "loss": 0.4992,
+      "step": 758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8212698412698417e-05,
+      "loss": 0.4967,
+      "step": 759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.820952380952381e-05,
+      "loss": 0.544,
+      "step": 760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8206349206349206e-05,
+      "loss": 0.5007,
+      "step": 761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8203174603174606e-05,
+      "loss": 0.489,
+      "step": 762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 0.4906,
+      "step": 763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8196825396825398e-05,
+      "loss": 0.4921,
+      "step": 764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8193650793650795e-05,
+      "loss": 0.5426,
+      "step": 765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.819047619047619e-05,
+      "loss": 0.5181,
+      "step": 766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.818730158730159e-05,
+      "loss": 0.5146,
+      "step": 767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8184126984126987e-05,
+      "loss": 0.5219,
+      "step": 768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.818095238095238e-05,
+      "loss": 0.5426,
+      "step": 769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.817777777777778e-05,
+      "loss": 0.5297,
+      "step": 770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8174603174603176e-05,
+      "loss": 0.4772,
+      "step": 771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8171428571428572e-05,
+      "loss": 0.5032,
+      "step": 772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8168253968253972e-05,
+      "loss": 0.5212,
+      "step": 773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8165079365079365e-05,
+      "loss": 0.5484,
+      "step": 774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.816190476190476e-05,
+      "loss": 0.4661,
+      "step": 775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.815873015873016e-05,
+      "loss": 0.4883,
+      "step": 776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8155555555555557e-05,
+      "loss": 0.5314,
+      "step": 777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8152380952380953e-05,
+      "loss": 0.4623,
+      "step": 778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.814920634920635e-05,
+      "loss": 0.5086,
+      "step": 779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8146031746031746e-05,
+      "loss": 0.5415,
+      "step": 780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8142857142857146e-05,
+      "loss": 0.5252,
+      "step": 781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8139682539682542e-05,
+      "loss": 0.5207,
+      "step": 782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.813650793650794e-05,
+      "loss": 0.4804,
+      "step": 783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 0.5301,
+      "step": 784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.813015873015873e-05,
+      "loss": 0.5051,
+      "step": 785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8126984126984127e-05,
+      "loss": 0.498,
+      "step": 786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8123809523809527e-05,
+      "loss": 0.4765,
+      "step": 787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8120634920634923e-05,
+      "loss": 0.4557,
+      "step": 788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.811746031746032e-05,
+      "loss": 0.5081,
+      "step": 789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8114285714285716e-05,
+      "loss": 0.5254,
+      "step": 790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8111111111111112e-05,
+      "loss": 0.5069,
+      "step": 791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.810793650793651e-05,
+      "loss": 0.5128,
+      "step": 792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.810476190476191e-05,
+      "loss": 0.4703,
+      "step": 793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.81015873015873e-05,
+      "loss": 0.4978,
+      "step": 794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.80984126984127e-05,
+      "loss": 0.4781,
+      "step": 795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 0.5392,
+      "step": 796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8092063492063494e-05,
+      "loss": 0.5039,
+      "step": 797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.808888888888889e-05,
+      "loss": 0.5354,
+      "step": 798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8085714285714286e-05,
+      "loss": 0.5137,
+      "step": 799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8082539682539683e-05,
+      "loss": 0.5147,
+      "step": 800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8079365079365082e-05,
+      "loss": 0.5109,
+      "step": 801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.807619047619048e-05,
+      "loss": 0.5134,
+      "step": 802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8073015873015875e-05,
+      "loss": 0.5359,
+      "step": 803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.806984126984127e-05,
+      "loss": 0.511,
+      "step": 804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8066666666666668e-05,
+      "loss": 0.4957,
+      "step": 805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8063492063492064e-05,
+      "loss": 0.5212,
+      "step": 806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8060317460317464e-05,
+      "loss": 0.5331,
+      "step": 807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8057142857142857e-05,
+      "loss": 0.5071,
+      "step": 808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8053968253968256e-05,
+      "loss": 0.4866,
+      "step": 809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8050793650793653e-05,
+      "loss": 0.5228,
+      "step": 810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.804761904761905e-05,
+      "loss": 0.4574,
+      "step": 811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8044444444444445e-05,
+      "loss": 0.5017,
+      "step": 812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.804126984126984e-05,
+      "loss": 0.5761,
+      "step": 813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8038095238095238e-05,
+      "loss": 0.486,
+      "step": 814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8034920634920638e-05,
+      "loss": 0.5212,
+      "step": 815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8031746031746034e-05,
+      "loss": 0.4982,
+      "step": 816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.802857142857143e-05,
+      "loss": 0.5256,
+      "step": 817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8025396825396827e-05,
+      "loss": 0.5066,
+      "step": 818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8022222222222223e-05,
+      "loss": 0.4995,
+      "step": 819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.801904761904762e-05,
+      "loss": 0.5239,
+      "step": 820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.801587301587302e-05,
+      "loss": 0.5278,
+      "step": 821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8012698412698415e-05,
+      "loss": 0.4629,
+      "step": 822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.800952380952381e-05,
+      "loss": 0.5171,
+      "step": 823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8006349206349208e-05,
+      "loss": 0.4791,
+      "step": 824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8003174603174604e-05,
+      "loss": 0.492,
+      "step": 825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5577,
+      "step": 826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.79968253968254e-05,
+      "loss": 0.4749,
+      "step": 827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7993650793650793e-05,
+      "loss": 0.534,
+      "step": 828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7990476190476193e-05,
+      "loss": 0.4984,
+      "step": 829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.798730158730159e-05,
+      "loss": 0.5066,
+      "step": 830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7984126984126985e-05,
+      "loss": 0.4873,
+      "step": 831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7980952380952382e-05,
+      "loss": 0.4993,
+      "step": 832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7977777777777778e-05,
+      "loss": 0.5119,
+      "step": 833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7974603174603174e-05,
+      "loss": 0.4656,
+      "step": 834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7971428571428574e-05,
+      "loss": 0.5081,
+      "step": 835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.796825396825397e-05,
+      "loss": 0.4906,
+      "step": 836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7965079365079367e-05,
+      "loss": 0.4944,
+      "step": 837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7961904761904763e-05,
+      "loss": 0.4927,
+      "step": 838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.795873015873016e-05,
+      "loss": 0.4588,
+      "step": 839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7955555555555556e-05,
+      "loss": 0.541,
+      "step": 840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7952380952380955e-05,
+      "loss": 0.5428,
+      "step": 841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7949206349206352e-05,
+      "loss": 0.5146,
+      "step": 842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7946031746031748e-05,
+      "loss": 0.5048,
+      "step": 843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7942857142857144e-05,
+      "loss": 0.5291,
+      "step": 844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.793968253968254e-05,
+      "loss": 0.4975,
+      "step": 845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7936507936507937e-05,
+      "loss": 0.4956,
+      "step": 846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7933333333333333e-05,
+      "loss": 0.5223,
+      "step": 847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.793015873015873e-05,
+      "loss": 0.4825,
+      "step": 848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.792698412698413e-05,
+      "loss": 0.5381,
+      "step": 849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7923809523809526e-05,
+      "loss": 0.4836,
+      "step": 850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7920634920634922e-05,
+      "loss": 0.4991,
+      "step": 851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7917460317460318e-05,
+      "loss": 0.4758,
+      "step": 852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7914285714285715e-05,
+      "loss": 0.5711,
+      "step": 853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.791111111111111e-05,
+      "loss": 0.5152,
+      "step": 854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.790793650793651e-05,
+      "loss": 0.489,
+      "step": 855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7904761904761907e-05,
+      "loss": 0.5073,
+      "step": 856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7901587301587303e-05,
+      "loss": 0.4717,
+      "step": 857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.78984126984127e-05,
+      "loss": 0.5024,
+      "step": 858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7895238095238096e-05,
+      "loss": 0.5318,
+      "step": 859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7892063492063496e-05,
+      "loss": 0.4955,
+      "step": 860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7888888888888892e-05,
+      "loss": 0.5518,
+      "step": 861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7885714285714285e-05,
+      "loss": 0.4984,
+      "step": 862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7882539682539685e-05,
+      "loss": 0.5216,
+      "step": 863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.787936507936508e-05,
+      "loss": 0.5315,
+      "step": 864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7876190476190477e-05,
+      "loss": 0.4598,
+      "step": 865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7873015873015877e-05,
+      "loss": 0.5039,
+      "step": 866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.786984126984127e-05,
+      "loss": 0.503,
+      "step": 867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 0.4797,
+      "step": 868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7863492063492066e-05,
+      "loss": 0.4619,
+      "step": 869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7860317460317462e-05,
+      "loss": 0.5172,
+      "step": 870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.5348,
+      "step": 871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7853968253968255e-05,
+      "loss": 0.4754,
+      "step": 872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.785079365079365e-05,
+      "loss": 0.5132,
+      "step": 873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.784761904761905e-05,
+      "loss": 0.4975,
+      "step": 874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7844444444444447e-05,
+      "loss": 0.5426,
+      "step": 875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7841269841269843e-05,
+      "loss": 0.5081,
+      "step": 876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.783809523809524e-05,
+      "loss": 0.4769,
+      "step": 877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7834920634920636e-05,
+      "loss": 0.5051,
+      "step": 878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7831746031746032e-05,
+      "loss": 0.5131,
+      "step": 879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7828571428571432e-05,
+      "loss": 0.5222,
+      "step": 880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.782539682539683e-05,
+      "loss": 0.5169,
+      "step": 881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.782222222222222e-05,
+      "loss": 0.4846,
+      "step": 882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.781904761904762e-05,
+      "loss": 0.5124,
+      "step": 883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7815873015873017e-05,
+      "loss": 0.4645,
+      "step": 884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7812698412698414e-05,
+      "loss": 0.5127,
+      "step": 885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.780952380952381e-05,
+      "loss": 0.5056,
+      "step": 886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7806349206349206e-05,
+      "loss": 0.4991,
+      "step": 887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7803174603174606e-05,
+      "loss": 0.5601,
+      "step": 888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 0.4788,
+      "step": 889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.77968253968254e-05,
+      "loss": 0.4929,
+      "step": 890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7793650793650795e-05,
+      "loss": 0.5191,
+      "step": 891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.779047619047619e-05,
+      "loss": 0.5223,
+      "step": 892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7787301587301588e-05,
+      "loss": 0.4858,
+      "step": 893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7784126984126987e-05,
+      "loss": 0.4818,
+      "step": 894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7780952380952384e-05,
+      "loss": 0.4891,
+      "step": 895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.5345,
+      "step": 896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7774603174603176e-05,
+      "loss": 0.5003,
+      "step": 897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7771428571428573e-05,
+      "loss": 0.5001,
+      "step": 898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.776825396825397e-05,
+      "loss": 0.5126,
+      "step": 899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.776507936507937e-05,
+      "loss": 0.5022,
+      "step": 900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.776190476190476e-05,
+      "loss": 0.4499,
+      "step": 901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.775873015873016e-05,
+      "loss": 0.4956,
+      "step": 902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7755555555555558e-05,
+      "loss": 0.5028,
+      "step": 903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7752380952380954e-05,
+      "loss": 0.5204,
+      "step": 904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.774920634920635e-05,
+      "loss": 0.5072,
+      "step": 905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7746031746031747e-05,
+      "loss": 0.4924,
+      "step": 906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7742857142857143e-05,
+      "loss": 0.4823,
+      "step": 907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7739682539682543e-05,
+      "loss": 0.4919,
+      "step": 908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.773650793650794e-05,
+      "loss": 0.5125,
+      "step": 909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7733333333333335e-05,
+      "loss": 0.4804,
+      "step": 910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.773015873015873e-05,
+      "loss": 0.5,
+      "step": 911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7726984126984128e-05,
+      "loss": 0.5274,
+      "step": 912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7723809523809524e-05,
+      "loss": 0.529,
+      "step": 913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7720634920634924e-05,
+      "loss": 0.4957,
+      "step": 914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.771746031746032e-05,
+      "loss": 0.5497,
+      "step": 915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7714285714285717e-05,
+      "loss": 0.5109,
+      "step": 916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7711111111111113e-05,
+      "loss": 0.5354,
+      "step": 917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.770793650793651e-05,
+      "loss": 0.5079,
+      "step": 918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7704761904761905e-05,
+      "loss": 0.5047,
+      "step": 919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7701587301587305e-05,
+      "loss": 0.4922,
+      "step": 920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7698412698412698e-05,
+      "loss": 0.4852,
+      "step": 921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7695238095238098e-05,
+      "loss": 0.5101,
+      "step": 922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7692063492063494e-05,
+      "loss": 0.4669,
+      "step": 923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.768888888888889e-05,
+      "loss": 0.5164,
+      "step": 924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7685714285714287e-05,
+      "loss": 0.5278,
+      "step": 925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7682539682539683e-05,
+      "loss": 0.5509,
+      "step": 926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.767936507936508e-05,
+      "loss": 0.4777,
+      "step": 927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.767619047619048e-05,
+      "loss": 0.5534,
+      "step": 928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7673015873015875e-05,
+      "loss": 0.4637,
+      "step": 929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7669841269841272e-05,
+      "loss": 0.5036,
+      "step": 930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7666666666666668e-05,
+      "loss": 0.5094,
+      "step": 931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7663492063492064e-05,
+      "loss": 0.508,
+      "step": 932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.766031746031746e-05,
+      "loss": 0.5,
+      "step": 933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.765714285714286e-05,
+      "loss": 0.4765,
+      "step": 934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7653968253968253e-05,
+      "loss": 0.5124,
+      "step": 935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7650793650793653e-05,
+      "loss": 0.5532,
+      "step": 936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.764761904761905e-05,
+      "loss": 0.5293,
+      "step": 937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7644444444444446e-05,
+      "loss": 0.5081,
+      "step": 938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7641269841269842e-05,
+      "loss": 0.5621,
+      "step": 939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7638095238095238e-05,
+      "loss": 0.5096,
+      "step": 940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7634920634920635e-05,
+      "loss": 0.5233,
+      "step": 941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7631746031746034e-05,
+      "loss": 0.5399,
+      "step": 942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.762857142857143e-05,
+      "loss": 0.4855,
+      "step": 943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7625396825396827e-05,
+      "loss": 0.4966,
+      "step": 944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7622222222222223e-05,
+      "loss": 0.4835,
+      "step": 945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.761904761904762e-05,
+      "loss": 0.4828,
+      "step": 946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7615873015873016e-05,
+      "loss": 0.491,
+      "step": 947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7612698412698416e-05,
+      "loss": 0.5149,
+      "step": 948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7609523809523812e-05,
+      "loss": 0.513,
+      "step": 949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7606349206349208e-05,
+      "loss": 0.5489,
+      "step": 950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7603174603174605e-05,
+      "loss": 0.563,
+      "step": 951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76e-05,
+      "loss": 0.514,
+      "step": 952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7596825396825397e-05,
+      "loss": 0.4736,
+      "step": 953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7593650793650797e-05,
+      "loss": 0.5274,
+      "step": 954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.759047619047619e-05,
+      "loss": 0.4934,
+      "step": 955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.758730158730159e-05,
+      "loss": 0.5018,
+      "step": 956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7584126984126986e-05,
+      "loss": 0.514,
+      "step": 957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7580952380952382e-05,
+      "loss": 0.4841,
+      "step": 958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7577777777777782e-05,
+      "loss": 0.4687,
+      "step": 959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7574603174603175e-05,
+      "loss": 0.52,
+      "step": 960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.757142857142857e-05,
+      "loss": 0.4951,
+      "step": 961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.756825396825397e-05,
+      "loss": 0.4995,
+      "step": 962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7565079365079367e-05,
+      "loss": 0.4842,
+      "step": 963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7561904761904763e-05,
+      "loss": 0.5486,
+      "step": 964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.755873015873016e-05,
+      "loss": 0.5396,
+      "step": 965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7555555555555556e-05,
+      "loss": 0.5192,
+      "step": 966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7552380952380952e-05,
+      "loss": 0.4609,
+      "step": 967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7549206349206352e-05,
+      "loss": 0.5382,
+      "step": 968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7546031746031745e-05,
+      "loss": 0.4992,
+      "step": 969
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7542857142857145e-05,
+      "loss": 0.4768,
+      "step": 970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.753968253968254e-05,
+      "loss": 0.4657,
+      "step": 971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7536507936507937e-05,
+      "loss": 0.4966,
+      "step": 972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7533333333333337e-05,
+      "loss": 0.4914,
+      "step": 973
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.753015873015873e-05,
+      "loss": 0.5344,
+      "step": 974
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7526984126984126e-05,
+      "loss": 0.5118,
+      "step": 975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7523809523809526e-05,
+      "loss": 0.5046,
+      "step": 976
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7520634920634922e-05,
+      "loss": 0.5198,
+      "step": 977
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.751746031746032e-05,
+      "loss": 0.4884,
+      "step": 978
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7514285714285715e-05,
+      "loss": 0.5079,
+      "step": 979
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.751111111111111e-05,
+      "loss": 0.5222,
+      "step": 980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7507936507936508e-05,
+      "loss": 0.5148,
+      "step": 981
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7504761904761907e-05,
+      "loss": 0.5141,
+      "step": 982
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7501587301587304e-05,
+      "loss": 0.4734,
+      "step": 983
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.74984126984127e-05,
+      "loss": 0.5461,
+      "step": 984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7495238095238096e-05,
+      "loss": 0.4807,
+      "step": 985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7492063492063493e-05,
+      "loss": 0.4909,
+      "step": 986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7488888888888892e-05,
+      "loss": 0.5185,
+      "step": 987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.748571428571429e-05,
+      "loss": 0.519,
+      "step": 988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.748253968253968e-05,
+      "loss": 0.4754,
+      "step": 989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.747936507936508e-05,
+      "loss": 0.5299,
+      "step": 990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7476190476190478e-05,
+      "loss": 0.491,
+      "step": 991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7473015873015874e-05,
+      "loss": 0.5128,
+      "step": 992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7469841269841274e-05,
+      "loss": 0.5348,
+      "step": 993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7466666666666667e-05,
+      "loss": 0.4785,
+      "step": 994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7463492063492066e-05,
+      "loss": 0.5144,
+      "step": 995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7460317460317463e-05,
+      "loss": 0.5033,
+      "step": 996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.745714285714286e-05,
+      "loss": 0.476,
+      "step": 997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7453968253968255e-05,
+      "loss": 0.479,
+      "step": 998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.745079365079365e-05,
+      "loss": 0.4541,
+      "step": 999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7447619047619048e-05,
+      "loss": 0.5186,
+      "step": 1000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7444444444444448e-05,
+      "loss": 0.4686,
+      "step": 1001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7441269841269844e-05,
+      "loss": 0.4909,
+      "step": 1002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.743809523809524e-05,
+      "loss": 0.5093,
+      "step": 1003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7434920634920637e-05,
+      "loss": 0.5227,
+      "step": 1004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7431746031746033e-05,
+      "loss": 0.508,
+      "step": 1005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.742857142857143e-05,
+      "loss": 0.508,
+      "step": 1006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.742539682539683e-05,
+      "loss": 0.4922,
+      "step": 1007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7422222222222222e-05,
+      "loss": 0.505,
+      "step": 1008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.741904761904762e-05,
+      "loss": 0.4861,
+      "step": 1009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7415873015873018e-05,
+      "loss": 0.4684,
+      "step": 1010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7412698412698414e-05,
+      "loss": 0.5175,
+      "step": 1011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.740952380952381e-05,
+      "loss": 0.4594,
+      "step": 1012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7406349206349207e-05,
+      "loss": 0.4864,
+      "step": 1013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7403174603174603e-05,
+      "loss": 0.4959,
+      "step": 1014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 0.5192,
+      "step": 1015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.73968253968254e-05,
+      "loss": 0.4935,
+      "step": 1016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7393650793650795e-05,
+      "loss": 0.4686,
+      "step": 1017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7390476190476192e-05,
+      "loss": 0.5105,
+      "step": 1018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7387301587301588e-05,
+      "loss": 0.4855,
+      "step": 1019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7384126984126984e-05,
+      "loss": 0.4926,
+      "step": 1020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7380952380952384e-05,
+      "loss": 0.4971,
+      "step": 1021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.737777777777778e-05,
+      "loss": 0.5138,
+      "step": 1022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7374603174603177e-05,
+      "loss": 0.5457,
+      "step": 1023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7371428571428573e-05,
+      "loss": 0.4861,
+      "step": 1024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.736825396825397e-05,
+      "loss": 0.4818,
+      "step": 1025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7365079365079366e-05,
+      "loss": 0.4972,
+      "step": 1026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7361904761904765e-05,
+      "loss": 0.5001,
+      "step": 1027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.735873015873016e-05,
+      "loss": 0.4917,
+      "step": 1028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7355555555555558e-05,
+      "loss": 0.4773,
+      "step": 1029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7352380952380954e-05,
+      "loss": 0.4694,
+      "step": 1030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.734920634920635e-05,
+      "loss": 0.4858,
+      "step": 1031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7346031746031747e-05,
+      "loss": 0.4702,
+      "step": 1032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7342857142857143e-05,
+      "loss": 0.4898,
+      "step": 1033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.733968253968254e-05,
+      "loss": 0.5248,
+      "step": 1034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.733650793650794e-05,
+      "loss": 0.5123,
+      "step": 1035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.4929,
+      "step": 1036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7330158730158732e-05,
+      "loss": 0.4779,
+      "step": 1037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7326984126984128e-05,
+      "loss": 0.4747,
+      "step": 1038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7323809523809525e-05,
+      "loss": 0.5445,
+      "step": 1039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.732063492063492e-05,
+      "loss": 0.5154,
+      "step": 1040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.731746031746032e-05,
+      "loss": 0.5224,
+      "step": 1041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7314285714285717e-05,
+      "loss": 0.4924,
+      "step": 1042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7311111111111113e-05,
+      "loss": 0.4674,
+      "step": 1043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.730793650793651e-05,
+      "loss": 0.5092,
+      "step": 1044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7304761904761906e-05,
+      "loss": 0.5223,
+      "step": 1045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7301587301587302e-05,
+      "loss": 0.5102,
+      "step": 1046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.72984126984127e-05,
+      "loss": 0.4965,
+      "step": 1047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7295238095238095e-05,
+      "loss": 0.5191,
+      "step": 1048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7292063492063495e-05,
+      "loss": 0.5378,
+      "step": 1049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.728888888888889e-05,
+      "loss": 0.5157,
+      "step": 1050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7285714285714287e-05,
+      "loss": 0.4635,
+      "step": 1051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7282539682539684e-05,
+      "loss": 0.5162,
+      "step": 1052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.727936507936508e-05,
+      "loss": 0.5118,
+      "step": 1053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7276190476190476e-05,
+      "loss": 0.5038,
+      "step": 1054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7273015873015876e-05,
+      "loss": 0.4994,
+      "step": 1055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7269841269841272e-05,
+      "loss": 0.5306,
+      "step": 1056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.726666666666667e-05,
+      "loss": 0.4797,
+      "step": 1057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7263492063492065e-05,
+      "loss": 0.5221,
+      "step": 1058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.726031746031746e-05,
+      "loss": 0.4965,
+      "step": 1059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7257142857142857e-05,
+      "loss": 0.4928,
+      "step": 1060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7253968253968257e-05,
+      "loss": 0.4819,
+      "step": 1061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.725079365079365e-05,
+      "loss": 0.4976,
+      "step": 1062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.724761904761905e-05,
+      "loss": 0.4939,
+      "step": 1063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7244444444444446e-05,
+      "loss": 0.5091,
+      "step": 1064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7241269841269842e-05,
+      "loss": 0.4991,
+      "step": 1065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.723809523809524e-05,
+      "loss": 0.5492,
+      "step": 1066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7234920634920635e-05,
+      "loss": 0.4978,
+      "step": 1067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.723174603174603e-05,
+      "loss": 0.4934,
+      "step": 1068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.722857142857143e-05,
+      "loss": 0.4899,
+      "step": 1069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7225396825396827e-05,
+      "loss": 0.5099,
+      "step": 1070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7222222222222224e-05,
+      "loss": 0.4907,
+      "step": 1071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.721904761904762e-05,
+      "loss": 0.4775,
+      "step": 1072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7215873015873016e-05,
+      "loss": 0.4885,
+      "step": 1073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7212698412698413e-05,
+      "loss": 0.5049,
+      "step": 1074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7209523809523812e-05,
+      "loss": 0.4807,
+      "step": 1075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.720634920634921e-05,
+      "loss": 0.5012,
+      "step": 1076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7203174603174605e-05,
+      "loss": 0.5148,
+      "step": 1077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.72e-05,
+      "loss": 0.4902,
+      "step": 1078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7196825396825398e-05,
+      "loss": 0.5263,
+      "step": 1079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7193650793650797e-05,
+      "loss": 0.4698,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7190476190476194e-05,
+      "loss": 0.5088,
+      "step": 1081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7187301587301587e-05,
+      "loss": 0.4729,
+      "step": 1082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7184126984126986e-05,
+      "loss": 0.5491,
+      "step": 1083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7180952380952383e-05,
+      "loss": 0.5132,
+      "step": 1084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.717777777777778e-05,
+      "loss": 0.4772,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7174603174603175e-05,
+      "loss": 0.5137,
+      "step": 1086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.717142857142857e-05,
+      "loss": 0.512,
+      "step": 1087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7168253968253968e-05,
+      "loss": 0.5371,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7165079365079368e-05,
+      "loss": 0.5272,
+      "step": 1089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7161904761904764e-05,
+      "loss": 0.5244,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.715873015873016e-05,
+      "loss": 0.4856,
+      "step": 1091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7155555555555557e-05,
+      "loss": 0.4894,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7152380952380953e-05,
+      "loss": 0.5014,
+      "step": 1093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7149206349206353e-05,
+      "loss": 0.5116,
+      "step": 1094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.714603174603175e-05,
+      "loss": 0.5126,
+      "step": 1095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.5195,
+      "step": 1096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.713968253968254e-05,
+      "loss": 0.5258,
+      "step": 1097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7136507936507938e-05,
+      "loss": 0.4828,
+      "step": 1098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7133333333333334e-05,
+      "loss": 0.4729,
+      "step": 1099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7130158730158734e-05,
+      "loss": 0.5269,
+      "step": 1100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7126984126984127e-05,
+      "loss": 0.5027,
+      "step": 1101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7123809523809523e-05,
+      "loss": 0.4629,
+      "step": 1102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7120634920634923e-05,
+      "loss": 0.4993,
+      "step": 1103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.711746031746032e-05,
+      "loss": 0.4745,
+      "step": 1104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7114285714285715e-05,
+      "loss": 0.5091,
+      "step": 1105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7111111111111112e-05,
+      "loss": 0.4719,
+      "step": 1106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7107936507936508e-05,
+      "loss": 0.4981,
+      "step": 1107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7104761904761908e-05,
+      "loss": 0.4946,
+      "step": 1108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7101587301587304e-05,
+      "loss": 0.4991,
+      "step": 1109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.70984126984127e-05,
+      "loss": 0.4998,
+      "step": 1110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7095238095238097e-05,
+      "loss": 0.5223,
+      "step": 1111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7092063492063493e-05,
+      "loss": 0.5145,
+      "step": 1112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.708888888888889e-05,
+      "loss": 0.4828,
+      "step": 1113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.708571428571429e-05,
+      "loss": 0.5279,
+      "step": 1114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7082539682539685e-05,
+      "loss": 0.5129,
+      "step": 1115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.707936507936508e-05,
+      "loss": 0.4835,
+      "step": 1116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7076190476190478e-05,
+      "loss": 0.4956,
+      "step": 1117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7073015873015874e-05,
+      "loss": 0.4911,
+      "step": 1118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.706984126984127e-05,
+      "loss": 0.5209,
+      "step": 1119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 0.4949,
+      "step": 1120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7063492063492063e-05,
+      "loss": 0.4992,
+      "step": 1121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7060317460317463e-05,
+      "loss": 0.5082,
+      "step": 1122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.705714285714286e-05,
+      "loss": 0.5061,
+      "step": 1123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7053968253968256e-05,
+      "loss": 0.5046,
+      "step": 1124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7050793650793652e-05,
+      "loss": 0.4782,
+      "step": 1125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.704761904761905e-05,
+      "loss": 0.5116,
+      "step": 1126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7044444444444445e-05,
+      "loss": 0.5338,
+      "step": 1127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7041269841269844e-05,
+      "loss": 0.4742,
+      "step": 1128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.703809523809524e-05,
+      "loss": 0.4957,
+      "step": 1129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7034920634920637e-05,
+      "loss": 0.483,
+      "step": 1130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7031746031746033e-05,
+      "loss": 0.488,
+      "step": 1131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.702857142857143e-05,
+      "loss": 0.4887,
+      "step": 1132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7025396825396826e-05,
+      "loss": 0.5163,
+      "step": 1133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7022222222222226e-05,
+      "loss": 0.5428,
+      "step": 1134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.701904761904762e-05,
+      "loss": 0.4582,
+      "step": 1135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7015873015873018e-05,
+      "loss": 0.498,
+      "step": 1136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7012698412698415e-05,
+      "loss": 0.5119,
+      "step": 1137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.700952380952381e-05,
+      "loss": 0.4832,
+      "step": 1138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7006349206349207e-05,
+      "loss": 0.498,
+      "step": 1139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7003174603174604e-05,
+      "loss": 0.5363,
+      "step": 1140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7e-05,
+      "loss": 0.5128,
+      "step": 1141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.69968253968254e-05,
+      "loss": 0.5022,
+      "step": 1142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6993650793650796e-05,
+      "loss": 0.497,
+      "step": 1143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6990476190476192e-05,
+      "loss": 0.4929,
+      "step": 1144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.698730158730159e-05,
+      "loss": 0.4814,
+      "step": 1145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6984126984126985e-05,
+      "loss": 0.5072,
+      "step": 1146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.698095238095238e-05,
+      "loss": 0.5087,
+      "step": 1147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.697777777777778e-05,
+      "loss": 0.4781,
+      "step": 1148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6974603174603177e-05,
+      "loss": 0.5186,
+      "step": 1149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6971428571428574e-05,
+      "loss": 0.5337,
+      "step": 1150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.696825396825397e-05,
+      "loss": 0.5309,
+      "step": 1151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6965079365079366e-05,
+      "loss": 0.4724,
+      "step": 1152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6961904761904762e-05,
+      "loss": 0.4857,
+      "step": 1153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6958730158730162e-05,
+      "loss": 0.4968,
+      "step": 1154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6955555555555555e-05,
+      "loss": 0.4997,
+      "step": 1155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6952380952380955e-05,
+      "loss": 0.4864,
+      "step": 1156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.694920634920635e-05,
+      "loss": 0.5216,
+      "step": 1157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6946031746031747e-05,
+      "loss": 0.5251,
+      "step": 1158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6942857142857144e-05,
+      "loss": 0.5206,
+      "step": 1159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.693968253968254e-05,
+      "loss": 0.4791,
+      "step": 1160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6936507936507936e-05,
+      "loss": 0.5249,
+      "step": 1161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6933333333333336e-05,
+      "loss": 0.4886,
+      "step": 1162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6930158730158732e-05,
+      "loss": 0.4796,
+      "step": 1163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.692698412698413e-05,
+      "loss": 0.5173,
+      "step": 1164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6923809523809525e-05,
+      "loss": 0.497,
+      "step": 1165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.692063492063492e-05,
+      "loss": 0.5082,
+      "step": 1166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6917460317460318e-05,
+      "loss": 0.5164,
+      "step": 1167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6914285714285717e-05,
+      "loss": 0.5237,
+      "step": 1168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.691111111111111e-05,
+      "loss": 0.5301,
+      "step": 1169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.690793650793651e-05,
+      "loss": 0.5278,
+      "step": 1170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6904761904761906e-05,
+      "loss": 0.5256,
+      "step": 1171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6901587301587303e-05,
+      "loss": 0.4927,
+      "step": 1172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.68984126984127e-05,
+      "loss": 0.5088,
+      "step": 1173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6895238095238095e-05,
+      "loss": 0.4946,
+      "step": 1174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.689206349206349e-05,
+      "loss": 0.5283,
+      "step": 1175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 0.4721,
+      "step": 1176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6885714285714288e-05,
+      "loss": 0.4759,
+      "step": 1177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6882539682539684e-05,
+      "loss": 0.4386,
+      "step": 1178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.687936507936508e-05,
+      "loss": 0.4555,
+      "step": 1179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6876190476190477e-05,
+      "loss": 0.4845,
+      "step": 1180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6873015873015873e-05,
+      "loss": 0.5241,
+      "step": 1181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6869841269841273e-05,
+      "loss": 0.4731,
+      "step": 1182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.686666666666667e-05,
+      "loss": 0.5,
+      "step": 1183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6863492063492065e-05,
+      "loss": 0.5017,
+      "step": 1184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.686031746031746e-05,
+      "loss": 0.5013,
+      "step": 1185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6857142857142858e-05,
+      "loss": 0.4877,
+      "step": 1186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6853968253968254e-05,
+      "loss": 0.5055,
+      "step": 1187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6850793650793654e-05,
+      "loss": 0.4786,
+      "step": 1188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6847619047619047e-05,
+      "loss": 0.496,
+      "step": 1189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6844444444444447e-05,
+      "loss": 0.4823,
+      "step": 1190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6841269841269843e-05,
+      "loss": 0.5094,
+      "step": 1191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.683809523809524e-05,
+      "loss": 0.4853,
+      "step": 1192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.683492063492064e-05,
+      "loss": 0.466,
+      "step": 1193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6831746031746032e-05,
+      "loss": 0.5128,
+      "step": 1194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6828571428571428e-05,
+      "loss": 0.4971,
+      "step": 1195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6825396825396828e-05,
+      "loss": 0.475,
+      "step": 1196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6822222222222224e-05,
+      "loss": 0.5295,
+      "step": 1197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.681904761904762e-05,
+      "loss": 0.4838,
+      "step": 1198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6815873015873017e-05,
+      "loss": 0.4981,
+      "step": 1199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6812698412698413e-05,
+      "loss": 0.469,
+      "step": 1200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.680952380952381e-05,
+      "loss": 0.5322,
+      "step": 1201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.680634920634921e-05,
+      "loss": 0.456,
+      "step": 1202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6803174603174602e-05,
+      "loss": 0.5061,
+      "step": 1203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.5162,
+      "step": 1204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6796825396825398e-05,
+      "loss": 0.5107,
+      "step": 1205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6793650793650794e-05,
+      "loss": 0.4885,
+      "step": 1206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6790476190476194e-05,
+      "loss": 0.5138,
+      "step": 1207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6787301587301587e-05,
+      "loss": 0.5062,
+      "step": 1208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6784126984126983e-05,
+      "loss": 0.5041,
+      "step": 1209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6780952380952383e-05,
+      "loss": 0.496,
+      "step": 1210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.677777777777778e-05,
+      "loss": 0.4706,
+      "step": 1211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6774603174603176e-05,
+      "loss": 0.483,
+      "step": 1212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6771428571428572e-05,
+      "loss": 0.4948,
+      "step": 1213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.676825396825397e-05,
+      "loss": 0.5147,
+      "step": 1214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6765079365079368e-05,
+      "loss": 0.501,
+      "step": 1215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6761904761904764e-05,
+      "loss": 0.5307,
+      "step": 1216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.675873015873016e-05,
+      "loss": 0.5015,
+      "step": 1217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6755555555555557e-05,
+      "loss": 0.4795,
+      "step": 1218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6752380952380953e-05,
+      "loss": 0.4979,
+      "step": 1219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.674920634920635e-05,
+      "loss": 0.5083,
+      "step": 1220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.674603174603175e-05,
+      "loss": 0.5036,
+      "step": 1221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6742857142857146e-05,
+      "loss": 0.5402,
+      "step": 1222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.673968253968254e-05,
+      "loss": 0.4934,
+      "step": 1223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6736507936507938e-05,
+      "loss": 0.5021,
+      "step": 1224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6733333333333335e-05,
+      "loss": 0.4985,
+      "step": 1225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673015873015873e-05,
+      "loss": 0.4646,
+      "step": 1226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.672698412698413e-05,
+      "loss": 0.4744,
+      "step": 1227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6723809523809524e-05,
+      "loss": 0.4818,
+      "step": 1228
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6720634920634923e-05,
+      "loss": 0.4956,
+      "step": 1229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.671746031746032e-05,
+      "loss": 0.5281,
+      "step": 1230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6714285714285716e-05,
+      "loss": 0.4949,
+      "step": 1231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6711111111111112e-05,
+      "loss": 0.4844,
+      "step": 1232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.670793650793651e-05,
+      "loss": 0.5159,
+      "step": 1233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6704761904761905e-05,
+      "loss": 0.5156,
+      "step": 1234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6701587301587305e-05,
+      "loss": 0.4773,
+      "step": 1235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.66984126984127e-05,
+      "loss": 0.5035,
+      "step": 1236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6695238095238097e-05,
+      "loss": 0.5115,
+      "step": 1237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6692063492063494e-05,
+      "loss": 0.4772,
+      "step": 1238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.668888888888889e-05,
+      "loss": 0.509,
+      "step": 1239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6685714285714286e-05,
+      "loss": 0.5163,
+      "step": 1240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6682539682539686e-05,
+      "loss": 0.5089,
+      "step": 1241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.667936507936508e-05,
+      "loss": 0.5314,
+      "step": 1242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.667619047619048e-05,
+      "loss": 0.4834,
+      "step": 1243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6673015873015875e-05,
+      "loss": 0.4667,
+      "step": 1244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.666984126984127e-05,
+      "loss": 0.4924,
+      "step": 1245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4888,
+      "step": 1246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6663492063492064e-05,
+      "loss": 0.5281,
+      "step": 1247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.666031746031746e-05,
+      "loss": 0.5085,
+      "step": 1248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665714285714286e-05,
+      "loss": 0.4835,
+      "step": 1249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6653968253968256e-05,
+      "loss": 0.4882,
+      "step": 1250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6650793650793652e-05,
+      "loss": 0.4593,
+      "step": 1251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664761904761905e-05,
+      "loss": 0.5769,
+      "step": 1252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6644444444444445e-05,
+      "loss": 0.4205,
+      "step": 1253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664126984126984e-05,
+      "loss": 0.5305,
+      "step": 1254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663809523809524e-05,
+      "loss": 0.5276,
+      "step": 1255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6634920634920637e-05,
+      "loss": 0.5038,
+      "step": 1256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6631746031746034e-05,
+      "loss": 0.486,
+      "step": 1257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662857142857143e-05,
+      "loss": 0.4611,
+      "step": 1258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6625396825396826e-05,
+      "loss": 0.4825,
+      "step": 1259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6622222222222223e-05,
+      "loss": 0.5035,
+      "step": 1260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6619047619047622e-05,
+      "loss": 0.4875,
+      "step": 1261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6615873015873015e-05,
+      "loss": 0.5205,
+      "step": 1262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6612698412698415e-05,
+      "loss": 0.5124,
+      "step": 1263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660952380952381e-05,
+      "loss": 0.4947,
+      "step": 1264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6606349206349208e-05,
+      "loss": 0.5059,
+      "step": 1265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6603174603174604e-05,
+      "loss": 0.4981,
+      "step": 1266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.66e-05,
+      "loss": 0.5568,
+      "step": 1267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6596825396825397e-05,
+      "loss": 0.4891,
+      "step": 1268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6593650793650796e-05,
+      "loss": 0.5067,
+      "step": 1269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6590476190476193e-05,
+      "loss": 0.5017,
+      "step": 1270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.658730158730159e-05,
+      "loss": 0.4858,
+      "step": 1271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6584126984126985e-05,
+      "loss": 0.5318,
+      "step": 1272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.658095238095238e-05,
+      "loss": 0.5208,
+      "step": 1273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6577777777777778e-05,
+      "loss": 0.4472,
+      "step": 1274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6574603174603178e-05,
+      "loss": 0.4816,
+      "step": 1275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6571428571428574e-05,
+      "loss": 0.4898,
+      "step": 1276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656825396825397e-05,
+      "loss": 0.5198,
+      "step": 1277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6565079365079367e-05,
+      "loss": 0.5336,
+      "step": 1278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6561904761904763e-05,
+      "loss": 0.4884,
+      "step": 1279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.655873015873016e-05,
+      "loss": 0.5216,
+      "step": 1280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6555555555555556e-05,
+      "loss": 0.4983,
+      "step": 1281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6552380952380952e-05,
+      "loss": 0.4947,
+      "step": 1282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654920634920635e-05,
+      "loss": 0.4607,
+      "step": 1283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6546031746031748e-05,
+      "loss": 0.5272,
+      "step": 1284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6542857142857144e-05,
+      "loss": 0.4982,
+      "step": 1285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.653968253968254e-05,
+      "loss": 0.4745,
+      "step": 1286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6536507936507937e-05,
+      "loss": 0.4915,
+      "step": 1287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 0.4884,
+      "step": 1288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6530158730158733e-05,
+      "loss": 0.4462,
+      "step": 1289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.652698412698413e-05,
+      "loss": 0.4962,
+      "step": 1290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6523809523809525e-05,
+      "loss": 0.5023,
+      "step": 1291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6520634920634922e-05,
+      "loss": 0.5297,
+      "step": 1292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6517460317460318e-05,
+      "loss": 0.4878,
+      "step": 1293
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6514285714285714e-05,
+      "loss": 0.496,
+      "step": 1294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6511111111111114e-05,
+      "loss": 0.4671,
+      "step": 1295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6507936507936507e-05,
+      "loss": 0.4895,
+      "step": 1296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6504761904761907e-05,
+      "loss": 0.4873,
+      "step": 1297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6501587301587303e-05,
+      "loss": 0.4904,
+      "step": 1298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.64984126984127e-05,
+      "loss": 0.5224,
+      "step": 1299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.64952380952381e-05,
+      "loss": 0.4965,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6492063492063492e-05,
+      "loss": 0.5023,
+      "step": 1301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.648888888888889e-05,
+      "loss": 0.4679,
+      "step": 1302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6485714285714288e-05,
+      "loss": 0.5112,
+      "step": 1303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6482539682539684e-05,
+      "loss": 0.4496,
+      "step": 1304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.647936507936508e-05,
+      "loss": 0.4727,
+      "step": 1305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6476190476190477e-05,
+      "loss": 0.5238,
+      "step": 1306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6473015873015873e-05,
+      "loss": 0.4709,
+      "step": 1307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646984126984127e-05,
+      "loss": 0.5257,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646666666666667e-05,
+      "loss": 0.5061,
+      "step": 1309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6463492063492066e-05,
+      "loss": 0.5192,
+      "step": 1310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6460317460317462e-05,
+      "loss": 0.4946,
+      "step": 1311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.645714285714286e-05,
+      "loss": 0.4639,
+      "step": 1312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6453968253968255e-05,
+      "loss": 0.5034,
+      "step": 1313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6450793650793654e-05,
+      "loss": 0.5438,
+      "step": 1314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.644761904761905e-05,
+      "loss": 0.5215,
+      "step": 1315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6444444444444444e-05,
+      "loss": 0.4981,
+      "step": 1316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6441269841269843e-05,
+      "loss": 0.4648,
+      "step": 1317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.643809523809524e-05,
+      "loss": 0.5118,
+      "step": 1318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6434920634920636e-05,
+      "loss": 0.5222,
+      "step": 1319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6431746031746032e-05,
+      "loss": 0.4777,
+      "step": 1320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.4739,
+      "step": 1321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6425396825396825e-05,
+      "loss": 0.5088,
+      "step": 1322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6422222222222225e-05,
+      "loss": 0.4716,
+      "step": 1323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.641904761904762e-05,
+      "loss": 0.5015,
+      "step": 1324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6415873015873017e-05,
+      "loss": 0.5118,
+      "step": 1325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6412698412698414e-05,
+      "loss": 0.5275,
+      "step": 1326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.640952380952381e-05,
+      "loss": 0.495,
+      "step": 1327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.640634920634921e-05,
+      "loss": 0.513,
+      "step": 1328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6403174603174606e-05,
+      "loss": 0.5241,
+      "step": 1329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.64e-05,
+      "loss": 0.5199,
+      "step": 1330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.63968253968254e-05,
+      "loss": 0.5161,
+      "step": 1331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6393650793650795e-05,
+      "loss": 0.5122,
+      "step": 1332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.639047619047619e-05,
+      "loss": 0.4754,
+      "step": 1333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.638730158730159e-05,
+      "loss": 0.5486,
+      "step": 1334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6384126984126984e-05,
+      "loss": 0.4948,
+      "step": 1335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6380952380952384e-05,
+      "loss": 0.4985,
+      "step": 1336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.637777777777778e-05,
+      "loss": 0.4918,
+      "step": 1337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6374603174603176e-05,
+      "loss": 0.5002,
+      "step": 1338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6371428571428572e-05,
+      "loss": 0.4655,
+      "step": 1339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.636825396825397e-05,
+      "loss": 0.4962,
+      "step": 1340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6365079365079365e-05,
+      "loss": 0.4994,
+      "step": 1341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6361904761904765e-05,
+      "loss": 0.5323,
+      "step": 1342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.635873015873016e-05,
+      "loss": 0.529,
+      "step": 1343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6355555555555557e-05,
+      "loss": 0.5014,
+      "step": 1344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6352380952380954e-05,
+      "loss": 0.4753,
+      "step": 1345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.634920634920635e-05,
+      "loss": 0.5294,
+      "step": 1346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6346031746031746e-05,
+      "loss": 0.4908,
+      "step": 1347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6342857142857146e-05,
+      "loss": 0.4669,
+      "step": 1348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6339682539682542e-05,
+      "loss": 0.482,
+      "step": 1349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.633650793650794e-05,
+      "loss": 0.4815,
+      "step": 1350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6333333333333335e-05,
+      "loss": 0.4916,
+      "step": 1351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.633015873015873e-05,
+      "loss": 0.5155,
+      "step": 1352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6326984126984128e-05,
+      "loss": 0.4852,
+      "step": 1353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6323809523809527e-05,
+      "loss": 0.5126,
+      "step": 1354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.632063492063492e-05,
+      "loss": 0.4662,
+      "step": 1355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.631746031746032e-05,
+      "loss": 0.4786,
+      "step": 1356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6314285714285716e-05,
+      "loss": 0.4905,
+      "step": 1357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6311111111111113e-05,
+      "loss": 0.5126,
+      "step": 1358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.630793650793651e-05,
+      "loss": 0.484,
+      "step": 1359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6304761904761905e-05,
+      "loss": 0.5221,
+      "step": 1360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.63015873015873e-05,
+      "loss": 0.4881,
+      "step": 1361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.62984126984127e-05,
+      "loss": 0.5024,
+      "step": 1362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6295238095238098e-05,
+      "loss": 0.5052,
+      "step": 1363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6292063492063494e-05,
+      "loss": 0.4774,
+      "step": 1364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.628888888888889e-05,
+      "loss": 0.5115,
+      "step": 1365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6285714285714287e-05,
+      "loss": 0.4985,
+      "step": 1366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6282539682539683e-05,
+      "loss": 0.4727,
+      "step": 1367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6279365079365083e-05,
+      "loss": 0.5404,
+      "step": 1368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6276190476190476e-05,
+      "loss": 0.4519,
+      "step": 1369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6273015873015875e-05,
+      "loss": 0.5127,
+      "step": 1370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.626984126984127e-05,
+      "loss": 0.5244,
+      "step": 1371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 0.4669,
+      "step": 1372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6263492063492064e-05,
+      "loss": 0.4956,
+      "step": 1373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.626031746031746e-05,
+      "loss": 0.5005,
+      "step": 1374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6257142857142857e-05,
+      "loss": 0.5047,
+      "step": 1375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6253968253968257e-05,
+      "loss": 0.5354,
+      "step": 1376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6250793650793653e-05,
+      "loss": 0.4937,
+      "step": 1377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.624761904761905e-05,
+      "loss": 0.4805,
+      "step": 1378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6244444444444446e-05,
+      "loss": 0.4816,
+      "step": 1379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6241269841269842e-05,
+      "loss": 0.4839,
+      "step": 1380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6238095238095238e-05,
+      "loss": 0.5207,
+      "step": 1381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6234920634920638e-05,
+      "loss": 0.4975,
+      "step": 1382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6231746031746034e-05,
+      "loss": 0.4846,
+      "step": 1383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.622857142857143e-05,
+      "loss": 0.4829,
+      "step": 1384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6225396825396827e-05,
+      "loss": 0.4858,
+      "step": 1385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6222222222222223e-05,
+      "loss": 0.4823,
+      "step": 1386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.621904761904762e-05,
+      "loss": 0.5447,
+      "step": 1387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.621587301587302e-05,
+      "loss": 0.5154,
+      "step": 1388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6212698412698412e-05,
+      "loss": 0.4856,
+      "step": 1389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6209523809523812e-05,
+      "loss": 0.4833,
+      "step": 1390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6206349206349208e-05,
+      "loss": 0.4659,
+      "step": 1391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6203174603174604e-05,
+      "loss": 0.4727,
+      "step": 1392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.62e-05,
+      "loss": 0.4885,
+      "step": 1393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6196825396825397e-05,
+      "loss": 0.5185,
+      "step": 1394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6193650793650793e-05,
+      "loss": 0.4795,
+      "step": 1395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 0.4731,
+      "step": 1396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.618730158730159e-05,
+      "loss": 0.4815,
+      "step": 1397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6184126984126986e-05,
+      "loss": 0.5192,
+      "step": 1398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6180952380952382e-05,
+      "loss": 0.4597,
+      "step": 1399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.617777777777778e-05,
+      "loss": 0.5293,
+      "step": 1400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6174603174603175e-05,
+      "loss": 0.4802,
+      "step": 1401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6171428571428574e-05,
+      "loss": 0.4564,
+      "step": 1402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6168253968253967e-05,
+      "loss": 0.4794,
+      "step": 1403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6165079365079367e-05,
+      "loss": 0.4834,
+      "step": 1404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6161904761904763e-05,
+      "loss": 0.4602,
+      "step": 1405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.615873015873016e-05,
+      "loss": 0.497,
+      "step": 1406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6155555555555556e-05,
+      "loss": 0.5312,
+      "step": 1407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6152380952380952e-05,
+      "loss": 0.4923,
+      "step": 1408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.614920634920635e-05,
+      "loss": 0.5167,
+      "step": 1409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.614603174603175e-05,
+      "loss": 0.4856,
+      "step": 1410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6142857142857145e-05,
+      "loss": 0.4609,
+      "step": 1411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.613968253968254e-05,
+      "loss": 0.5136,
+      "step": 1412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6136507936507937e-05,
+      "loss": 0.5281,
+      "step": 1413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6133333333333334e-05,
+      "loss": 0.4955,
+      "step": 1414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.613015873015873e-05,
+      "loss": 0.5258,
+      "step": 1415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.612698412698413e-05,
+      "loss": 0.4749,
+      "step": 1416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6123809523809526e-05,
+      "loss": 0.5098,
+      "step": 1417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6120634920634922e-05,
+      "loss": 0.4798,
+      "step": 1418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.611746031746032e-05,
+      "loss": 0.4882,
+      "step": 1419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6114285714285715e-05,
+      "loss": 0.4988,
+      "step": 1420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6111111111111115e-05,
+      "loss": 0.519,
+      "step": 1421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.610793650793651e-05,
+      "loss": 0.5012,
+      "step": 1422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6104761904761904e-05,
+      "loss": 0.4773,
+      "step": 1423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6101587301587304e-05,
+      "loss": 0.5073,
+      "step": 1424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.60984126984127e-05,
+      "loss": 0.462,
+      "step": 1425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6095238095238096e-05,
+      "loss": 0.5029,
+      "step": 1426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6092063492063496e-05,
+      "loss": 0.5005,
+      "step": 1427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.608888888888889e-05,
+      "loss": 0.5329,
+      "step": 1428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6085714285714285e-05,
+      "loss": 0.4684,
+      "step": 1429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6082539682539685e-05,
+      "loss": 0.5253,
+      "step": 1430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.607936507936508e-05,
+      "loss": 0.4889,
+      "step": 1431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6076190476190477e-05,
+      "loss": 0.504,
+      "step": 1432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6073015873015874e-05,
+      "loss": 0.4934,
+      "step": 1433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.606984126984127e-05,
+      "loss": 0.4843,
+      "step": 1434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.606666666666667e-05,
+      "loss": 0.4966,
+      "step": 1435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6063492063492066e-05,
+      "loss": 0.4816,
+      "step": 1436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.606031746031746e-05,
+      "loss": 0.4933,
+      "step": 1437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.605714285714286e-05,
+      "loss": 0.5092,
+      "step": 1438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6053968253968255e-05,
+      "loss": 0.5033,
+      "step": 1439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.605079365079365e-05,
+      "loss": 0.4996,
+      "step": 1440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.604761904761905e-05,
+      "loss": 0.5181,
+      "step": 1441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6044444444444444e-05,
+      "loss": 0.4859,
+      "step": 1442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.604126984126984e-05,
+      "loss": 0.4872,
+      "step": 1443
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.603809523809524e-05,
+      "loss": 0.5085,
+      "step": 1444
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6034920634920636e-05,
+      "loss": 0.4996,
+      "step": 1445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6031746031746033e-05,
+      "loss": 0.5174,
+      "step": 1446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.602857142857143e-05,
+      "loss": 0.4781,
+      "step": 1447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6025396825396825e-05,
+      "loss": 0.4979,
+      "step": 1448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6022222222222225e-05,
+      "loss": 0.4833,
+      "step": 1449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.601904761904762e-05,
+      "loss": 0.4764,
+      "step": 1450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6015873015873018e-05,
+      "loss": 0.4792,
+      "step": 1451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6012698412698414e-05,
+      "loss": 0.4758,
+      "step": 1452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.600952380952381e-05,
+      "loss": 0.5183,
+      "step": 1453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6006349206349207e-05,
+      "loss": 0.5156,
+      "step": 1454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6003174603174606e-05,
+      "loss": 0.4723,
+      "step": 1455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.5353,
+      "step": 1456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5996825396825396e-05,
+      "loss": 0.4794,
+      "step": 1457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5993650793650795e-05,
+      "loss": 0.4845,
+      "step": 1458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.599047619047619e-05,
+      "loss": 0.4465,
+      "step": 1459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5987301587301588e-05,
+      "loss": 0.4957,
+      "step": 1460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5984126984126988e-05,
+      "loss": 0.4942,
+      "step": 1461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.598095238095238e-05,
+      "loss": 0.5239,
+      "step": 1462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.597777777777778e-05,
+      "loss": 0.5165,
+      "step": 1463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5974603174603177e-05,
+      "loss": 0.4964,
+      "step": 1464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5971428571428573e-05,
+      "loss": 0.506,
+      "step": 1465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.596825396825397e-05,
+      "loss": 0.502,
+      "step": 1466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5965079365079366e-05,
+      "loss": 0.5747,
+      "step": 1467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5961904761904762e-05,
+      "loss": 0.5113,
+      "step": 1468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.595873015873016e-05,
+      "loss": 0.4565,
+      "step": 1469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5955555555555558e-05,
+      "loss": 0.5229,
+      "step": 1470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5952380952380954e-05,
+      "loss": 0.4741,
+      "step": 1471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.594920634920635e-05,
+      "loss": 0.5068,
+      "step": 1472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5946031746031747e-05,
+      "loss": 0.5326,
+      "step": 1473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5942857142857143e-05,
+      "loss": 0.4682,
+      "step": 1474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5939682539682543e-05,
+      "loss": 0.4857,
+      "step": 1475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5936507936507936e-05,
+      "loss": 0.4882,
+      "step": 1476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5933333333333336e-05,
+      "loss": 0.4777,
+      "step": 1477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5930158730158732e-05,
+      "loss": 0.5287,
+      "step": 1478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5926984126984128e-05,
+      "loss": 0.5071,
+      "step": 1479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5923809523809524e-05,
+      "loss": 0.4664,
+      "step": 1480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.592063492063492e-05,
+      "loss": 0.4768,
+      "step": 1481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5917460317460317e-05,
+      "loss": 0.4715,
+      "step": 1482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5914285714285717e-05,
+      "loss": 0.5399,
+      "step": 1483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5911111111111113e-05,
+      "loss": 0.4896,
+      "step": 1484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.590793650793651e-05,
+      "loss": 0.499,
+      "step": 1485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5904761904761906e-05,
+      "loss": 0.4872,
+      "step": 1486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5901587301587302e-05,
+      "loss": 0.5037,
+      "step": 1487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.58984126984127e-05,
+      "loss": 0.4585,
+      "step": 1488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5895238095238098e-05,
+      "loss": 0.441,
+      "step": 1489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5892063492063494e-05,
+      "loss": 0.5168,
+      "step": 1490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.588888888888889e-05,
+      "loss": 0.4987,
+      "step": 1491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5885714285714287e-05,
+      "loss": 0.4741,
+      "step": 1492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5882539682539683e-05,
+      "loss": 0.4705,
+      "step": 1493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.587936507936508e-05,
+      "loss": 0.4853,
+      "step": 1494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.587619047619048e-05,
+      "loss": 0.5114,
+      "step": 1495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5873015873015872e-05,
+      "loss": 0.4813,
+      "step": 1496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5869841269841272e-05,
+      "loss": 0.4824,
+      "step": 1497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.586666666666667e-05,
+      "loss": 0.4736,
+      "step": 1498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5863492063492065e-05,
+      "loss": 0.4836,
+      "step": 1499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.586031746031746e-05,
+      "loss": 0.4739,
+      "step": 1500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5857142857142857e-05,
+      "loss": 0.5258,
+      "step": 1501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5853968253968254e-05,
+      "loss": 0.4702,
+      "step": 1502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5850793650793653e-05,
+      "loss": 0.4557,
+      "step": 1503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.584761904761905e-05,
+      "loss": 0.4877,
+      "step": 1504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5844444444444446e-05,
+      "loss": 0.5301,
+      "step": 1505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5841269841269842e-05,
+      "loss": 0.5368,
+      "step": 1506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.583809523809524e-05,
+      "loss": 0.4992,
+      "step": 1507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5834920634920635e-05,
+      "loss": 0.4842,
+      "step": 1508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5831746031746035e-05,
+      "loss": 0.5256,
+      "step": 1509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.582857142857143e-05,
+      "loss": 0.4825,
+      "step": 1510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5825396825396827e-05,
+      "loss": 0.5117,
+      "step": 1511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5822222222222224e-05,
+      "loss": 0.4767,
+      "step": 1512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.581904761904762e-05,
+      "loss": 0.5027,
+      "step": 1513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5815873015873016e-05,
+      "loss": 0.489,
+      "step": 1514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5812698412698416e-05,
+      "loss": 0.4803,
+      "step": 1515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.580952380952381e-05,
+      "loss": 0.4951,
+      "step": 1516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.580634920634921e-05,
+      "loss": 0.4827,
+      "step": 1517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5803174603174605e-05,
+      "loss": 0.4807,
+      "step": 1518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.58e-05,
+      "loss": 0.4513,
+      "step": 1519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5796825396825398e-05,
+      "loss": 0.5226,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5793650793650794e-05,
+      "loss": 0.4962,
+      "step": 1521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.579047619047619e-05,
+      "loss": 0.4835,
+      "step": 1522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.578730158730159e-05,
+      "loss": 0.493,
+      "step": 1523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5784126984126986e-05,
+      "loss": 0.4836,
+      "step": 1524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5780952380952382e-05,
+      "loss": 0.4802,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.577777777777778e-05,
+      "loss": 0.541,
+      "step": 1526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5774603174603175e-05,
+      "loss": 0.5117,
+      "step": 1527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.577142857142857e-05,
+      "loss": 0.4706,
+      "step": 1528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.576825396825397e-05,
+      "loss": 0.5083,
+      "step": 1529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5765079365079364e-05,
+      "loss": 0.4399,
+      "step": 1530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5761904761904764e-05,
+      "loss": 0.5064,
+      "step": 1531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.575873015873016e-05,
+      "loss": 0.5036,
+      "step": 1532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5755555555555556e-05,
+      "loss": 0.4952,
+      "step": 1533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5752380952380956e-05,
+      "loss": 0.4789,
+      "step": 1534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.574920634920635e-05,
+      "loss": 0.4704,
+      "step": 1535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5746031746031745e-05,
+      "loss": 0.444,
+      "step": 1536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5742857142857145e-05,
+      "loss": 0.4612,
+      "step": 1537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.573968253968254e-05,
+      "loss": 0.5288,
+      "step": 1538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5736507936507938e-05,
+      "loss": 0.4971,
+      "step": 1539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 0.4565,
+      "step": 1540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.573015873015873e-05,
+      "loss": 0.4897,
+      "step": 1541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5726984126984127e-05,
+      "loss": 0.4607,
+      "step": 1542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5723809523809526e-05,
+      "loss": 0.48,
+      "step": 1543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5720634920634923e-05,
+      "loss": 0.488,
+      "step": 1544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.571746031746032e-05,
+      "loss": 0.476,
+      "step": 1545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.5055,
+      "step": 1546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.571111111111111e-05,
+      "loss": 0.4998,
+      "step": 1547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.570793650793651e-05,
+      "loss": 0.4851,
+      "step": 1548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5704761904761908e-05,
+      "loss": 0.5046,
+      "step": 1549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.57015873015873e-05,
+      "loss": 0.4633,
+      "step": 1550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.56984126984127e-05,
+      "loss": 0.4819,
+      "step": 1551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5695238095238097e-05,
+      "loss": 0.4522,
+      "step": 1552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5692063492063493e-05,
+      "loss": 0.4673,
+      "step": 1553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5688888888888893e-05,
+      "loss": 0.4879,
+      "step": 1554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5685714285714286e-05,
+      "loss": 0.5069,
+      "step": 1555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5682539682539685e-05,
+      "loss": 0.5043,
+      "step": 1556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.567936507936508e-05,
+      "loss": 0.4863,
+      "step": 1557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5676190476190478e-05,
+      "loss": 0.4745,
+      "step": 1558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5673015873015874e-05,
+      "loss": 0.4849,
+      "step": 1559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.566984126984127e-05,
+      "loss": 0.5056,
+      "step": 1560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5666666666666667e-05,
+      "loss": 0.5083,
+      "step": 1561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5663492063492067e-05,
+      "loss": 0.4979,
+      "step": 1562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5660317460317463e-05,
+      "loss": 0.4767,
+      "step": 1563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5657142857142856e-05,
+      "loss": 0.4705,
+      "step": 1564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5653968253968256e-05,
+      "loss": 0.4799,
+      "step": 1565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5650793650793652e-05,
+      "loss": 0.4903,
+      "step": 1566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5647619047619048e-05,
+      "loss": 0.4917,
+      "step": 1567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5644444444444448e-05,
+      "loss": 0.5301,
+      "step": 1568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.564126984126984e-05,
+      "loss": 0.4989,
+      "step": 1569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.563809523809524e-05,
+      "loss": 0.4631,
+      "step": 1570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5634920634920637e-05,
+      "loss": 0.5257,
+      "step": 1571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5631746031746033e-05,
+      "loss": 0.483,
+      "step": 1572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.562857142857143e-05,
+      "loss": 0.4818,
+      "step": 1573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5625396825396826e-05,
+      "loss": 0.4919,
+      "step": 1574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5622222222222222e-05,
+      "loss": 0.5067,
+      "step": 1575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5619047619047622e-05,
+      "loss": 0.509,
+      "step": 1576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5615873015873018e-05,
+      "loss": 0.4867,
+      "step": 1577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5612698412698414e-05,
+      "loss": 0.5213,
+      "step": 1578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.560952380952381e-05,
+      "loss": 0.4522,
+      "step": 1579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5606349206349207e-05,
+      "loss": 0.4652,
+      "step": 1580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5603174603174603e-05,
+      "loss": 0.4796,
+      "step": 1581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 0.4399,
+      "step": 1582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.55968253968254e-05,
+      "loss": 0.4817,
+      "step": 1583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5593650793650796e-05,
+      "loss": 0.4857,
+      "step": 1584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5590476190476192e-05,
+      "loss": 0.533,
+      "step": 1585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.558730158730159e-05,
+      "loss": 0.4868,
+      "step": 1586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5584126984126985e-05,
+      "loss": 0.4446,
+      "step": 1587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5580952380952384e-05,
+      "loss": 0.5232,
+      "step": 1588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5577777777777777e-05,
+      "loss": 0.5056,
+      "step": 1589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5574603174603177e-05,
+      "loss": 0.512,
+      "step": 1590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5571428571428573e-05,
+      "loss": 0.5321,
+      "step": 1591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.556825396825397e-05,
+      "loss": 0.4792,
+      "step": 1592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5565079365079366e-05,
+      "loss": 0.5133,
+      "step": 1593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5561904761904762e-05,
+      "loss": 0.5262,
+      "step": 1594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.555873015873016e-05,
+      "loss": 0.5096,
+      "step": 1595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.4667,
+      "step": 1596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5552380952380955e-05,
+      "loss": 0.4508,
+      "step": 1597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.554920634920635e-05,
+      "loss": 0.4787,
+      "step": 1598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5546031746031747e-05,
+      "loss": 0.5012,
+      "step": 1599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5542857142857144e-05,
+      "loss": 0.4775,
+      "step": 1600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.553968253968254e-05,
+      "loss": 0.4878,
+      "step": 1601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.553650793650794e-05,
+      "loss": 0.4775,
+      "step": 1602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5533333333333333e-05,
+      "loss": 0.4979,
+      "step": 1603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5530158730158732e-05,
+      "loss": 0.4971,
+      "step": 1604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.552698412698413e-05,
+      "loss": 0.444,
+      "step": 1605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5523809523809525e-05,
+      "loss": 0.5108,
+      "step": 1606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.552063492063492e-05,
+      "loss": 0.5081,
+      "step": 1607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5517460317460318e-05,
+      "loss": 0.4789,
+      "step": 1608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5514285714285714e-05,
+      "loss": 0.5086,
+      "step": 1609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5511111111111114e-05,
+      "loss": 0.5037,
+      "step": 1610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.550793650793651e-05,
+      "loss": 0.5107,
+      "step": 1611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5504761904761906e-05,
+      "loss": 0.4707,
+      "step": 1612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5501587301587303e-05,
+      "loss": 0.4935,
+      "step": 1613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.54984126984127e-05,
+      "loss": 0.4659,
+      "step": 1614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5495238095238095e-05,
+      "loss": 0.4827,
+      "step": 1615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5492063492063495e-05,
+      "loss": 0.5127,
+      "step": 1616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548888888888889e-05,
+      "loss": 0.4637,
+      "step": 1617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5485714285714287e-05,
+      "loss": 0.5193,
+      "step": 1618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5482539682539684e-05,
+      "loss": 0.5131,
+      "step": 1619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.547936507936508e-05,
+      "loss": 0.4812,
+      "step": 1620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.5107,
+      "step": 1621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5473015873015876e-05,
+      "loss": 0.5299,
+      "step": 1622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.546984126984127e-05,
+      "loss": 0.5041,
+      "step": 1623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 0.4783,
+      "step": 1624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5463492063492065e-05,
+      "loss": 0.5214,
+      "step": 1625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.546031746031746e-05,
+      "loss": 0.5033,
+      "step": 1626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545714285714286e-05,
+      "loss": 0.4888,
+      "step": 1627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5453968253968254e-05,
+      "loss": 0.4878,
+      "step": 1628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545079365079365e-05,
+      "loss": 0.4926,
+      "step": 1629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.544761904761905e-05,
+      "loss": 0.4605,
+      "step": 1630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5444444444444446e-05,
+      "loss": 0.4545,
+      "step": 1631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5441269841269843e-05,
+      "loss": 0.5062,
+      "step": 1632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.543809523809524e-05,
+      "loss": 0.4616,
+      "step": 1633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5434920634920635e-05,
+      "loss": 0.4751,
+      "step": 1634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.543174603174603e-05,
+      "loss": 0.4546,
+      "step": 1635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.542857142857143e-05,
+      "loss": 0.4714,
+      "step": 1636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5425396825396824e-05,
+      "loss": 0.4818,
+      "step": 1637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5422222222222224e-05,
+      "loss": 0.502,
+      "step": 1638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.541904761904762e-05,
+      "loss": 0.5109,
+      "step": 1639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5415873015873017e-05,
+      "loss": 0.4705,
+      "step": 1640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5412698412698416e-05,
+      "loss": 0.4841,
+      "step": 1641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.540952380952381e-05,
+      "loss": 0.5037,
+      "step": 1642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5406349206349206e-05,
+      "loss": 0.5119,
+      "step": 1643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5403174603174605e-05,
+      "loss": 0.4841,
+      "step": 1644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.54e-05,
+      "loss": 0.4611,
+      "step": 1645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5396825396825398e-05,
+      "loss": 0.4524,
+      "step": 1646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5393650793650794e-05,
+      "loss": 0.5197,
+      "step": 1647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.539047619047619e-05,
+      "loss": 0.5103,
+      "step": 1648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5387301587301587e-05,
+      "loss": 0.4847,
+      "step": 1649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5384126984126987e-05,
+      "loss": 0.4716,
+      "step": 1650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5380952380952383e-05,
+      "loss": 0.4637,
+      "step": 1651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.537777777777778e-05,
+      "loss": 0.5109,
+      "step": 1652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5374603174603176e-05,
+      "loss": 0.4679,
+      "step": 1653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5371428571428572e-05,
+      "loss": 0.5014,
+      "step": 1654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.536825396825397e-05,
+      "loss": 0.4463,
+      "step": 1655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5365079365079368e-05,
+      "loss": 0.484,
+      "step": 1656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.536190476190476e-05,
+      "loss": 0.4944,
+      "step": 1657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.535873015873016e-05,
+      "loss": 0.5392,
+      "step": 1658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5355555555555557e-05,
+      "loss": 0.4708,
+      "step": 1659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5352380952380953e-05,
+      "loss": 0.4907,
+      "step": 1660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5349206349206353e-05,
+      "loss": 0.4726,
+      "step": 1661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5346031746031746e-05,
+      "loss": 0.4805,
+      "step": 1662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5342857142857142e-05,
+      "loss": 0.4845,
+      "step": 1663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5339682539682542e-05,
+      "loss": 0.5234,
+      "step": 1664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5336507936507938e-05,
+      "loss": 0.4749,
+      "step": 1665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 0.4822,
+      "step": 1666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.533015873015873e-05,
+      "loss": 0.4565,
+      "step": 1667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5326984126984127e-05,
+      "loss": 0.4612,
+      "step": 1668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5323809523809527e-05,
+      "loss": 0.5031,
+      "step": 1669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5320634920634923e-05,
+      "loss": 0.4839,
+      "step": 1670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.531746031746032e-05,
+      "loss": 0.4862,
+      "step": 1671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5314285714285716e-05,
+      "loss": 0.477,
+      "step": 1672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5311111111111112e-05,
+      "loss": 0.4923,
+      "step": 1673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.530793650793651e-05,
+      "loss": 0.485,
+      "step": 1674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5304761904761908e-05,
+      "loss": 0.5204,
+      "step": 1675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.53015873015873e-05,
+      "loss": 0.5092,
+      "step": 1676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5298412698412697e-05,
+      "loss": 0.5085,
+      "step": 1677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5295238095238097e-05,
+      "loss": 0.4592,
+      "step": 1678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5292063492063493e-05,
+      "loss": 0.5146,
+      "step": 1679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.528888888888889e-05,
+      "loss": 0.4579,
+      "step": 1680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5285714285714286e-05,
+      "loss": 0.5183,
+      "step": 1681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5282539682539682e-05,
+      "loss": 0.4959,
+      "step": 1682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5279365079365082e-05,
+      "loss": 0.4972,
+      "step": 1683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.527619047619048e-05,
+      "loss": 0.5202,
+      "step": 1684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5273015873015875e-05,
+      "loss": 0.5116,
+      "step": 1685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.526984126984127e-05,
+      "loss": 0.4679,
+      "step": 1686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5266666666666667e-05,
+      "loss": 0.5116,
+      "step": 1687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5263492063492064e-05,
+      "loss": 0.4872,
+      "step": 1688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5260317460317463e-05,
+      "loss": 0.5095,
+      "step": 1689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.525714285714286e-05,
+      "loss": 0.5027,
+      "step": 1690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5253968253968254e-05,
+      "loss": 0.4716,
+      "step": 1691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5250793650793652e-05,
+      "loss": 0.4869,
+      "step": 1692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5247619047619049e-05,
+      "loss": 0.5258,
+      "step": 1693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5244444444444447e-05,
+      "loss": 0.4891,
+      "step": 1694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5241269841269843e-05,
+      "loss": 0.4948,
+      "step": 1695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 0.5168,
+      "step": 1696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5234920634920636e-05,
+      "loss": 0.49,
+      "step": 1697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5231746031746034e-05,
+      "loss": 0.5047,
+      "step": 1698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.522857142857143e-05,
+      "loss": 0.5041,
+      "step": 1699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5225396825396828e-05,
+      "loss": 0.4961,
+      "step": 1700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5222222222222223e-05,
+      "loss": 0.4616,
+      "step": 1701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.521904761904762e-05,
+      "loss": 0.5036,
+      "step": 1702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5215873015873017e-05,
+      "loss": 0.5132,
+      "step": 1703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5212698412698415e-05,
+      "loss": 0.4813,
+      "step": 1704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5209523809523811e-05,
+      "loss": 0.4616,
+      "step": 1705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5206349206349208e-05,
+      "loss": 0.5271,
+      "step": 1706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5203174603174604e-05,
+      "loss": 0.5057,
+      "step": 1707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.4697,
+      "step": 1708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5196825396825398e-05,
+      "loss": 0.4771,
+      "step": 1709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5193650793650796e-05,
+      "loss": 0.4542,
+      "step": 1710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.519047619047619e-05,
+      "loss": 0.4406,
+      "step": 1711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5187301587301589e-05,
+      "loss": 0.4898,
+      "step": 1712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5184126984126985e-05,
+      "loss": 0.5033,
+      "step": 1713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5180952380952383e-05,
+      "loss": 0.5245,
+      "step": 1714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5177777777777778e-05,
+      "loss": 0.4886,
+      "step": 1715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5174603174603176e-05,
+      "loss": 0.4934,
+      "step": 1716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5171428571428572e-05,
+      "loss": 0.4779,
+      "step": 1717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.516825396825397e-05,
+      "loss": 0.4716,
+      "step": 1718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5165079365079366e-05,
+      "loss": 0.4913,
+      "step": 1719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5161904761904763e-05,
+      "loss": 0.5126,
+      "step": 1720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5158730158730159e-05,
+      "loss": 0.477,
+      "step": 1721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5155555555555557e-05,
+      "loss": 0.4894,
+      "step": 1722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5152380952380953e-05,
+      "loss": 0.5165,
+      "step": 1723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5149206349206351e-05,
+      "loss": 0.4734,
+      "step": 1724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5146031746031746e-05,
+      "loss": 0.4682,
+      "step": 1725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5142857142857144e-05,
+      "loss": 0.5217,
+      "step": 1726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.513968253968254e-05,
+      "loss": 0.4816,
+      "step": 1727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5136507936507938e-05,
+      "loss": 0.4849,
+      "step": 1728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5133333333333335e-05,
+      "loss": 0.5064,
+      "step": 1729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5130158730158731e-05,
+      "loss": 0.5068,
+      "step": 1730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5126984126984127e-05,
+      "loss": 0.4641,
+      "step": 1731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5123809523809525e-05,
+      "loss": 0.4788,
+      "step": 1732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5120634920634922e-05,
+      "loss": 0.5056,
+      "step": 1733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.511746031746032e-05,
+      "loss": 0.4574,
+      "step": 1734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5114285714285714e-05,
+      "loss": 0.5244,
+      "step": 1735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.4732,
+      "step": 1736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5107936507936509e-05,
+      "loss": 0.4652,
+      "step": 1737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5104761904761907e-05,
+      "loss": 0.4634,
+      "step": 1738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5101587301587303e-05,
+      "loss": 0.4431,
+      "step": 1739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.50984126984127e-05,
+      "loss": 0.5137,
+      "step": 1740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5095238095238096e-05,
+      "loss": 0.4626,
+      "step": 1741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5092063492063494e-05,
+      "loss": 0.4803,
+      "step": 1742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.508888888888889e-05,
+      "loss": 0.4638,
+      "step": 1743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5085714285714288e-05,
+      "loss": 0.4588,
+      "step": 1744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5082539682539683e-05,
+      "loss": 0.5608,
+      "step": 1745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.507936507936508e-05,
+      "loss": 0.4635,
+      "step": 1746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5076190476190477e-05,
+      "loss": 0.4848,
+      "step": 1747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5073015873015875e-05,
+      "loss": 0.4964,
+      "step": 1748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5069841269841273e-05,
+      "loss": 0.4934,
+      "step": 1749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5066666666666668e-05,
+      "loss": 0.4876,
+      "step": 1750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5063492063492064e-05,
+      "loss": 0.5079,
+      "step": 1751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5060317460317462e-05,
+      "loss": 0.4825,
+      "step": 1752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.505714285714286e-05,
+      "loss": 0.4817,
+      "step": 1753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5053968253968255e-05,
+      "loss": 0.4777,
+      "step": 1754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.505079365079365e-05,
+      "loss": 0.4779,
+      "step": 1755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5047619047619049e-05,
+      "loss": 0.5245,
+      "step": 1756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5044444444444445e-05,
+      "loss": 0.4976,
+      "step": 1757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5041269841269843e-05,
+      "loss": 0.4842,
+      "step": 1758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5038095238095238e-05,
+      "loss": 0.4947,
+      "step": 1759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5034920634920636e-05,
+      "loss": 0.4977,
+      "step": 1760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5031746031746032e-05,
+      "loss": 0.4871,
+      "step": 1761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.502857142857143e-05,
+      "loss": 0.4849,
+      "step": 1762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5025396825396828e-05,
+      "loss": 0.4613,
+      "step": 1763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5022222222222223e-05,
+      "loss": 0.4615,
+      "step": 1764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5019047619047619e-05,
+      "loss": 0.5166,
+      "step": 1765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5015873015873017e-05,
+      "loss": 0.5253,
+      "step": 1766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5012698412698415e-05,
+      "loss": 0.4776,
+      "step": 1767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5009523809523811e-05,
+      "loss": 0.501,
+      "step": 1768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5006349206349206e-05,
+      "loss": 0.4889,
+      "step": 1769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5003174603174604e-05,
+      "loss": 0.4735,
+      "step": 1770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4479,
+      "step": 1771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4996825396825398e-05,
+      "loss": 0.4717,
+      "step": 1772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4993650793650796e-05,
+      "loss": 0.4764,
+      "step": 1773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4990476190476191e-05,
+      "loss": 0.4571,
+      "step": 1774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4987301587301587e-05,
+      "loss": 0.4654,
+      "step": 1775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4984126984126985e-05,
+      "loss": 0.4802,
+      "step": 1776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4980952380952383e-05,
+      "loss": 0.4855,
+      "step": 1777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.497777777777778e-05,
+      "loss": 0.5009,
+      "step": 1778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4974603174603174e-05,
+      "loss": 0.4985,
+      "step": 1779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4971428571428572e-05,
+      "loss": 0.495,
+      "step": 1780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.496825396825397e-05,
+      "loss": 0.4838,
+      "step": 1781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4965079365079367e-05,
+      "loss": 0.4605,
+      "step": 1782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4961904761904765e-05,
+      "loss": 0.4799,
+      "step": 1783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.495873015873016e-05,
+      "loss": 0.4655,
+      "step": 1784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4955555555555557e-05,
+      "loss": 0.5449,
+      "step": 1785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4952380952380954e-05,
+      "loss": 0.4523,
+      "step": 1786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4949206349206352e-05,
+      "loss": 0.5148,
+      "step": 1787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4946031746031748e-05,
+      "loss": 0.5046,
+      "step": 1788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4942857142857143e-05,
+      "loss": 0.494,
+      "step": 1789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493968253968254e-05,
+      "loss": 0.5011,
+      "step": 1790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4936507936507939e-05,
+      "loss": 0.4741,
+      "step": 1791
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 0.4997,
+      "step": 1792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493015873015873e-05,
+      "loss": 0.5101,
+      "step": 1793
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4926984126984128e-05,
+      "loss": 0.5108,
+      "step": 1794
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4923809523809526e-05,
+      "loss": 0.4847,
+      "step": 1795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4920634920634922e-05,
+      "loss": 0.4748,
+      "step": 1796
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491746031746032e-05,
+      "loss": 0.4931,
+      "step": 1797
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4914285714285715e-05,
+      "loss": 0.459,
+      "step": 1798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4911111111111113e-05,
+      "loss": 0.4594,
+      "step": 1799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4907936507936509e-05,
+      "loss": 0.492,
+      "step": 1800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4904761904761907e-05,
+      "loss": 0.4991,
+      "step": 1801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4901587301587303e-05,
+      "loss": 0.4601,
+      "step": 1802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.48984126984127e-05,
+      "loss": 0.4886,
+      "step": 1803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4895238095238096e-05,
+      "loss": 0.4748,
+      "step": 1804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4892063492063494e-05,
+      "loss": 0.4881,
+      "step": 1805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.488888888888889e-05,
+      "loss": 0.4823,
+      "step": 1806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4885714285714288e-05,
+      "loss": 0.5108,
+      "step": 1807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4882539682539683e-05,
+      "loss": 0.4546,
+      "step": 1808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.487936507936508e-05,
+      "loss": 0.4928,
+      "step": 1809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4876190476190477e-05,
+      "loss": 0.4831,
+      "step": 1810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4873015873015875e-05,
+      "loss": 0.4796,
+      "step": 1811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4869841269841271e-05,
+      "loss": 0.4743,
+      "step": 1812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4866666666666668e-05,
+      "loss": 0.4912,
+      "step": 1813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4863492063492064e-05,
+      "loss": 0.5013,
+      "step": 1814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4860317460317462e-05,
+      "loss": 0.4898,
+      "step": 1815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 0.4556,
+      "step": 1816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4853968253968256e-05,
+      "loss": 0.4893,
+      "step": 1817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4850793650793651e-05,
+      "loss": 0.475,
+      "step": 1818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4847619047619049e-05,
+      "loss": 0.4851,
+      "step": 1819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4844444444444445e-05,
+      "loss": 0.475,
+      "step": 1820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4841269841269843e-05,
+      "loss": 0.4856,
+      "step": 1821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.483809523809524e-05,
+      "loss": 0.5046,
+      "step": 1822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4834920634920636e-05,
+      "loss": 0.4959,
+      "step": 1823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4831746031746032e-05,
+      "loss": 0.452,
+      "step": 1824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.482857142857143e-05,
+      "loss": 0.4856,
+      "step": 1825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4825396825396827e-05,
+      "loss": 0.4882,
+      "step": 1826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4822222222222225e-05,
+      "loss": 0.5117,
+      "step": 1827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.481904761904762e-05,
+      "loss": 0.4364,
+      "step": 1828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4815873015873017e-05,
+      "loss": 0.4974,
+      "step": 1829
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4812698412698414e-05,
+      "loss": 0.4751,
+      "step": 1830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4809523809523812e-05,
+      "loss": 0.4913,
+      "step": 1831
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4806349206349206e-05,
+      "loss": 0.4936,
+      "step": 1832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4803174603174604e-05,
+      "loss": 0.4774,
+      "step": 1833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.48e-05,
+      "loss": 0.4815,
+      "step": 1834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4796825396825399e-05,
+      "loss": 0.4929,
+      "step": 1835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4793650793650795e-05,
+      "loss": 0.5065,
+      "step": 1836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4790476190476191e-05,
+      "loss": 0.472,
+      "step": 1837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4787301587301588e-05,
+      "loss": 0.4968,
+      "step": 1838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4784126984126986e-05,
+      "loss": 0.4827,
+      "step": 1839
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4780952380952382e-05,
+      "loss": 0.5083,
+      "step": 1840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.477777777777778e-05,
+      "loss": 0.4261,
+      "step": 1841
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4774603174603175e-05,
+      "loss": 0.4883,
+      "step": 1842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4771428571428573e-05,
+      "loss": 0.5103,
+      "step": 1843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4768253968253969e-05,
+      "loss": 0.5041,
+      "step": 1844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4765079365079367e-05,
+      "loss": 0.5532,
+      "step": 1845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4761904761904763e-05,
+      "loss": 0.4859,
+      "step": 1846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.475873015873016e-05,
+      "loss": 0.4874,
+      "step": 1847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4755555555555556e-05,
+      "loss": 0.4902,
+      "step": 1848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4752380952380954e-05,
+      "loss": 0.4563,
+      "step": 1849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.474920634920635e-05,
+      "loss": 0.4873,
+      "step": 1850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4746031746031748e-05,
+      "loss": 0.4873,
+      "step": 1851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4742857142857143e-05,
+      "loss": 0.4637,
+      "step": 1852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.473968253968254e-05,
+      "loss": 0.4782,
+      "step": 1853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4736507936507937e-05,
+      "loss": 0.5019,
+      "step": 1854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4733333333333335e-05,
+      "loss": 0.4661,
+      "step": 1855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4730158730158733e-05,
+      "loss": 0.4986,
+      "step": 1856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4726984126984128e-05,
+      "loss": 0.4941,
+      "step": 1857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4723809523809524e-05,
+      "loss": 0.4662,
+      "step": 1858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4720634920634922e-05,
+      "loss": 0.4725,
+      "step": 1859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4717460317460318e-05,
+      "loss": 0.4399,
+      "step": 1860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4714285714285716e-05,
+      "loss": 0.4817,
+      "step": 1861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4711111111111111e-05,
+      "loss": 0.4865,
+      "step": 1862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4707936507936509e-05,
+      "loss": 0.5197,
+      "step": 1863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4704761904761905e-05,
+      "loss": 0.5027,
+      "step": 1864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4701587301587303e-05,
+      "loss": 0.5127,
+      "step": 1865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4698412698412701e-05,
+      "loss": 0.5166,
+      "step": 1866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4695238095238096e-05,
+      "loss": 0.4767,
+      "step": 1867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4692063492063492e-05,
+      "loss": 0.4956,
+      "step": 1868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.468888888888889e-05,
+      "loss": 0.5075,
+      "step": 1869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4685714285714288e-05,
+      "loss": 0.4788,
+      "step": 1870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4682539682539683e-05,
+      "loss": 0.5008,
+      "step": 1871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.467936507936508e-05,
+      "loss": 0.4664,
+      "step": 1872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4676190476190477e-05,
+      "loss": 0.4967,
+      "step": 1873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4673015873015874e-05,
+      "loss": 0.4935,
+      "step": 1874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4669841269841272e-05,
+      "loss": 0.4879,
+      "step": 1875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.4383,
+      "step": 1876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4663492063492064e-05,
+      "loss": 0.5146,
+      "step": 1877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.466031746031746e-05,
+      "loss": 0.4705,
+      "step": 1878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4657142857142859e-05,
+      "loss": 0.5112,
+      "step": 1879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4653968253968257e-05,
+      "loss": 0.4604,
+      "step": 1880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4650793650793651e-05,
+      "loss": 0.5184,
+      "step": 1881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4647619047619048e-05,
+      "loss": 0.5022,
+      "step": 1882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4644444444444446e-05,
+      "loss": 0.4888,
+      "step": 1883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4641269841269844e-05,
+      "loss": 0.4832,
+      "step": 1884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.463809523809524e-05,
+      "loss": 0.4558,
+      "step": 1885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4634920634920635e-05,
+      "loss": 0.4959,
+      "step": 1886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4631746031746033e-05,
+      "loss": 0.5133,
+      "step": 1887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.462857142857143e-05,
+      "loss": 0.4549,
+      "step": 1888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4625396825396827e-05,
+      "loss": 0.4701,
+      "step": 1889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4622222222222225e-05,
+      "loss": 0.481,
+      "step": 1890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.461904761904762e-05,
+      "loss": 0.4963,
+      "step": 1891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4615873015873016e-05,
+      "loss": 0.4604,
+      "step": 1892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4612698412698414e-05,
+      "loss": 0.4932,
+      "step": 1893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4609523809523812e-05,
+      "loss": 0.4925,
+      "step": 1894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4606349206349208e-05,
+      "loss": 0.4822,
+      "step": 1895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4603174603174603e-05,
+      "loss": 0.5043,
+      "step": 1896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.46e-05,
+      "loss": 0.4794,
+      "step": 1897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4596825396825399e-05,
+      "loss": 0.4833,
+      "step": 1898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4593650793650795e-05,
+      "loss": 0.4689,
+      "step": 1899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4590476190476193e-05,
+      "loss": 0.5212,
+      "step": 1900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4587301587301588e-05,
+      "loss": 0.5054,
+      "step": 1901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4584126984126986e-05,
+      "loss": 0.4959,
+      "step": 1902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4580952380952382e-05,
+      "loss": 0.4952,
+      "step": 1903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.457777777777778e-05,
+      "loss": 0.5288,
+      "step": 1904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4574603174603176e-05,
+      "loss": 0.4926,
+      "step": 1905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4571428571428573e-05,
+      "loss": 0.5121,
+      "step": 1906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4568253968253969e-05,
+      "loss": 0.4376,
+      "step": 1907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4565079365079367e-05,
+      "loss": 0.4564,
+      "step": 1908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4561904761904763e-05,
+      "loss": 0.4929,
+      "step": 1909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4558730158730158e-05,
+      "loss": 0.5112,
+      "step": 1910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4555555555555556e-05,
+      "loss": 0.4472,
+      "step": 1911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4552380952380954e-05,
+      "loss": 0.4334,
+      "step": 1912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.454920634920635e-05,
+      "loss": 0.4733,
+      "step": 1913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4546031746031748e-05,
+      "loss": 0.5139,
+      "step": 1914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4542857142857143e-05,
+      "loss": 0.4962,
+      "step": 1915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4539682539682541e-05,
+      "loss": 0.457,
+      "step": 1916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4536507936507937e-05,
+      "loss": 0.4721,
+      "step": 1917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4533333333333335e-05,
+      "loss": 0.4768,
+      "step": 1918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4530158730158732e-05,
+      "loss": 0.4878,
+      "step": 1919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4526984126984128e-05,
+      "loss": 0.4831,
+      "step": 1920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4523809523809524e-05,
+      "loss": 0.4802,
+      "step": 1921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4520634920634922e-05,
+      "loss": 0.4822,
+      "step": 1922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4517460317460319e-05,
+      "loss": 0.4953,
+      "step": 1923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4514285714285717e-05,
+      "loss": 0.5064,
+      "step": 1924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4511111111111111e-05,
+      "loss": 0.4499,
+      "step": 1925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.450793650793651e-05,
+      "loss": 0.4955,
+      "step": 1926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4504761904761906e-05,
+      "loss": 0.5189,
+      "step": 1927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4501587301587304e-05,
+      "loss": 0.4554,
+      "step": 1928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.44984126984127e-05,
+      "loss": 0.5115,
+      "step": 1929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4495238095238096e-05,
+      "loss": 0.4935,
+      "step": 1930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4492063492063493e-05,
+      "loss": 0.5259,
+      "step": 1931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.448888888888889e-05,
+      "loss": 0.4847,
+      "step": 1932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4485714285714287e-05,
+      "loss": 0.4756,
+      "step": 1933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4482539682539685e-05,
+      "loss": 0.46,
+      "step": 1934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.447936507936508e-05,
+      "loss": 0.4998,
+      "step": 1935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4476190476190478e-05,
+      "loss": 0.4558,
+      "step": 1936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4473015873015874e-05,
+      "loss": 0.4567,
+      "step": 1937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4469841269841272e-05,
+      "loss": 0.4638,
+      "step": 1938
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4466666666666668e-05,
+      "loss": 0.4926,
+      "step": 1939
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4463492063492065e-05,
+      "loss": 0.5116,
+      "step": 1940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.446031746031746e-05,
+      "loss": 0.5132,
+      "step": 1941
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4457142857142859e-05,
+      "loss": 0.4759,
+      "step": 1942
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4453968253968255e-05,
+      "loss": 0.5112,
+      "step": 1943
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4450793650793653e-05,
+      "loss": 0.4656,
+      "step": 1944
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4447619047619048e-05,
+      "loss": 0.4884,
+      "step": 1945
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 0.473,
+      "step": 1946
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4441269841269842e-05,
+      "loss": 0.5068,
+      "step": 1947
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.443809523809524e-05,
+      "loss": 0.5228,
+      "step": 1948
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4434920634920635e-05,
+      "loss": 0.4722,
+      "step": 1949
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4431746031746033e-05,
+      "loss": 0.4631,
+      "step": 1950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4428571428571429e-05,
+      "loss": 0.4925,
+      "step": 1951
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4425396825396827e-05,
+      "loss": 0.4846,
+      "step": 1952
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4422222222222223e-05,
+      "loss": 0.4574,
+      "step": 1953
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.441904761904762e-05,
+      "loss": 0.5145,
+      "step": 1954
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4415873015873016e-05,
+      "loss": 0.5056,
+      "step": 1955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4412698412698414e-05,
+      "loss": 0.4898,
+      "step": 1956
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.440952380952381e-05,
+      "loss": 0.5125,
+      "step": 1957
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4406349206349208e-05,
+      "loss": 0.4742,
+      "step": 1958
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4403174603174603e-05,
+      "loss": 0.4689,
+      "step": 1959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.4938,
+      "step": 1960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4396825396825397e-05,
+      "loss": 0.4804,
+      "step": 1961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4393650793650795e-05,
+      "loss": 0.4616,
+      "step": 1962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4390476190476192e-05,
+      "loss": 0.4982,
+      "step": 1963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4387301587301588e-05,
+      "loss": 0.4803,
+      "step": 1964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4384126984126984e-05,
+      "loss": 0.4983,
+      "step": 1965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4380952380952382e-05,
+      "loss": 0.4942,
+      "step": 1966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4377777777777779e-05,
+      "loss": 0.4451,
+      "step": 1967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4374603174603177e-05,
+      "loss": 0.4737,
+      "step": 1968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4371428571428571e-05,
+      "loss": 0.516,
+      "step": 1969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.436825396825397e-05,
+      "loss": 0.4802,
+      "step": 1970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4365079365079366e-05,
+      "loss": 0.4913,
+      "step": 1971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4361904761904764e-05,
+      "loss": 0.5256,
+      "step": 1972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4358730158730162e-05,
+      "loss": 0.4964,
+      "step": 1973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4355555555555556e-05,
+      "loss": 0.4791,
+      "step": 1974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4352380952380953e-05,
+      "loss": 0.4641,
+      "step": 1975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.434920634920635e-05,
+      "loss": 0.495,
+      "step": 1976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4346031746031747e-05,
+      "loss": 0.4881,
+      "step": 1977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4342857142857145e-05,
+      "loss": 0.4906,
+      "step": 1978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.433968253968254e-05,
+      "loss": 0.5116,
+      "step": 1979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4336507936507938e-05,
+      "loss": 0.4858,
+      "step": 1980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4333333333333334e-05,
+      "loss": 0.481,
+      "step": 1981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4330158730158732e-05,
+      "loss": 0.4716,
+      "step": 1982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.432698412698413e-05,
+      "loss": 0.492,
+      "step": 1983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4323809523809525e-05,
+      "loss": 0.4777,
+      "step": 1984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4320634920634921e-05,
+      "loss": 0.4763,
+      "step": 1985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4317460317460319e-05,
+      "loss": 0.4797,
+      "step": 1986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4314285714285717e-05,
+      "loss": 0.4811,
+      "step": 1987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4311111111111111e-05,
+      "loss": 0.5066,
+      "step": 1988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4307936507936508e-05,
+      "loss": 0.5339,
+      "step": 1989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4304761904761906e-05,
+      "loss": 0.4973,
+      "step": 1990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4301587301587304e-05,
+      "loss": 0.4865,
+      "step": 1991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.42984126984127e-05,
+      "loss": 0.4788,
+      "step": 1992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4295238095238095e-05,
+      "loss": 0.4665,
+      "step": 1993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4292063492063493e-05,
+      "loss": 0.4485,
+      "step": 1994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4288888888888889e-05,
+      "loss": 0.4784,
+      "step": 1995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.4746,
+      "step": 1996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4282539682539685e-05,
+      "loss": 0.4718,
+      "step": 1997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.427936507936508e-05,
+      "loss": 0.484,
+      "step": 1998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4276190476190476e-05,
+      "loss": 0.4661,
+      "step": 1999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4273015873015874e-05,
+      "loss": 0.5006,
+      "step": 2000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4269841269841272e-05,
+      "loss": 0.4546,
+      "step": 2001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4266666666666668e-05,
+      "loss": 0.4801,
+      "step": 2002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4263492063492063e-05,
+      "loss": 0.4989,
+      "step": 2003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4260317460317461e-05,
+      "loss": 0.4892,
+      "step": 2004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4257142857142859e-05,
+      "loss": 0.4934,
+      "step": 2005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4253968253968255e-05,
+      "loss": 0.4738,
+      "step": 2006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4250793650793653e-05,
+      "loss": 0.4514,
+      "step": 2007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4247619047619048e-05,
+      "loss": 0.4788,
+      "step": 2008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4244444444444444e-05,
+      "loss": 0.471,
+      "step": 2009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4241269841269842e-05,
+      "loss": 0.4763,
+      "step": 2010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.423809523809524e-05,
+      "loss": 0.5151,
+      "step": 2011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4234920634920637e-05,
+      "loss": 0.4818,
+      "step": 2012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4231746031746031e-05,
+      "loss": 0.4946,
+      "step": 2013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.422857142857143e-05,
+      "loss": 0.4438,
+      "step": 2014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4225396825396827e-05,
+      "loss": 0.4808,
+      "step": 2015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.5238,
+      "step": 2016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4219047619047622e-05,
+      "loss": 0.4866,
+      "step": 2017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4215873015873016e-05,
+      "loss": 0.5256,
+      "step": 2018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4212698412698414e-05,
+      "loss": 0.5366,
+      "step": 2019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.420952380952381e-05,
+      "loss": 0.4842,
+      "step": 2020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4206349206349209e-05,
+      "loss": 0.5002,
+      "step": 2021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4203174603174605e-05,
+      "loss": 0.5057,
+      "step": 2022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 0.5383,
+      "step": 2023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4196825396825398e-05,
+      "loss": 0.4893,
+      "step": 2024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4193650793650796e-05,
+      "loss": 0.451,
+      "step": 2025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4190476190476192e-05,
+      "loss": 0.4908,
+      "step": 2026
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4187301587301587e-05,
+      "loss": 0.4844,
+      "step": 2027
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4184126984126985e-05,
+      "loss": 0.5266,
+      "step": 2028
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4180952380952383e-05,
+      "loss": 0.4601,
+      "step": 2029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4177777777777779e-05,
+      "loss": 0.5089,
+      "step": 2030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4174603174603177e-05,
+      "loss": 0.5207,
+      "step": 2031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4171428571428572e-05,
+      "loss": 0.481,
+      "step": 2032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.416825396825397e-05,
+      "loss": 0.4444,
+      "step": 2033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4165079365079366e-05,
+      "loss": 0.4646,
+      "step": 2034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4161904761904764e-05,
+      "loss": 0.5075,
+      "step": 2035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.415873015873016e-05,
+      "loss": 0.4486,
+      "step": 2036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4155555555555556e-05,
+      "loss": 0.4946,
+      "step": 2037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4152380952380953e-05,
+      "loss": 0.4682,
+      "step": 2038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.414920634920635e-05,
+      "loss": 0.4915,
+      "step": 2039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4146031746031747e-05,
+      "loss": 0.4727,
+      "step": 2040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4142857142857145e-05,
+      "loss": 0.498,
+      "step": 2041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.413968253968254e-05,
+      "loss": 0.5039,
+      "step": 2042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4136507936507938e-05,
+      "loss": 0.4828,
+      "step": 2043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 0.5049,
+      "step": 2044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4130158730158732e-05,
+      "loss": 0.51,
+      "step": 2045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4126984126984128e-05,
+      "loss": 0.4899,
+      "step": 2046
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4123809523809525e-05,
+      "loss": 0.5051,
+      "step": 2047
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4120634920634921e-05,
+      "loss": 0.4962,
+      "step": 2048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4117460317460319e-05,
+      "loss": 0.5182,
+      "step": 2049
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4114285714285715e-05,
+      "loss": 0.4875,
+      "step": 2050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4111111111111113e-05,
+      "loss": 0.4678,
+      "step": 2051
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4107936507936508e-05,
+      "loss": 0.4927,
+      "step": 2052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4104761904761906e-05,
+      "loss": 0.4739,
+      "step": 2053
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4101587301587302e-05,
+      "loss": 0.5156,
+      "step": 2054
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.40984126984127e-05,
+      "loss": 0.4499,
+      "step": 2055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4095238095238097e-05,
+      "loss": 0.469,
+      "step": 2056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4092063492063493e-05,
+      "loss": 0.4864,
+      "step": 2057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.408888888888889e-05,
+      "loss": 0.4836,
+      "step": 2058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4085714285714287e-05,
+      "loss": 0.4664,
+      "step": 2059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4082539682539684e-05,
+      "loss": 0.4623,
+      "step": 2060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4079365079365082e-05,
+      "loss": 0.4484,
+      "step": 2061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4076190476190476e-05,
+      "loss": 0.4749,
+      "step": 2062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4073015873015874e-05,
+      "loss": 0.4902,
+      "step": 2063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.406984126984127e-05,
+      "loss": 0.482,
+      "step": 2064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4066666666666669e-05,
+      "loss": 0.5035,
+      "step": 2065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4063492063492065e-05,
+      "loss": 0.4497,
+      "step": 2066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4060317460317461e-05,
+      "loss": 0.4985,
+      "step": 2067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4057142857142858e-05,
+      "loss": 0.4793,
+      "step": 2068
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4053968253968256e-05,
+      "loss": 0.4935,
+      "step": 2069
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4050793650793652e-05,
+      "loss": 0.5306,
+      "step": 2070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4047619047619048e-05,
+      "loss": 0.468,
+      "step": 2071
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4044444444444445e-05,
+      "loss": 0.4611,
+      "step": 2072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4041269841269843e-05,
+      "loss": 0.5054,
+      "step": 2073
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4038095238095239e-05,
+      "loss": 0.4788,
+      "step": 2074
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4034920634920637e-05,
+      "loss": 0.4623,
+      "step": 2075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4031746031746032e-05,
+      "loss": 0.4822,
+      "step": 2076
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.402857142857143e-05,
+      "loss": 0.4964,
+      "step": 2077
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4025396825396826e-05,
+      "loss": 0.4896,
+      "step": 2078
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4022222222222224e-05,
+      "loss": 0.5002,
+      "step": 2079
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.401904761904762e-05,
+      "loss": 0.4849,
+      "step": 2080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4015873015873017e-05,
+      "loss": 0.45,
+      "step": 2081
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4012698412698413e-05,
+      "loss": 0.4611,
+      "step": 2082
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.400952380952381e-05,
+      "loss": 0.4996,
+      "step": 2083
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4006349206349207e-05,
+      "loss": 0.4928,
+      "step": 2084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4003174603174605e-05,
+      "loss": 0.4755,
+      "step": 2085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4e-05,
+      "loss": 0.4824,
+      "step": 2086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3996825396825398e-05,
+      "loss": 0.4738,
+      "step": 2087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3993650793650794e-05,
+      "loss": 0.5001,
+      "step": 2088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3990476190476192e-05,
+      "loss": 0.4984,
+      "step": 2089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.398730158730159e-05,
+      "loss": 0.4901,
+      "step": 2090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3984126984126985e-05,
+      "loss": 0.4607,
+      "step": 2091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3980952380952381e-05,
+      "loss": 0.4731,
+      "step": 2092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3977777777777779e-05,
+      "loss": 0.4691,
+      "step": 2093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3974603174603177e-05,
+      "loss": 0.5009,
+      "step": 2094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3971428571428573e-05,
+      "loss": 0.4884,
+      "step": 2095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3968253968253968e-05,
+      "loss": 0.4931,
+      "step": 2096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3965079365079366e-05,
+      "loss": 0.4818,
+      "step": 2097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3961904761904762e-05,
+      "loss": 0.4928,
+      "step": 2098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.395873015873016e-05,
+      "loss": 0.493,
+      "step": 2099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3955555555555558e-05,
+      "loss": 0.4481,
+      "step": 2100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3952380952380953e-05,
+      "loss": 0.5141,
+      "step": 2101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.394920634920635e-05,
+      "loss": 0.4875,
+      "step": 2102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3946031746031747e-05,
+      "loss": 0.4742,
+      "step": 2103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3942857142857145e-05,
+      "loss": 0.5025,
+      "step": 2104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3939682539682542e-05,
+      "loss": 0.4555,
+      "step": 2105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3936507936507936e-05,
+      "loss": 0.4977,
+      "step": 2106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3933333333333334e-05,
+      "loss": 0.5204,
+      "step": 2107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3930158730158732e-05,
+      "loss": 0.4921,
+      "step": 2108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3926984126984129e-05,
+      "loss": 0.5133,
+      "step": 2109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3923809523809523e-05,
+      "loss": 0.4707,
+      "step": 2110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3920634920634921e-05,
+      "loss": 0.469,
+      "step": 2111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3917460317460318e-05,
+      "loss": 0.4913,
+      "step": 2112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3914285714285716e-05,
+      "loss": 0.4989,
+      "step": 2113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3911111111111114e-05,
+      "loss": 0.4913,
+      "step": 2114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3907936507936508e-05,
+      "loss": 0.5028,
+      "step": 2115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3904761904761905e-05,
+      "loss": 0.485,
+      "step": 2116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3901587301587303e-05,
+      "loss": 0.5153,
+      "step": 2117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.38984126984127e-05,
+      "loss": 0.4559,
+      "step": 2118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3895238095238097e-05,
+      "loss": 0.4852,
+      "step": 2119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3892063492063492e-05,
+      "loss": 0.4697,
+      "step": 2120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.486,
+      "step": 2121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3885714285714288e-05,
+      "loss": 0.4805,
+      "step": 2122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3882539682539684e-05,
+      "loss": 0.4714,
+      "step": 2123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3879365079365082e-05,
+      "loss": 0.4689,
+      "step": 2124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3876190476190477e-05,
+      "loss": 0.4591,
+      "step": 2125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3873015873015875e-05,
+      "loss": 0.481,
+      "step": 2126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3869841269841271e-05,
+      "loss": 0.4744,
+      "step": 2127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 0.4895,
+      "step": 2128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3863492063492065e-05,
+      "loss": 0.468,
+      "step": 2129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.386031746031746e-05,
+      "loss": 0.5169,
+      "step": 2130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3857142857142858e-05,
+      "loss": 0.4913,
+      "step": 2131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3853968253968256e-05,
+      "loss": 0.4717,
+      "step": 2132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3850793650793652e-05,
+      "loss": 0.506,
+      "step": 2133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.384761904761905e-05,
+      "loss": 0.4734,
+      "step": 2134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3844444444444445e-05,
+      "loss": 0.4983,
+      "step": 2135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3841269841269843e-05,
+      "loss": 0.5105,
+      "step": 2136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3838095238095239e-05,
+      "loss": 0.5077,
+      "step": 2137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3834920634920637e-05,
+      "loss": 0.5224,
+      "step": 2138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3831746031746033e-05,
+      "loss": 0.4498,
+      "step": 2139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.382857142857143e-05,
+      "loss": 0.4211,
+      "step": 2140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3825396825396826e-05,
+      "loss": 0.4884,
+      "step": 2141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3822222222222224e-05,
+      "loss": 0.4588,
+      "step": 2142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.381904761904762e-05,
+      "loss": 0.5025,
+      "step": 2143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3815873015873018e-05,
+      "loss": 0.5199,
+      "step": 2144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3812698412698413e-05,
+      "loss": 0.5064,
+      "step": 2145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3809523809523811e-05,
+      "loss": 0.4747,
+      "step": 2146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3806349206349207e-05,
+      "loss": 0.4361,
+      "step": 2147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3803174603174605e-05,
+      "loss": 0.4823,
+      "step": 2148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.38e-05,
+      "loss": 0.4914,
+      "step": 2149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3796825396825398e-05,
+      "loss": 0.5092,
+      "step": 2150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3793650793650794e-05,
+      "loss": 0.5396,
+      "step": 2151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3790476190476192e-05,
+      "loss": 0.4632,
+      "step": 2152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3787301587301589e-05,
+      "loss": 0.4738,
+      "step": 2153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3784126984126985e-05,
+      "loss": 0.4971,
+      "step": 2154
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3780952380952381e-05,
+      "loss": 0.4542,
+      "step": 2155
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.377777777777778e-05,
+      "loss": 0.4777,
+      "step": 2156
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3774603174603176e-05,
+      "loss": 0.5518,
+      "step": 2157
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3771428571428574e-05,
+      "loss": 0.4408,
+      "step": 2158
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3768253968253968e-05,
+      "loss": 0.5121,
+      "step": 2159
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3765079365079366e-05,
+      "loss": 0.5019,
+      "step": 2160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3761904761904763e-05,
+      "loss": 0.5378,
+      "step": 2161
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.375873015873016e-05,
+      "loss": 0.4877,
+      "step": 2162
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3755555555555557e-05,
+      "loss": 0.4647,
+      "step": 2163
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3752380952380953e-05,
+      "loss": 0.4737,
+      "step": 2164
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.374920634920635e-05,
+      "loss": 0.4434,
+      "step": 2165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3746031746031748e-05,
+      "loss": 0.4392,
+      "step": 2166
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3742857142857144e-05,
+      "loss": 0.3816,
+      "step": 2167
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3739682539682542e-05,
+      "loss": 0.3714,
+      "step": 2168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3736507936507937e-05,
+      "loss": 0.4054,
+      "step": 2169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3733333333333335e-05,
+      "loss": 0.3695,
+      "step": 2170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3730158730158731e-05,
+      "loss": 0.3669,
+      "step": 2171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3726984126984129e-05,
+      "loss": 0.3619,
+      "step": 2172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3723809523809525e-05,
+      "loss": 0.3799,
+      "step": 2173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3720634920634922e-05,
+      "loss": 0.3369,
+      "step": 2174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3717460317460318e-05,
+      "loss": 0.3254,
+      "step": 2175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.3539,
+      "step": 2176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3711111111111112e-05,
+      "loss": 0.3642,
+      "step": 2177
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.370793650793651e-05,
+      "loss": 0.354,
+      "step": 2178
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3704761904761905e-05,
+      "loss": 0.3639,
+      "step": 2179
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3701587301587303e-05,
+      "loss": 0.3645,
+      "step": 2180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3698412698412699e-05,
+      "loss": 0.3828,
+      "step": 2181
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3695238095238097e-05,
+      "loss": 0.339,
+      "step": 2182
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3692063492063493e-05,
+      "loss": 0.3752,
+      "step": 2183
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.368888888888889e-05,
+      "loss": 0.3419,
+      "step": 2184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3685714285714286e-05,
+      "loss": 0.3503,
+      "step": 2185
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3682539682539684e-05,
+      "loss": 0.3361,
+      "step": 2186
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.367936507936508e-05,
+      "loss": 0.3683,
+      "step": 2187
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3676190476190477e-05,
+      "loss": 0.3571,
+      "step": 2188
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3673015873015873e-05,
+      "loss": 0.3557,
+      "step": 2189
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3669841269841271e-05,
+      "loss": 0.3373,
+      "step": 2190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3666666666666667e-05,
+      "loss": 0.351,
+      "step": 2191
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3663492063492065e-05,
+      "loss": 0.3346,
+      "step": 2192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.366031746031746e-05,
+      "loss": 0.3515,
+      "step": 2193
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3657142857142858e-05,
+      "loss": 0.3392,
+      "step": 2194
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3653968253968254e-05,
+      "loss": 0.3513,
+      "step": 2195
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3650793650793652e-05,
+      "loss": 0.3529,
+      "step": 2196
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.364761904761905e-05,
+      "loss": 0.3767,
+      "step": 2197
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3644444444444445e-05,
+      "loss": 0.3556,
+      "step": 2198
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3641269841269841e-05,
+      "loss": 0.359,
+      "step": 2199
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.363809523809524e-05,
+      "loss": 0.3516,
+      "step": 2200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3634920634920636e-05,
+      "loss": 0.3373,
+      "step": 2201
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3631746031746034e-05,
+      "loss": 0.3643,
+      "step": 2202
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3628571428571428e-05,
+      "loss": 0.3297,
+      "step": 2203
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3625396825396826e-05,
+      "loss": 0.3536,
+      "step": 2204
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3622222222222223e-05,
+      "loss": 0.3319,
+      "step": 2205
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.361904761904762e-05,
+      "loss": 0.3283,
+      "step": 2206
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3615873015873019e-05,
+      "loss": 0.3495,
+      "step": 2207
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3612698412698413e-05,
+      "loss": 0.3613,
+      "step": 2208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.360952380952381e-05,
+      "loss": 0.381,
+      "step": 2209
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3606349206349208e-05,
+      "loss": 0.3399,
+      "step": 2210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3603174603174606e-05,
+      "loss": 0.3335,
+      "step": 2211
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.3482,
+      "step": 2212
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3596825396825397e-05,
+      "loss": 0.3787,
+      "step": 2213
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3593650793650795e-05,
+      "loss": 0.3824,
+      "step": 2214
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3590476190476191e-05,
+      "loss": 0.3607,
+      "step": 2215
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3587301587301589e-05,
+      "loss": 0.3231,
+      "step": 2216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3584126984126987e-05,
+      "loss": 0.36,
+      "step": 2217
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3580952380952382e-05,
+      "loss": 0.3459,
+      "step": 2218
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3577777777777778e-05,
+      "loss": 0.319,
+      "step": 2219
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3574603174603176e-05,
+      "loss": 0.3448,
+      "step": 2220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 0.3657,
+      "step": 2221
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.356825396825397e-05,
+      "loss": 0.3351,
+      "step": 2222
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3565079365079365e-05,
+      "loss": 0.3724,
+      "step": 2223
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3561904761904763e-05,
+      "loss": 0.3468,
+      "step": 2224
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.355873015873016e-05,
+      "loss": 0.3475,
+      "step": 2225
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3555555555555557e-05,
+      "loss": 0.3468,
+      "step": 2226
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3552380952380952e-05,
+      "loss": 0.3586,
+      "step": 2227
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.354920634920635e-05,
+      "loss": 0.3484,
+      "step": 2228
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3546031746031748e-05,
+      "loss": 0.3682,
+      "step": 2229
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3542857142857144e-05,
+      "loss": 0.3332,
+      "step": 2230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3539682539682542e-05,
+      "loss": 0.3266,
+      "step": 2231
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3536507936507937e-05,
+      "loss": 0.3401,
+      "step": 2232
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3533333333333333e-05,
+      "loss": 0.3033,
+      "step": 2233
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3530158730158731e-05,
+      "loss": 0.337,
+      "step": 2234
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3526984126984129e-05,
+      "loss": 0.3502,
+      "step": 2235
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3523809523809525e-05,
+      "loss": 0.3424,
+      "step": 2236
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.352063492063492e-05,
+      "loss": 0.3676,
+      "step": 2237
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3517460317460318e-05,
+      "loss": 0.3635,
+      "step": 2238
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3514285714285716e-05,
+      "loss": 0.3433,
+      "step": 2239
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3511111111111112e-05,
+      "loss": 0.348,
+      "step": 2240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.350793650793651e-05,
+      "loss": 0.3609,
+      "step": 2241
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3504761904761905e-05,
+      "loss": 0.3477,
+      "step": 2242
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3501587301587303e-05,
+      "loss": 0.3552,
+      "step": 2243
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.34984126984127e-05,
+      "loss": 0.3602,
+      "step": 2244
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3495238095238097e-05,
+      "loss": 0.3467,
+      "step": 2245
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3492063492063494e-05,
+      "loss": 0.3516,
+      "step": 2246
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3488888888888888e-05,
+      "loss": 0.339,
+      "step": 2247
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3485714285714286e-05,
+      "loss": 0.3828,
+      "step": 2248
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3482539682539684e-05,
+      "loss": 0.3257,
+      "step": 2249
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.347936507936508e-05,
+      "loss": 0.3653,
+      "step": 2250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3476190476190479e-05,
+      "loss": 0.3314,
+      "step": 2251
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3473015873015873e-05,
+      "loss": 0.3381,
+      "step": 2252
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3469841269841271e-05,
+      "loss": 0.3799,
+      "step": 2253
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3466666666666668e-05,
+      "loss": 0.3439,
+      "step": 2254
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3463492063492066e-05,
+      "loss": 0.3219,
+      "step": 2255
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3460317460317462e-05,
+      "loss": 0.3293,
+      "step": 2256
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3457142857142858e-05,
+      "loss": 0.3555,
+      "step": 2257
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3453968253968255e-05,
+      "loss": 0.3447,
+      "step": 2258
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3450793650793653e-05,
+      "loss": 0.3161,
+      "step": 2259
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3447619047619049e-05,
+      "loss": 0.3571,
+      "step": 2260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3444444444444447e-05,
+      "loss": 0.3683,
+      "step": 2261
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3441269841269842e-05,
+      "loss": 0.3399,
+      "step": 2262
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.343809523809524e-05,
+      "loss": 0.3601,
+      "step": 2263
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3434920634920636e-05,
+      "loss": 0.3693,
+      "step": 2264
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3431746031746034e-05,
+      "loss": 0.3402,
+      "step": 2265
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3428571428571429e-05,
+      "loss": 0.3403,
+      "step": 2266
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3425396825396827e-05,
+      "loss": 0.3247,
+      "step": 2267
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3422222222222223e-05,
+      "loss": 0.3559,
+      "step": 2268
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3419047619047621e-05,
+      "loss": 0.364,
+      "step": 2269
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3415873015873017e-05,
+      "loss": 0.345,
+      "step": 2270
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3412698412698413e-05,
+      "loss": 0.3302,
+      "step": 2271
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.340952380952381e-05,
+      "loss": 0.3376,
+      "step": 2272
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3406349206349208e-05,
+      "loss": 0.3599,
+      "step": 2273
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3403174603174604e-05,
+      "loss": 0.3335,
+      "step": 2274
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 0.3498,
+      "step": 2275
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3396825396825397e-05,
+      "loss": 0.3078,
+      "step": 2276
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3393650793650795e-05,
+      "loss": 0.3489,
+      "step": 2277
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3390476190476191e-05,
+      "loss": 0.349,
+      "step": 2278
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3387301587301589e-05,
+      "loss": 0.338,
+      "step": 2279
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3384126984126985e-05,
+      "loss": 0.3791,
+      "step": 2280
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3380952380952382e-05,
+      "loss": 0.3377,
+      "step": 2281
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3377777777777778e-05,
+      "loss": 0.3511,
+      "step": 2282
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3374603174603176e-05,
+      "loss": 0.3408,
+      "step": 2283
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3371428571428572e-05,
+      "loss": 0.377,
+      "step": 2284
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.336825396825397e-05,
+      "loss": 0.3523,
+      "step": 2285
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3365079365079365e-05,
+      "loss": 0.3577,
+      "step": 2286
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3361904761904763e-05,
+      "loss": 0.3276,
+      "step": 2287
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.335873015873016e-05,
+      "loss": 0.3524,
+      "step": 2288
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3355555555555557e-05,
+      "loss": 0.3378,
+      "step": 2289
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3352380952380954e-05,
+      "loss": 0.3466,
+      "step": 2290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.334920634920635e-05,
+      "loss": 0.3545,
+      "step": 2291
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3346031746031746e-05,
+      "loss": 0.3733,
+      "step": 2292
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3342857142857144e-05,
+      "loss": 0.3547,
+      "step": 2293
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.333968253968254e-05,
+      "loss": 0.3581,
+      "step": 2294
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3336507936507939e-05,
+      "loss": 0.3281,
+      "step": 2295
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.3286,
+      "step": 2296
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3330158730158731e-05,
+      "loss": 0.3394,
+      "step": 2297
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3326984126984128e-05,
+      "loss": 0.321,
+      "step": 2298
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3323809523809526e-05,
+      "loss": 0.3406,
+      "step": 2299
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3320634920634924e-05,
+      "loss": 0.356,
+      "step": 2300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3317460317460318e-05,
+      "loss": 0.3468,
+      "step": 2301
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3314285714285715e-05,
+      "loss": 0.3711,
+      "step": 2302
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3311111111111113e-05,
+      "loss": 0.3425,
+      "step": 2303
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3307936507936509e-05,
+      "loss": 0.3645,
+      "step": 2304
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3304761904761905e-05,
+      "loss": 0.3658,
+      "step": 2305
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3301587301587302e-05,
+      "loss": 0.3299,
+      "step": 2306
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.32984126984127e-05,
+      "loss": 0.3459,
+      "step": 2307
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3295238095238096e-05,
+      "loss": 0.3238,
+      "step": 2308
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3292063492063494e-05,
+      "loss": 0.3472,
+      "step": 2309
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3288888888888889e-05,
+      "loss": 0.3473,
+      "step": 2310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3285714285714287e-05,
+      "loss": 0.3607,
+      "step": 2311
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3282539682539683e-05,
+      "loss": 0.3283,
+      "step": 2312
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3279365079365081e-05,
+      "loss": 0.3482,
+      "step": 2313
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3276190476190479e-05,
+      "loss": 0.3503,
+      "step": 2314
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3273015873015874e-05,
+      "loss": 0.3511,
+      "step": 2315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.326984126984127e-05,
+      "loss": 0.3322,
+      "step": 2316
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3266666666666668e-05,
+      "loss": 0.3182,
+      "step": 2317
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3263492063492064e-05,
+      "loss": 0.359,
+      "step": 2318
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3260317460317462e-05,
+      "loss": 0.3543,
+      "step": 2319
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3257142857142857e-05,
+      "loss": 0.3495,
+      "step": 2320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3253968253968255e-05,
+      "loss": 0.3517,
+      "step": 2321
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3250793650793651e-05,
+      "loss": 0.3457,
+      "step": 2322
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3247619047619049e-05,
+      "loss": 0.3435,
+      "step": 2323
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3244444444444447e-05,
+      "loss": 0.3374,
+      "step": 2324
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3241269841269842e-05,
+      "loss": 0.339,
+      "step": 2325
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3238095238095238e-05,
+      "loss": 0.3612,
+      "step": 2326
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3234920634920636e-05,
+      "loss": 0.3162,
+      "step": 2327
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3231746031746034e-05,
+      "loss": 0.3864,
+      "step": 2328
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.322857142857143e-05,
+      "loss": 0.3367,
+      "step": 2329
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3225396825396825e-05,
+      "loss": 0.3783,
+      "step": 2330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3222222222222223e-05,
+      "loss": 0.3511,
+      "step": 2331
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3219047619047621e-05,
+      "loss": 0.3642,
+      "step": 2332
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3215873015873017e-05,
+      "loss": 0.3464,
+      "step": 2333
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3212698412698415e-05,
+      "loss": 0.3261,
+      "step": 2334
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.320952380952381e-05,
+      "loss": 0.3428,
+      "step": 2335
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3206349206349206e-05,
+      "loss": 0.3639,
+      "step": 2336
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3203174603174604e-05,
+      "loss": 0.3236,
+      "step": 2337
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 0.358,
+      "step": 2338
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3196825396825399e-05,
+      "loss": 0.3483,
+      "step": 2339
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3193650793650793e-05,
+      "loss": 0.3513,
+      "step": 2340
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3190476190476191e-05,
+      "loss": 0.3489,
+      "step": 2341
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.318730158730159e-05,
+      "loss": 0.3785,
+      "step": 2342
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3184126984126986e-05,
+      "loss": 0.3522,
+      "step": 2343
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.318095238095238e-05,
+      "loss": 0.3657,
+      "step": 2344
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3177777777777778e-05,
+      "loss": 0.33,
+      "step": 2345
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3174603174603176e-05,
+      "loss": 0.3535,
+      "step": 2346
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3171428571428573e-05,
+      "loss": 0.3673,
+      "step": 2347
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.316825396825397e-05,
+      "loss": 0.3659,
+      "step": 2348
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3165079365079365e-05,
+      "loss": 0.3613,
+      "step": 2349
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3161904761904762e-05,
+      "loss": 0.3557,
+      "step": 2350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.315873015873016e-05,
+      "loss": 0.3591,
+      "step": 2351
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3155555555555558e-05,
+      "loss": 0.3397,
+      "step": 2352
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3152380952380954e-05,
+      "loss": 0.3712,
+      "step": 2353
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3149206349206349e-05,
+      "loss": 0.3563,
+      "step": 2354
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3146031746031747e-05,
+      "loss": 0.3551,
+      "step": 2355
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3142857142857145e-05,
+      "loss": 0.3774,
+      "step": 2356
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3139682539682541e-05,
+      "loss": 0.3467,
+      "step": 2357
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3136507936507939e-05,
+      "loss": 0.33,
+      "step": 2358
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3133333333333334e-05,
+      "loss": 0.3352,
+      "step": 2359
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3130158730158732e-05,
+      "loss": 0.3707,
+      "step": 2360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3126984126984128e-05,
+      "loss": 0.3576,
+      "step": 2361
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3123809523809526e-05,
+      "loss": 0.3515,
+      "step": 2362
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3120634920634922e-05,
+      "loss": 0.3117,
+      "step": 2363
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3117460317460318e-05,
+      "loss": 0.3626,
+      "step": 2364
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3114285714285715e-05,
+      "loss": 0.3514,
+      "step": 2365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3111111111111113e-05,
+      "loss": 0.3564,
+      "step": 2366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3107936507936509e-05,
+      "loss": 0.3709,
+      "step": 2367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3104761904761907e-05,
+      "loss": 0.3626,
+      "step": 2368
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3101587301587302e-05,
+      "loss": 0.3411,
+      "step": 2369
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.30984126984127e-05,
+      "loss": 0.3641,
+      "step": 2370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3095238095238096e-05,
+      "loss": 0.3526,
+      "step": 2371
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3092063492063494e-05,
+      "loss": 0.3311,
+      "step": 2372
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.308888888888889e-05,
+      "loss": 0.3386,
+      "step": 2373
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3085714285714287e-05,
+      "loss": 0.3307,
+      "step": 2374
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3082539682539683e-05,
+      "loss": 0.3305,
+      "step": 2375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3079365079365081e-05,
+      "loss": 0.3597,
+      "step": 2376
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3076190476190477e-05,
+      "loss": 0.3571,
+      "step": 2377
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3073015873015875e-05,
+      "loss": 0.3534,
+      "step": 2378
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.306984126984127e-05,
+      "loss": 0.3387,
+      "step": 2379
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 0.3511,
+      "step": 2380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3063492063492064e-05,
+      "loss": 0.3238,
+      "step": 2381
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3060317460317462e-05,
+      "loss": 0.3446,
+      "step": 2382
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3057142857142857e-05,
+      "loss": 0.3241,
+      "step": 2383
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3053968253968255e-05,
+      "loss": 0.3506,
+      "step": 2384
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3050793650793651e-05,
+      "loss": 0.3598,
+      "step": 2385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.304761904761905e-05,
+      "loss": 0.3501,
+      "step": 2386
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3044444444444446e-05,
+      "loss": 0.3125,
+      "step": 2387
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3041269841269842e-05,
+      "loss": 0.3431,
+      "step": 2388
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3038095238095238e-05,
+      "loss": 0.3602,
+      "step": 2389
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3034920634920636e-05,
+      "loss": 0.361,
+      "step": 2390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3031746031746033e-05,
+      "loss": 0.3442,
+      "step": 2391
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.302857142857143e-05,
+      "loss": 0.3604,
+      "step": 2392
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3025396825396825e-05,
+      "loss": 0.3285,
+      "step": 2393
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3022222222222223e-05,
+      "loss": 0.3388,
+      "step": 2394
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.301904761904762e-05,
+      "loss": 0.3598,
+      "step": 2395
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3015873015873018e-05,
+      "loss": 0.3652,
+      "step": 2396
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3012698412698414e-05,
+      "loss": 0.3468,
+      "step": 2397
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.300952380952381e-05,
+      "loss": 0.369,
+      "step": 2398
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3006349206349207e-05,
+      "loss": 0.348,
+      "step": 2399
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3003174603174605e-05,
+      "loss": 0.366,
+      "step": 2400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.3568,
+      "step": 2401
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2996825396825399e-05,
+      "loss": 0.3592,
+      "step": 2402
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2993650793650794e-05,
+      "loss": 0.3353,
+      "step": 2403
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2990476190476192e-05,
+      "loss": 0.358,
+      "step": 2404
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2987301587301588e-05,
+      "loss": 0.3768,
+      "step": 2405
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2984126984126986e-05,
+      "loss": 0.3496,
+      "step": 2406
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2980952380952382e-05,
+      "loss": 0.3732,
+      "step": 2407
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2977777777777779e-05,
+      "loss": 0.3375,
+      "step": 2408
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2974603174603175e-05,
+      "loss": 0.3541,
+      "step": 2409
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2971428571428573e-05,
+      "loss": 0.3468,
+      "step": 2410
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.296825396825397e-05,
+      "loss": 0.3496,
+      "step": 2411
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2965079365079367e-05,
+      "loss": 0.34,
+      "step": 2412
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2961904761904762e-05,
+      "loss": 0.3157,
+      "step": 2413
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.295873015873016e-05,
+      "loss": 0.3441,
+      "step": 2414
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2955555555555556e-05,
+      "loss": 0.3539,
+      "step": 2415
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2952380952380954e-05,
+      "loss": 0.3489,
+      "step": 2416
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2949206349206352e-05,
+      "loss": 0.3639,
+      "step": 2417
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2946031746031747e-05,
+      "loss": 0.3477,
+      "step": 2418
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2942857142857143e-05,
+      "loss": 0.3607,
+      "step": 2419
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2939682539682541e-05,
+      "loss": 0.3432,
+      "step": 2420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2936507936507937e-05,
+      "loss": 0.3646,
+      "step": 2421
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2933333333333334e-05,
+      "loss": 0.3639,
+      "step": 2422
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.293015873015873e-05,
+      "loss": 0.3447,
+      "step": 2423
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2926984126984128e-05,
+      "loss": 0.3433,
+      "step": 2424
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2923809523809524e-05,
+      "loss": 0.3628,
+      "step": 2425
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2920634920634922e-05,
+      "loss": 0.3312,
+      "step": 2426
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2917460317460317e-05,
+      "loss": 0.3252,
+      "step": 2427
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2914285714285715e-05,
+      "loss": 0.3527,
+      "step": 2428
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2911111111111111e-05,
+      "loss": 0.3408,
+      "step": 2429
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.290793650793651e-05,
+      "loss": 0.3697,
+      "step": 2430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2904761904761907e-05,
+      "loss": 0.3475,
+      "step": 2431
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2901587301587302e-05,
+      "loss": 0.3482,
+      "step": 2432
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2898412698412698e-05,
+      "loss": 0.3724,
+      "step": 2433
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2895238095238096e-05,
+      "loss": 0.4007,
+      "step": 2434
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2892063492063494e-05,
+      "loss": 0.3386,
+      "step": 2435
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.288888888888889e-05,
+      "loss": 0.3322,
+      "step": 2436
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2885714285714285e-05,
+      "loss": 0.358,
+      "step": 2437
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2882539682539683e-05,
+      "loss": 0.3562,
+      "step": 2438
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.287936507936508e-05,
+      "loss": 0.3511,
+      "step": 2439
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2876190476190478e-05,
+      "loss": 0.3511,
+      "step": 2440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2873015873015876e-05,
+      "loss": 0.3559,
+      "step": 2441
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.286984126984127e-05,
+      "loss": 0.351,
+      "step": 2442
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2866666666666667e-05,
+      "loss": 0.3446,
+      "step": 2443
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2863492063492065e-05,
+      "loss": 0.3214,
+      "step": 2444
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2860317460317463e-05,
+      "loss": 0.3427,
+      "step": 2445
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.3357,
+      "step": 2446
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2853968253968254e-05,
+      "loss": 0.3367,
+      "step": 2447
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2850793650793652e-05,
+      "loss": 0.3353,
+      "step": 2448
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.284761904761905e-05,
+      "loss": 0.3512,
+      "step": 2449
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2844444444444446e-05,
+      "loss": 0.3594,
+      "step": 2450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2841269841269844e-05,
+      "loss": 0.3371,
+      "step": 2451
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2838095238095239e-05,
+      "loss": 0.3484,
+      "step": 2452
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2834920634920635e-05,
+      "loss": 0.3514,
+      "step": 2453
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2831746031746033e-05,
+      "loss": 0.3508,
+      "step": 2454
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2828571428571431e-05,
+      "loss": 0.3197,
+      "step": 2455
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2825396825396827e-05,
+      "loss": 0.3206,
+      "step": 2456
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2822222222222222e-05,
+      "loss": 0.3502,
+      "step": 2457
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.281904761904762e-05,
+      "loss": 0.3503,
+      "step": 2458
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2815873015873018e-05,
+      "loss": 0.3505,
+      "step": 2459
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2812698412698414e-05,
+      "loss": 0.3619,
+      "step": 2460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2809523809523809e-05,
+      "loss": 0.3366,
+      "step": 2461
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2806349206349207e-05,
+      "loss": 0.348,
+      "step": 2462
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2803174603174605e-05,
+      "loss": 0.3307,
+      "step": 2463
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.3409,
+      "step": 2464
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2796825396825399e-05,
+      "loss": 0.358,
+      "step": 2465
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2793650793650794e-05,
+      "loss": 0.3495,
+      "step": 2466
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2790476190476192e-05,
+      "loss": 0.344,
+      "step": 2467
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2787301587301588e-05,
+      "loss": 0.371,
+      "step": 2468
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2784126984126986e-05,
+      "loss": 0.3351,
+      "step": 2469
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2780952380952382e-05,
+      "loss": 0.3504,
+      "step": 2470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2777777777777777e-05,
+      "loss": 0.358,
+      "step": 2471
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2774603174603175e-05,
+      "loss": 0.3564,
+      "step": 2472
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2771428571428573e-05,
+      "loss": 0.3645,
+      "step": 2473
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.276825396825397e-05,
+      "loss": 0.3331,
+      "step": 2474
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2765079365079367e-05,
+      "loss": 0.3533,
+      "step": 2475
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2761904761904762e-05,
+      "loss": 0.352,
+      "step": 2476
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.275873015873016e-05,
+      "loss": 0.373,
+      "step": 2477
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2755555555555556e-05,
+      "loss": 0.3527,
+      "step": 2478
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2752380952380954e-05,
+      "loss": 0.3827,
+      "step": 2479
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.274920634920635e-05,
+      "loss": 0.3511,
+      "step": 2480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2746031746031747e-05,
+      "loss": 0.3614,
+      "step": 2481
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2742857142857143e-05,
+      "loss": 0.3196,
+      "step": 2482
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2739682539682541e-05,
+      "loss": 0.358,
+      "step": 2483
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2736507936507938e-05,
+      "loss": 0.3626,
+      "step": 2484
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2733333333333336e-05,
+      "loss": 0.3406,
+      "step": 2485
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.273015873015873e-05,
+      "loss": 0.3369,
+      "step": 2486
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2726984126984128e-05,
+      "loss": 0.3439,
+      "step": 2487
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2723809523809525e-05,
+      "loss": 0.3543,
+      "step": 2488
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2720634920634923e-05,
+      "loss": 0.3564,
+      "step": 2489
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2717460317460319e-05,
+      "loss": 0.3565,
+      "step": 2490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2714285714285715e-05,
+      "loss": 0.3414,
+      "step": 2491
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2711111111111112e-05,
+      "loss": 0.3285,
+      "step": 2492
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.270793650793651e-05,
+      "loss": 0.3486,
+      "step": 2493
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2704761904761906e-05,
+      "loss": 0.3505,
+      "step": 2494
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2701587301587304e-05,
+      "loss": 0.3548,
+      "step": 2495
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2698412698412699e-05,
+      "loss": 0.3617,
+      "step": 2496
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2695238095238097e-05,
+      "loss": 0.3556,
+      "step": 2497
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2692063492063493e-05,
+      "loss": 0.3351,
+      "step": 2498
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2688888888888891e-05,
+      "loss": 0.3711,
+      "step": 2499
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2685714285714286e-05,
+      "loss": 0.3395,
+      "step": 2500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2682539682539684e-05,
+      "loss": 0.3573,
+      "step": 2501
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.267936507936508e-05,
+      "loss": 0.3997,
+      "step": 2502
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2676190476190478e-05,
+      "loss": 0.3466,
+      "step": 2503
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2673015873015874e-05,
+      "loss": 0.33,
+      "step": 2504
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.266984126984127e-05,
+      "loss": 0.3245,
+      "step": 2505
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 0.3379,
+      "step": 2506
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2663492063492065e-05,
+      "loss": 0.3796,
+      "step": 2507
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2660317460317461e-05,
+      "loss": 0.334,
+      "step": 2508
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2657142857142859e-05,
+      "loss": 0.337,
+      "step": 2509
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2653968253968254e-05,
+      "loss": 0.3241,
+      "step": 2510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2650793650793652e-05,
+      "loss": 0.3448,
+      "step": 2511
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2647619047619048e-05,
+      "loss": 0.3447,
+      "step": 2512
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2644444444444446e-05,
+      "loss": 0.3341,
+      "step": 2513
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2641269841269842e-05,
+      "loss": 0.3672,
+      "step": 2514
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2638095238095239e-05,
+      "loss": 0.3363,
+      "step": 2515
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2634920634920635e-05,
+      "loss": 0.352,
+      "step": 2516
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2631746031746033e-05,
+      "loss": 0.3621,
+      "step": 2517
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.262857142857143e-05,
+      "loss": 0.3448,
+      "step": 2518
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2625396825396827e-05,
+      "loss": 0.3753,
+      "step": 2519
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2622222222222222e-05,
+      "loss": 0.3558,
+      "step": 2520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.261904761904762e-05,
+      "loss": 0.381,
+      "step": 2521
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2615873015873016e-05,
+      "loss": 0.3822,
+      "step": 2522
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2612698412698414e-05,
+      "loss": 0.3122,
+      "step": 2523
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.260952380952381e-05,
+      "loss": 0.3533,
+      "step": 2524
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2606349206349207e-05,
+      "loss": 0.36,
+      "step": 2525
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2603174603174603e-05,
+      "loss": 0.3303,
+      "step": 2526
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 0.3644,
+      "step": 2527
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2596825396825398e-05,
+      "loss": 0.351,
+      "step": 2528
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2593650793650796e-05,
+      "loss": 0.3727,
+      "step": 2529
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.259047619047619e-05,
+      "loss": 0.3451,
+      "step": 2530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2587301587301588e-05,
+      "loss": 0.353,
+      "step": 2531
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2584126984126985e-05,
+      "loss": 0.3368,
+      "step": 2532
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2580952380952383e-05,
+      "loss": 0.3555,
+      "step": 2533
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.257777777777778e-05,
+      "loss": 0.3557,
+      "step": 2534
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2574603174603175e-05,
+      "loss": 0.3132,
+      "step": 2535
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2571428571428572e-05,
+      "loss": 0.354,
+      "step": 2536
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.256825396825397e-05,
+      "loss": 0.357,
+      "step": 2537
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2565079365079368e-05,
+      "loss": 0.3475,
+      "step": 2538
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2561904761904762e-05,
+      "loss": 0.365,
+      "step": 2539
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2558730158730159e-05,
+      "loss": 0.353,
+      "step": 2540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2555555555555557e-05,
+      "loss": 0.3414,
+      "step": 2541
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2552380952380953e-05,
+      "loss": 0.3243,
+      "step": 2542
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2549206349206351e-05,
+      "loss": 0.3763,
+      "step": 2543
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2546031746031746e-05,
+      "loss": 0.3211,
+      "step": 2544
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2542857142857144e-05,
+      "loss": 0.361,
+      "step": 2545
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.253968253968254e-05,
+      "loss": 0.3552,
+      "step": 2546
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2536507936507938e-05,
+      "loss": 0.3555,
+      "step": 2547
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 0.3434,
+      "step": 2548
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.253015873015873e-05,
+      "loss": 0.3396,
+      "step": 2549
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2526984126984127e-05,
+      "loss": 0.3261,
+      "step": 2550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2523809523809525e-05,
+      "loss": 0.351,
+      "step": 2551
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2520634920634923e-05,
+      "loss": 0.3281,
+      "step": 2552
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.251746031746032e-05,
+      "loss": 0.3542,
+      "step": 2553
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2514285714285714e-05,
+      "loss": 0.3347,
+      "step": 2554
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2511111111111112e-05,
+      "loss": 0.3515,
+      "step": 2555
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2507936507936508e-05,
+      "loss": 0.3548,
+      "step": 2556
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2504761904761906e-05,
+      "loss": 0.3647,
+      "step": 2557
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2501587301587304e-05,
+      "loss": 0.3736,
+      "step": 2558
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2498412698412699e-05,
+      "loss": 0.338,
+      "step": 2559
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2495238095238095e-05,
+      "loss": 0.3648,
+      "step": 2560
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2492063492063493e-05,
+      "loss": 0.3344,
+      "step": 2561
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2488888888888891e-05,
+      "loss": 0.3557,
+      "step": 2562
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2485714285714287e-05,
+      "loss": 0.3218,
+      "step": 2563
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2482539682539682e-05,
+      "loss": 0.3474,
+      "step": 2564
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.247936507936508e-05,
+      "loss": 0.3609,
+      "step": 2565
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2476190476190478e-05,
+      "loss": 0.3911,
+      "step": 2566
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2473015873015874e-05,
+      "loss": 0.3378,
+      "step": 2567
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2469841269841272e-05,
+      "loss": 0.3621,
+      "step": 2568
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2466666666666667e-05,
+      "loss": 0.3572,
+      "step": 2569
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2463492063492065e-05,
+      "loss": 0.36,
+      "step": 2570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2460317460317461e-05,
+      "loss": 0.3472,
+      "step": 2571
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.245714285714286e-05,
+      "loss": 0.3345,
+      "step": 2572
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2453968253968256e-05,
+      "loss": 0.3323,
+      "step": 2573
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.245079365079365e-05,
+      "loss": 0.3329,
+      "step": 2574
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2447619047619048e-05,
+      "loss": 0.3422,
+      "step": 2575
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 0.3511,
+      "step": 2576
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2441269841269843e-05,
+      "loss": 0.3858,
+      "step": 2577
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.243809523809524e-05,
+      "loss": 0.3466,
+      "step": 2578
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2434920634920635e-05,
+      "loss": 0.3521,
+      "step": 2579
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2431746031746033e-05,
+      "loss": 0.3414,
+      "step": 2580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.242857142857143e-05,
+      "loss": 0.3422,
+      "step": 2581
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2425396825396828e-05,
+      "loss": 0.3592,
+      "step": 2582
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2422222222222222e-05,
+      "loss": 0.3671,
+      "step": 2583
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.241904761904762e-05,
+      "loss": 0.3371,
+      "step": 2584
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2415873015873017e-05,
+      "loss": 0.3591,
+      "step": 2585
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2412698412698415e-05,
+      "loss": 0.3571,
+      "step": 2586
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2409523809523811e-05,
+      "loss": 0.3391,
+      "step": 2587
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2406349206349206e-05,
+      "loss": 0.3394,
+      "step": 2588
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2403174603174604e-05,
+      "loss": 0.3608,
+      "step": 2589
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 0.3426,
+      "step": 2590
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2396825396825398e-05,
+      "loss": 0.3654,
+      "step": 2591
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2393650793650796e-05,
+      "loss": 0.3558,
+      "step": 2592
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.239047619047619e-05,
+      "loss": 0.3458,
+      "step": 2593
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2387301587301589e-05,
+      "loss": 0.3649,
+      "step": 2594
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2384126984126985e-05,
+      "loss": 0.339,
+      "step": 2595
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 0.3745,
+      "step": 2596
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.237777777777778e-05,
+      "loss": 0.3606,
+      "step": 2597
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2374603174603175e-05,
+      "loss": 0.3429,
+      "step": 2598
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2371428571428572e-05,
+      "loss": 0.3759,
+      "step": 2599
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.236825396825397e-05,
+      "loss": 0.3448,
+      "step": 2600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2365079365079366e-05,
+      "loss": 0.3624,
+      "step": 2601
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2361904761904764e-05,
+      "loss": 0.3406,
+      "step": 2602
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2358730158730159e-05,
+      "loss": 0.3518,
+      "step": 2603
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2355555555555557e-05,
+      "loss": 0.3583,
+      "step": 2604
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2352380952380953e-05,
+      "loss": 0.3746,
+      "step": 2605
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2349206349206351e-05,
+      "loss": 0.356,
+      "step": 2606
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2346031746031747e-05,
+      "loss": 0.3717,
+      "step": 2607
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2342857142857144e-05,
+      "loss": 0.362,
+      "step": 2608
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.233968253968254e-05,
+      "loss": 0.3334,
+      "step": 2609
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2336507936507938e-05,
+      "loss": 0.3731,
+      "step": 2610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2333333333333334e-05,
+      "loss": 0.3587,
+      "step": 2611
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2330158730158732e-05,
+      "loss": 0.3627,
+      "step": 2612
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2326984126984127e-05,
+      "loss": 0.3596,
+      "step": 2613
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2323809523809525e-05,
+      "loss": 0.3685,
+      "step": 2614
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2320634920634921e-05,
+      "loss": 0.3642,
+      "step": 2615
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.231746031746032e-05,
+      "loss": 0.3293,
+      "step": 2616
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2314285714285716e-05,
+      "loss": 0.3451,
+      "step": 2617
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2311111111111112e-05,
+      "loss": 0.3333,
+      "step": 2618
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2307936507936508e-05,
+      "loss": 0.3473,
+      "step": 2619
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2304761904761906e-05,
+      "loss": 0.3435,
+      "step": 2620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2301587301587303e-05,
+      "loss": 0.3668,
+      "step": 2621
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2298412698412699e-05,
+      "loss": 0.3479,
+      "step": 2622
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2295238095238095e-05,
+      "loss": 0.3366,
+      "step": 2623
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2292063492063493e-05,
+      "loss": 0.3487,
+      "step": 2624
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.228888888888889e-05,
+      "loss": 0.3304,
+      "step": 2625
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2285714285714288e-05,
+      "loss": 0.3677,
+      "step": 2626
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2282539682539682e-05,
+      "loss": 0.3711,
+      "step": 2627
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.227936507936508e-05,
+      "loss": 0.339,
+      "step": 2628
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2276190476190477e-05,
+      "loss": 0.3355,
+      "step": 2629
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2273015873015875e-05,
+      "loss": 0.331,
+      "step": 2630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2269841269841271e-05,
+      "loss": 0.355,
+      "step": 2631
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 0.3375,
+      "step": 2632
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2263492063492064e-05,
+      "loss": 0.3467,
+      "step": 2633
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2260317460317462e-05,
+      "loss": 0.3377,
+      "step": 2634
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2257142857142858e-05,
+      "loss": 0.3435,
+      "step": 2635
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2253968253968256e-05,
+      "loss": 0.3513,
+      "step": 2636
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.225079365079365e-05,
+      "loss": 0.3388,
+      "step": 2637
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2247619047619049e-05,
+      "loss": 0.3572,
+      "step": 2638
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2244444444444445e-05,
+      "loss": 0.3693,
+      "step": 2639
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2241269841269843e-05,
+      "loss": 0.3474,
+      "step": 2640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.223809523809524e-05,
+      "loss": 0.3805,
+      "step": 2641
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2234920634920636e-05,
+      "loss": 0.3348,
+      "step": 2642
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2231746031746032e-05,
+      "loss": 0.3608,
+      "step": 2643
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.222857142857143e-05,
+      "loss": 0.3182,
+      "step": 2644
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2225396825396826e-05,
+      "loss": 0.3587,
+      "step": 2645
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 0.3451,
+      "step": 2646
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2219047619047619e-05,
+      "loss": 0.3677,
+      "step": 2647
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2215873015873017e-05,
+      "loss": 0.3549,
+      "step": 2648
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2212698412698413e-05,
+      "loss": 0.3516,
+      "step": 2649
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2209523809523811e-05,
+      "loss": 0.3608,
+      "step": 2650
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2206349206349209e-05,
+      "loss": 0.3399,
+      "step": 2651
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2203174603174604e-05,
+      "loss": 0.3405,
+      "step": 2652
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.22e-05,
+      "loss": 0.3302,
+      "step": 2653
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2196825396825398e-05,
+      "loss": 0.3421,
+      "step": 2654
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2193650793650796e-05,
+      "loss": 0.3427,
+      "step": 2655
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2190476190476192e-05,
+      "loss": 0.3347,
+      "step": 2656
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2187301587301587e-05,
+      "loss": 0.3501,
+      "step": 2657
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2184126984126985e-05,
+      "loss": 0.3349,
+      "step": 2658
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2180952380952381e-05,
+      "loss": 0.3228,
+      "step": 2659
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.217777777777778e-05,
+      "loss": 0.3394,
+      "step": 2660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2174603174603174e-05,
+      "loss": 0.3661,
+      "step": 2661
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2171428571428572e-05,
+      "loss": 0.3514,
+      "step": 2662
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2168253968253968e-05,
+      "loss": 0.3444,
+      "step": 2663
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2165079365079366e-05,
+      "loss": 0.3513,
+      "step": 2664
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2161904761904764e-05,
+      "loss": 0.3704,
+      "step": 2665
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2158730158730159e-05,
+      "loss": 0.3327,
+      "step": 2666
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2155555555555555e-05,
+      "loss": 0.359,
+      "step": 2667
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2152380952380953e-05,
+      "loss": 0.3593,
+      "step": 2668
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2149206349206351e-05,
+      "loss": 0.3381,
+      "step": 2669
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2146031746031748e-05,
+      "loss": 0.3428,
+      "step": 2670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 0.3422,
+      "step": 2671
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.213968253968254e-05,
+      "loss": 0.3593,
+      "step": 2672
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2136507936507938e-05,
+      "loss": 0.3398,
+      "step": 2673
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2133333333333335e-05,
+      "loss": 0.3209,
+      "step": 2674
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2130158730158733e-05,
+      "loss": 0.3882,
+      "step": 2675
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2126984126984127e-05,
+      "loss": 0.3252,
+      "step": 2676
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2123809523809524e-05,
+      "loss": 0.3519,
+      "step": 2677
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2120634920634922e-05,
+      "loss": 0.3573,
+      "step": 2678
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.211746031746032e-05,
+      "loss": 0.349,
+      "step": 2679
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2114285714285716e-05,
+      "loss": 0.3344,
+      "step": 2680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.211111111111111e-05,
+      "loss": 0.3609,
+      "step": 2681
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2107936507936509e-05,
+      "loss": 0.3264,
+      "step": 2682
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2104761904761907e-05,
+      "loss": 0.3448,
+      "step": 2683
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2101587301587303e-05,
+      "loss": 0.3533,
+      "step": 2684
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2098412698412701e-05,
+      "loss": 0.3477,
+      "step": 2685
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2095238095238096e-05,
+      "loss": 0.3536,
+      "step": 2686
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2092063492063494e-05,
+      "loss": 0.3541,
+      "step": 2687
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.208888888888889e-05,
+      "loss": 0.3723,
+      "step": 2688
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2085714285714288e-05,
+      "loss": 0.3586,
+      "step": 2689
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2082539682539684e-05,
+      "loss": 0.3243,
+      "step": 2690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2079365079365079e-05,
+      "loss": 0.3261,
+      "step": 2691
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2076190476190477e-05,
+      "loss": 0.351,
+      "step": 2692
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2073015873015875e-05,
+      "loss": 0.3563,
+      "step": 2693
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2069841269841271e-05,
+      "loss": 0.3531,
+      "step": 2694
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.206666666666667e-05,
+      "loss": 0.3305,
+      "step": 2695
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2063492063492064e-05,
+      "loss": 0.3541,
+      "step": 2696
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2060317460317462e-05,
+      "loss": 0.3706,
+      "step": 2697
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2057142857142858e-05,
+      "loss": 0.3654,
+      "step": 2698
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2053968253968256e-05,
+      "loss": 0.3375,
+      "step": 2699
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.205079365079365e-05,
+      "loss": 0.3644,
+      "step": 2700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2047619047619049e-05,
+      "loss": 0.3172,
+      "step": 2701
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2044444444444445e-05,
+      "loss": 0.3523,
+      "step": 2702
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2041269841269843e-05,
+      "loss": 0.3453,
+      "step": 2703
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.203809523809524e-05,
+      "loss": 0.3452,
+      "step": 2704
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2034920634920636e-05,
+      "loss": 0.3612,
+      "step": 2705
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2031746031746032e-05,
+      "loss": 0.3517,
+      "step": 2706
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.202857142857143e-05,
+      "loss": 0.3544,
+      "step": 2707
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2025396825396826e-05,
+      "loss": 0.329,
+      "step": 2708
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2022222222222224e-05,
+      "loss": 0.3364,
+      "step": 2709
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2019047619047619e-05,
+      "loss": 0.3344,
+      "step": 2710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2015873015873017e-05,
+      "loss": 0.3408,
+      "step": 2711
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2012698412698413e-05,
+      "loss": 0.3292,
+      "step": 2712
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2009523809523811e-05,
+      "loss": 0.3458,
+      "step": 2713
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2006349206349208e-05,
+      "loss": 0.3491,
+      "step": 2714
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2003174603174604e-05,
+      "loss": 0.3493,
+      "step": 2715
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2e-05,
+      "loss": 0.3285,
+      "step": 2716
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1996825396825398e-05,
+      "loss": 0.3439,
+      "step": 2717
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1993650793650795e-05,
+      "loss": 0.3291,
+      "step": 2718
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1990476190476193e-05,
+      "loss": 0.3573,
+      "step": 2719
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1987301587301587e-05,
+      "loss": 0.3505,
+      "step": 2720
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1984126984126985e-05,
+      "loss": 0.368,
+      "step": 2721
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1980952380952382e-05,
+      "loss": 0.3429,
+      "step": 2722
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.197777777777778e-05,
+      "loss": 0.3458,
+      "step": 2723
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1974603174603176e-05,
+      "loss": 0.3335,
+      "step": 2724
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1971428571428572e-05,
+      "loss": 0.3406,
+      "step": 2725
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1968253968253969e-05,
+      "loss": 0.3342,
+      "step": 2726
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1965079365079367e-05,
+      "loss": 0.3561,
+      "step": 2727
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1961904761904763e-05,
+      "loss": 0.3279,
+      "step": 2728
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1958730158730161e-05,
+      "loss": 0.3415,
+      "step": 2729
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1955555555555556e-05,
+      "loss": 0.337,
+      "step": 2730
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1952380952380954e-05,
+      "loss": 0.3452,
+      "step": 2731
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.194920634920635e-05,
+      "loss": 0.3411,
+      "step": 2732
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1946031746031748e-05,
+      "loss": 0.3221,
+      "step": 2733
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1942857142857144e-05,
+      "loss": 0.3614,
+      "step": 2734
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.193968253968254e-05,
+      "loss": 0.3398,
+      "step": 2735
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1936507936507937e-05,
+      "loss": 0.3567,
+      "step": 2736
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1933333333333335e-05,
+      "loss": 0.3229,
+      "step": 2737
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1930158730158731e-05,
+      "loss": 0.3434,
+      "step": 2738
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1926984126984127e-05,
+      "loss": 0.3907,
+      "step": 2739
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1923809523809524e-05,
+      "loss": 0.3361,
+      "step": 2740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1920634920634922e-05,
+      "loss": 0.3538,
+      "step": 2741
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1917460317460318e-05,
+      "loss": 0.363,
+      "step": 2742
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1914285714285716e-05,
+      "loss": 0.3396,
+      "step": 2743
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.191111111111111e-05,
+      "loss": 0.3306,
+      "step": 2744
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1907936507936509e-05,
+      "loss": 0.3475,
+      "step": 2745
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.3632,
+      "step": 2746
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1901587301587303e-05,
+      "loss": 0.3433,
+      "step": 2747
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.18984126984127e-05,
+      "loss": 0.3151,
+      "step": 2748
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1895238095238096e-05,
+      "loss": 0.3588,
+      "step": 2749
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1892063492063492e-05,
+      "loss": 0.3287,
+      "step": 2750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.188888888888889e-05,
+      "loss": 0.3437,
+      "step": 2751
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1885714285714286e-05,
+      "loss": 0.3446,
+      "step": 2752
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1882539682539684e-05,
+      "loss": 0.371,
+      "step": 2753
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1879365079365079e-05,
+      "loss": 0.3818,
+      "step": 2754
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1876190476190477e-05,
+      "loss": 0.3324,
+      "step": 2755
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1873015873015873e-05,
+      "loss": 0.3427,
+      "step": 2756
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1869841269841271e-05,
+      "loss": 0.3653,
+      "step": 2757
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.186666666666667e-05,
+      "loss": 0.346,
+      "step": 2758
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1863492063492064e-05,
+      "loss": 0.3316,
+      "step": 2759
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.186031746031746e-05,
+      "loss": 0.3155,
+      "step": 2760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1857142857142858e-05,
+      "loss": 0.3532,
+      "step": 2761
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1853968253968255e-05,
+      "loss": 0.3641,
+      "step": 2762
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1850793650793653e-05,
+      "loss": 0.347,
+      "step": 2763
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1847619047619047e-05,
+      "loss": 0.3522,
+      "step": 2764
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1844444444444445e-05,
+      "loss": 0.348,
+      "step": 2765
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1841269841269842e-05,
+      "loss": 0.3362,
+      "step": 2766
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.183809523809524e-05,
+      "loss": 0.3591,
+      "step": 2767
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1834920634920638e-05,
+      "loss": 0.3345,
+      "step": 2768
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1831746031746032e-05,
+      "loss": 0.3338,
+      "step": 2769
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1828571428571429e-05,
+      "loss": 0.384,
+      "step": 2770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1825396825396827e-05,
+      "loss": 0.3554,
+      "step": 2771
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1822222222222225e-05,
+      "loss": 0.3397,
+      "step": 2772
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1819047619047621e-05,
+      "loss": 0.3669,
+      "step": 2773
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1815873015873016e-05,
+      "loss": 0.3412,
+      "step": 2774
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1812698412698414e-05,
+      "loss": 0.3653,
+      "step": 2775
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.180952380952381e-05,
+      "loss": 0.3301,
+      "step": 2776
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1806349206349208e-05,
+      "loss": 0.3726,
+      "step": 2777
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1803174603174603e-05,
+      "loss": 0.34,
+      "step": 2778
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.18e-05,
+      "loss": 0.3517,
+      "step": 2779
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1796825396825397e-05,
+      "loss": 0.3717,
+      "step": 2780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1793650793650795e-05,
+      "loss": 0.3689,
+      "step": 2781
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1790476190476193e-05,
+      "loss": 0.3489,
+      "step": 2782
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1787301587301587e-05,
+      "loss": 0.3267,
+      "step": 2783
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1784126984126984e-05,
+      "loss": 0.3494,
+      "step": 2784
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1780952380952382e-05,
+      "loss": 0.3525,
+      "step": 2785
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.177777777777778e-05,
+      "loss": 0.3419,
+      "step": 2786
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1774603174603176e-05,
+      "loss": 0.3672,
+      "step": 2787
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.177142857142857e-05,
+      "loss": 0.3715,
+      "step": 2788
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1768253968253969e-05,
+      "loss": 0.3276,
+      "step": 2789
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1765079365079367e-05,
+      "loss": 0.3761,
+      "step": 2790
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1761904761904763e-05,
+      "loss": 0.3265,
+      "step": 2791
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1758730158730161e-05,
+      "loss": 0.3408,
+      "step": 2792
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1755555555555556e-05,
+      "loss": 0.3668,
+      "step": 2793
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1752380952380952e-05,
+      "loss": 0.3438,
+      "step": 2794
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.174920634920635e-05,
+      "loss": 0.3613,
+      "step": 2795
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1746031746031748e-05,
+      "loss": 0.3414,
+      "step": 2796
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1742857142857144e-05,
+      "loss": 0.3432,
+      "step": 2797
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1739682539682539e-05,
+      "loss": 0.3595,
+      "step": 2798
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1736507936507937e-05,
+      "loss": 0.3643,
+      "step": 2799
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 0.3607,
+      "step": 2800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1730158730158731e-05,
+      "loss": 0.354,
+      "step": 2801
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.172698412698413e-05,
+      "loss": 0.3556,
+      "step": 2802
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1723809523809524e-05,
+      "loss": 0.3536,
+      "step": 2803
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1720634920634922e-05,
+      "loss": 0.3459,
+      "step": 2804
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1717460317460318e-05,
+      "loss": 0.3376,
+      "step": 2805
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1714285714285716e-05,
+      "loss": 0.3598,
+      "step": 2806
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1711111111111113e-05,
+      "loss": 0.3489,
+      "step": 2807
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1707936507936509e-05,
+      "loss": 0.3536,
+      "step": 2808
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1704761904761905e-05,
+      "loss": 0.3337,
+      "step": 2809
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1701587301587303e-05,
+      "loss": 0.3481,
+      "step": 2810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.16984126984127e-05,
+      "loss": 0.3655,
+      "step": 2811
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1695238095238098e-05,
+      "loss": 0.3464,
+      "step": 2812
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1692063492063492e-05,
+      "loss": 0.3551,
+      "step": 2813
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.168888888888889e-05,
+      "loss": 0.3334,
+      "step": 2814
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1685714285714287e-05,
+      "loss": 0.3228,
+      "step": 2815
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1682539682539685e-05,
+      "loss": 0.3402,
+      "step": 2816
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.167936507936508e-05,
+      "loss": 0.3693,
+      "step": 2817
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1676190476190477e-05,
+      "loss": 0.363,
+      "step": 2818
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1673015873015874e-05,
+      "loss": 0.375,
+      "step": 2819
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1669841269841272e-05,
+      "loss": 0.3513,
+      "step": 2820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.3555,
+      "step": 2821
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1663492063492064e-05,
+      "loss": 0.3533,
+      "step": 2822
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.166031746031746e-05,
+      "loss": 0.3639,
+      "step": 2823
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1657142857142859e-05,
+      "loss": 0.3614,
+      "step": 2824
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1653968253968255e-05,
+      "loss": 0.3578,
+      "step": 2825
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1650793650793653e-05,
+      "loss": 0.3299,
+      "step": 2826
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1647619047619048e-05,
+      "loss": 0.3394,
+      "step": 2827
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1644444444444446e-05,
+      "loss": 0.3394,
+      "step": 2828
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1641269841269842e-05,
+      "loss": 0.3738,
+      "step": 2829
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.163809523809524e-05,
+      "loss": 0.3492,
+      "step": 2830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1634920634920636e-05,
+      "loss": 0.3544,
+      "step": 2831
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1631746031746032e-05,
+      "loss": 0.3314,
+      "step": 2832
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1628571428571429e-05,
+      "loss": 0.3558,
+      "step": 2833
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1625396825396827e-05,
+      "loss": 0.3345,
+      "step": 2834
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1622222222222223e-05,
+      "loss": 0.3673,
+      "step": 2835
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1619047619047621e-05,
+      "loss": 0.3691,
+      "step": 2836
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1615873015873016e-05,
+      "loss": 0.3664,
+      "step": 2837
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1612698412698414e-05,
+      "loss": 0.3618,
+      "step": 2838
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.160952380952381e-05,
+      "loss": 0.3696,
+      "step": 2839
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1606349206349208e-05,
+      "loss": 0.3354,
+      "step": 2840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1603174603174604e-05,
+      "loss": 0.3645,
+      "step": 2841
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.16e-05,
+      "loss": 0.3426,
+      "step": 2842
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1596825396825397e-05,
+      "loss": 0.3615,
+      "step": 2843
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1593650793650795e-05,
+      "loss": 0.3721,
+      "step": 2844
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1590476190476191e-05,
+      "loss": 0.3661,
+      "step": 2845
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.158730158730159e-05,
+      "loss": 0.381,
+      "step": 2846
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1584126984126984e-05,
+      "loss": 0.3566,
+      "step": 2847
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1580952380952382e-05,
+      "loss": 0.3353,
+      "step": 2848
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1577777777777778e-05,
+      "loss": 0.365,
+      "step": 2849
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1574603174603176e-05,
+      "loss": 0.3525,
+      "step": 2850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1571428571428573e-05,
+      "loss": 0.3716,
+      "step": 2851
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1568253968253969e-05,
+      "loss": 0.3286,
+      "step": 2852
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1565079365079365e-05,
+      "loss": 0.3395,
+      "step": 2853
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1561904761904763e-05,
+      "loss": 0.3562,
+      "step": 2854
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.155873015873016e-05,
+      "loss": 0.359,
+      "step": 2855
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 0.3599,
+      "step": 2856
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1552380952380952e-05,
+      "loss": 0.36,
+      "step": 2857
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.154920634920635e-05,
+      "loss": 0.333,
+      "step": 2858
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1546031746031747e-05,
+      "loss": 0.3581,
+      "step": 2859
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1542857142857145e-05,
+      "loss": 0.3514,
+      "step": 2860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.153968253968254e-05,
+      "loss": 0.3614,
+      "step": 2861
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1536507936507937e-05,
+      "loss": 0.3472,
+      "step": 2862
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1533333333333334e-05,
+      "loss": 0.3621,
+      "step": 2863
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1530158730158732e-05,
+      "loss": 0.3674,
+      "step": 2864
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1526984126984128e-05,
+      "loss": 0.3583,
+      "step": 2865
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1523809523809524e-05,
+      "loss": 0.3662,
+      "step": 2866
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.152063492063492e-05,
+      "loss": 0.359,
+      "step": 2867
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1517460317460319e-05,
+      "loss": 0.361,
+      "step": 2868
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1514285714285715e-05,
+      "loss": 0.3456,
+      "step": 2869
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1511111111111113e-05,
+      "loss": 0.3931,
+      "step": 2870
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1507936507936508e-05,
+      "loss": 0.3429,
+      "step": 2871
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1504761904761906e-05,
+      "loss": 0.3595,
+      "step": 2872
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1501587301587302e-05,
+      "loss": 0.3449,
+      "step": 2873
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.14984126984127e-05,
+      "loss": 0.3509,
+      "step": 2874
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1495238095238098e-05,
+      "loss": 0.3552,
+      "step": 2875
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1492063492063492e-05,
+      "loss": 0.3615,
+      "step": 2876
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1488888888888889e-05,
+      "loss": 0.3517,
+      "step": 2877
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1485714285714287e-05,
+      "loss": 0.3384,
+      "step": 2878
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1482539682539683e-05,
+      "loss": 0.3556,
+      "step": 2879
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1479365079365081e-05,
+      "loss": 0.3478,
+      "step": 2880
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1476190476190476e-05,
+      "loss": 0.3453,
+      "step": 2881
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1473015873015874e-05,
+      "loss": 0.3366,
+      "step": 2882
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.146984126984127e-05,
+      "loss": 0.3342,
+      "step": 2883
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 0.3331,
+      "step": 2884
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1463492063492066e-05,
+      "loss": 0.3556,
+      "step": 2885
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.146031746031746e-05,
+      "loss": 0.34,
+      "step": 2886
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1457142857142857e-05,
+      "loss": 0.3436,
+      "step": 2887
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1453968253968255e-05,
+      "loss": 0.3548,
+      "step": 2888
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1450793650793653e-05,
+      "loss": 0.3707,
+      "step": 2889
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.144761904761905e-05,
+      "loss": 0.3615,
+      "step": 2890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1444444444444444e-05,
+      "loss": 0.354,
+      "step": 2891
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1441269841269842e-05,
+      "loss": 0.3825,
+      "step": 2892
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.143809523809524e-05,
+      "loss": 0.3496,
+      "step": 2893
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1434920634920636e-05,
+      "loss": 0.3638,
+      "step": 2894
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1431746031746031e-05,
+      "loss": 0.3552,
+      "step": 2895
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.352,
+      "step": 2896
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1425396825396825e-05,
+      "loss": 0.3581,
+      "step": 2897
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1422222222222223e-05,
+      "loss": 0.3779,
+      "step": 2898
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1419047619047621e-05,
+      "loss": 0.3389,
+      "step": 2899
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1415873015873016e-05,
+      "loss": 0.354,
+      "step": 2900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1412698412698412e-05,
+      "loss": 0.3582,
+      "step": 2901
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.140952380952381e-05,
+      "loss": 0.3221,
+      "step": 2902
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1406349206349208e-05,
+      "loss": 0.3448,
+      "step": 2903
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1403174603174605e-05,
+      "loss": 0.3712,
+      "step": 2904
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.14e-05,
+      "loss": 0.3623,
+      "step": 2905
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1396825396825397e-05,
+      "loss": 0.3522,
+      "step": 2906
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1393650793650795e-05,
+      "loss": 0.3238,
+      "step": 2907
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1390476190476192e-05,
+      "loss": 0.3284,
+      "step": 2908
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.138730158730159e-05,
+      "loss": 0.3545,
+      "step": 2909
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1384126984126984e-05,
+      "loss": 0.3453,
+      "step": 2910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1380952380952382e-05,
+      "loss": 0.3679,
+      "step": 2911
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1377777777777779e-05,
+      "loss": 0.3385,
+      "step": 2912
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1374603174603177e-05,
+      "loss": 0.3355,
+      "step": 2913
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1371428571428573e-05,
+      "loss": 0.3402,
+      "step": 2914
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1368253968253968e-05,
+      "loss": 0.3754,
+      "step": 2915
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1365079365079366e-05,
+      "loss": 0.32,
+      "step": 2916
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1361904761904764e-05,
+      "loss": 0.3342,
+      "step": 2917
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.135873015873016e-05,
+      "loss": 0.3416,
+      "step": 2918
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1355555555555558e-05,
+      "loss": 0.3267,
+      "step": 2919
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1352380952380953e-05,
+      "loss": 0.3424,
+      "step": 2920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.134920634920635e-05,
+      "loss": 0.3466,
+      "step": 2921
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1346031746031747e-05,
+      "loss": 0.343,
+      "step": 2922
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1342857142857145e-05,
+      "loss": 0.3867,
+      "step": 2923
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1339682539682541e-05,
+      "loss": 0.3369,
+      "step": 2924
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1336507936507937e-05,
+      "loss": 0.3347,
+      "step": 2925
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 0.3634,
+      "step": 2926
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1330158730158732e-05,
+      "loss": 0.3323,
+      "step": 2927
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1326984126984128e-05,
+      "loss": 0.3519,
+      "step": 2928
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1323809523809526e-05,
+      "loss": 0.3424,
+      "step": 2929
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.132063492063492e-05,
+      "loss": 0.3372,
+      "step": 2930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1317460317460319e-05,
+      "loss": 0.3484,
+      "step": 2931
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1314285714285715e-05,
+      "loss": 0.3318,
+      "step": 2932
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1311111111111113e-05,
+      "loss": 0.3467,
+      "step": 2933
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1307936507936508e-05,
+      "loss": 0.3379,
+      "step": 2934
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1304761904761906e-05,
+      "loss": 0.3223,
+      "step": 2935
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1301587301587302e-05,
+      "loss": 0.3463,
+      "step": 2936
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.12984126984127e-05,
+      "loss": 0.3179,
+      "step": 2937
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1295238095238096e-05,
+      "loss": 0.3604,
+      "step": 2938
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1292063492063493e-05,
+      "loss": 0.3416,
+      "step": 2939
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1288888888888889e-05,
+      "loss": 0.3603,
+      "step": 2940
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1285714285714287e-05,
+      "loss": 0.3739,
+      "step": 2941
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1282539682539683e-05,
+      "loss": 0.3664,
+      "step": 2942
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1279365079365081e-05,
+      "loss": 0.356,
+      "step": 2943
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1276190476190476e-05,
+      "loss": 0.3523,
+      "step": 2944
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1273015873015874e-05,
+      "loss": 0.345,
+      "step": 2945
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.126984126984127e-05,
+      "loss": 0.3704,
+      "step": 2946
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1266666666666668e-05,
+      "loss": 0.365,
+      "step": 2947
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1263492063492065e-05,
+      "loss": 0.3424,
+      "step": 2948
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1260317460317461e-05,
+      "loss": 0.3481,
+      "step": 2949
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1257142857142857e-05,
+      "loss": 0.3566,
+      "step": 2950
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1253968253968255e-05,
+      "loss": 0.3255,
+      "step": 2951
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1250793650793652e-05,
+      "loss": 0.3509,
+      "step": 2952
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.124761904761905e-05,
+      "loss": 0.3372,
+      "step": 2953
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1244444444444444e-05,
+      "loss": 0.3559,
+      "step": 2954
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1241269841269842e-05,
+      "loss": 0.375,
+      "step": 2955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1238095238095239e-05,
+      "loss": 0.344,
+      "step": 2956
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1234920634920637e-05,
+      "loss": 0.3521,
+      "step": 2957
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1231746031746033e-05,
+      "loss": 0.3278,
+      "step": 2958
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.122857142857143e-05,
+      "loss": 0.3422,
+      "step": 2959
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1225396825396826e-05,
+      "loss": 0.346,
+      "step": 2960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1222222222222224e-05,
+      "loss": 0.3657,
+      "step": 2961
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.121904761904762e-05,
+      "loss": 0.3562,
+      "step": 2962
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1215873015873018e-05,
+      "loss": 0.3713,
+      "step": 2963
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1212698412698413e-05,
+      "loss": 0.3438,
+      "step": 2964
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.120952380952381e-05,
+      "loss": 0.355,
+      "step": 2965
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1206349206349207e-05,
+      "loss": 0.3379,
+      "step": 2966
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1203174603174605e-05,
+      "loss": 0.3814,
+      "step": 2967
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.3335,
+      "step": 2968
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1196825396825398e-05,
+      "loss": 0.3213,
+      "step": 2969
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1193650793650794e-05,
+      "loss": 0.3717,
+      "step": 2970
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1190476190476192e-05,
+      "loss": 0.3717,
+      "step": 2971
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1187301587301588e-05,
+      "loss": 0.3587,
+      "step": 2972
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1184126984126984e-05,
+      "loss": 0.3373,
+      "step": 2973
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.118095238095238e-05,
+      "loss": 0.3438,
+      "step": 2974
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1177777777777779e-05,
+      "loss": 0.3531,
+      "step": 2975
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1174603174603175e-05,
+      "loss": 0.3328,
+      "step": 2976
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1171428571428573e-05,
+      "loss": 0.3194,
+      "step": 2977
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1168253968253968e-05,
+      "loss": 0.3449,
+      "step": 2978
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1165079365079366e-05,
+      "loss": 0.3333,
+      "step": 2979
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1161904761904762e-05,
+      "loss": 0.3382,
+      "step": 2980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.115873015873016e-05,
+      "loss": 0.3278,
+      "step": 2981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1155555555555556e-05,
+      "loss": 0.3654,
+      "step": 2982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1152380952380953e-05,
+      "loss": 0.3507,
+      "step": 2983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1149206349206349e-05,
+      "loss": 0.3467,
+      "step": 2984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1146031746031747e-05,
+      "loss": 0.3536,
+      "step": 2985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1142857142857143e-05,
+      "loss": 0.3238,
+      "step": 2986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1139682539682541e-05,
+      "loss": 0.3217,
+      "step": 2987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1136507936507936e-05,
+      "loss": 0.3589,
+      "step": 2988
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1133333333333334e-05,
+      "loss": 0.3232,
+      "step": 2989
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.113015873015873e-05,
+      "loss": 0.3565,
+      "step": 2990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1126984126984128e-05,
+      "loss": 0.346,
+      "step": 2991
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1123809523809526e-05,
+      "loss": 0.3662,
+      "step": 2992
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1120634920634921e-05,
+      "loss": 0.3765,
+      "step": 2993
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1117460317460317e-05,
+      "loss": 0.3457,
+      "step": 2994
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1114285714285715e-05,
+      "loss": 0.3567,
+      "step": 2995
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.3385,
+      "step": 2996
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.110793650793651e-05,
+      "loss": 0.3692,
+      "step": 2997
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1104761904761904e-05,
+      "loss": 0.3399,
+      "step": 2998
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1101587301587302e-05,
+      "loss": 0.3582,
+      "step": 2999
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1098412698412699e-05,
+      "loss": 0.3553,
+      "step": 3000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1095238095238097e-05,
+      "loss": 0.3457,
+      "step": 3001
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1092063492063495e-05,
+      "loss": 0.3476,
+      "step": 3002
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.108888888888889e-05,
+      "loss": 0.3752,
+      "step": 3003
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1085714285714286e-05,
+      "loss": 0.3761,
+      "step": 3004
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1082539682539684e-05,
+      "loss": 0.3397,
+      "step": 3005
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1079365079365082e-05,
+      "loss": 0.3327,
+      "step": 3006
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1076190476190478e-05,
+      "loss": 0.3419,
+      "step": 3007
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1073015873015873e-05,
+      "loss": 0.3568,
+      "step": 3008
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.106984126984127e-05,
+      "loss": 0.3834,
+      "step": 3009
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1066666666666669e-05,
+      "loss": 0.3661,
+      "step": 3010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1063492063492065e-05,
+      "loss": 0.3516,
+      "step": 3011
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.106031746031746e-05,
+      "loss": 0.3429,
+      "step": 3012
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1057142857142858e-05,
+      "loss": 0.3694,
+      "step": 3013
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1053968253968254e-05,
+      "loss": 0.3444,
+      "step": 3014
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1050793650793652e-05,
+      "loss": 0.3295,
+      "step": 3015
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.104761904761905e-05,
+      "loss": 0.3459,
+      "step": 3016
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1044444444444444e-05,
+      "loss": 0.3581,
+      "step": 3017
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.104126984126984e-05,
+      "loss": 0.3667,
+      "step": 3018
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1038095238095239e-05,
+      "loss": 0.3513,
+      "step": 3019
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1034920634920637e-05,
+      "loss": 0.3392,
+      "step": 3020
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1031746031746033e-05,
+      "loss": 0.3453,
+      "step": 3021
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1028571428571428e-05,
+      "loss": 0.3377,
+      "step": 3022
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1025396825396826e-05,
+      "loss": 0.3385,
+      "step": 3023
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1022222222222224e-05,
+      "loss": 0.3399,
+      "step": 3024
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.101904761904762e-05,
+      "loss": 0.3284,
+      "step": 3025
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1015873015873018e-05,
+      "loss": 0.3505,
+      "step": 3026
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1012698412698413e-05,
+      "loss": 0.353,
+      "step": 3027
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.100952380952381e-05,
+      "loss": 0.3251,
+      "step": 3028
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1006349206349207e-05,
+      "loss": 0.3427,
+      "step": 3029
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1003174603174605e-05,
+      "loss": 0.3465,
+      "step": 3030
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.3708,
+      "step": 3031
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0996825396825396e-05,
+      "loss": 0.3461,
+      "step": 3032
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0993650793650794e-05,
+      "loss": 0.3483,
+      "step": 3033
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0990476190476192e-05,
+      "loss": 0.3568,
+      "step": 3034
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0987301587301588e-05,
+      "loss": 0.3509,
+      "step": 3035
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0984126984126986e-05,
+      "loss": 0.3565,
+      "step": 3036
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0980952380952381e-05,
+      "loss": 0.366,
+      "step": 3037
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0977777777777779e-05,
+      "loss": 0.3566,
+      "step": 3038
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0974603174603175e-05,
+      "loss": 0.3479,
+      "step": 3039
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0971428571428573e-05,
+      "loss": 0.3518,
+      "step": 3040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.096825396825397e-05,
+      "loss": 0.3468,
+      "step": 3041
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0965079365079366e-05,
+      "loss": 0.3765,
+      "step": 3042
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0961904761904762e-05,
+      "loss": 0.3741,
+      "step": 3043
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.095873015873016e-05,
+      "loss": 0.3772,
+      "step": 3044
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0955555555555557e-05,
+      "loss": 0.344,
+      "step": 3045
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0952380952380955e-05,
+      "loss": 0.3389,
+      "step": 3046
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.094920634920635e-05,
+      "loss": 0.3271,
+      "step": 3047
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0946031746031747e-05,
+      "loss": 0.3396,
+      "step": 3048
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0942857142857144e-05,
+      "loss": 0.3203,
+      "step": 3049
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0939682539682542e-05,
+      "loss": 0.3516,
+      "step": 3050
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0936507936507936e-05,
+      "loss": 0.3345,
+      "step": 3051
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 0.3443,
+      "step": 3052
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.093015873015873e-05,
+      "loss": 0.351,
+      "step": 3053
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0926984126984129e-05,
+      "loss": 0.3296,
+      "step": 3054
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0923809523809525e-05,
+      "loss": 0.3492,
+      "step": 3055
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0920634920634921e-05,
+      "loss": 0.385,
+      "step": 3056
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0917460317460318e-05,
+      "loss": 0.3234,
+      "step": 3057
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0914285714285716e-05,
+      "loss": 0.3625,
+      "step": 3058
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0911111111111112e-05,
+      "loss": 0.362,
+      "step": 3059
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.090793650793651e-05,
+      "loss": 0.334,
+      "step": 3060
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0904761904761905e-05,
+      "loss": 0.3129,
+      "step": 3061
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0901587301587303e-05,
+      "loss": 0.3578,
+      "step": 3062
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0898412698412699e-05,
+      "loss": 0.3474,
+      "step": 3063
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0895238095238097e-05,
+      "loss": 0.3353,
+      "step": 3064
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0892063492063493e-05,
+      "loss": 0.3547,
+      "step": 3065
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.088888888888889e-05,
+      "loss": 0.3351,
+      "step": 3066
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0885714285714286e-05,
+      "loss": 0.3529,
+      "step": 3067
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0882539682539684e-05,
+      "loss": 0.3649,
+      "step": 3068
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.087936507936508e-05,
+      "loss": 0.3558,
+      "step": 3069
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0876190476190478e-05,
+      "loss": 0.3285,
+      "step": 3070
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0873015873015873e-05,
+      "loss": 0.3444,
+      "step": 3071
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.086984126984127e-05,
+      "loss": 0.3649,
+      "step": 3072
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0866666666666667e-05,
+      "loss": 0.3601,
+      "step": 3073
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0863492063492065e-05,
+      "loss": 0.3478,
+      "step": 3074
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0860317460317461e-05,
+      "loss": 0.3538,
+      "step": 3075
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 0.34,
+      "step": 3076
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0853968253968254e-05,
+      "loss": 0.3557,
+      "step": 3077
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0850793650793652e-05,
+      "loss": 0.3501,
+      "step": 3078
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0847619047619048e-05,
+      "loss": 0.3469,
+      "step": 3079
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0844444444444446e-05,
+      "loss": 0.3643,
+      "step": 3080
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0841269841269841e-05,
+      "loss": 0.3152,
+      "step": 3081
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0838095238095239e-05,
+      "loss": 0.3665,
+      "step": 3082
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0834920634920635e-05,
+      "loss": 0.3327,
+      "step": 3083
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0831746031746033e-05,
+      "loss": 0.3598,
+      "step": 3084
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.082857142857143e-05,
+      "loss": 0.3607,
+      "step": 3085
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0825396825396826e-05,
+      "loss": 0.3334,
+      "step": 3086
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0822222222222222e-05,
+      "loss": 0.3353,
+      "step": 3087
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.081904761904762e-05,
+      "loss": 0.3639,
+      "step": 3088
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0815873015873017e-05,
+      "loss": 0.3275,
+      "step": 3089
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0812698412698415e-05,
+      "loss": 0.352,
+      "step": 3090
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.080952380952381e-05,
+      "loss": 0.3521,
+      "step": 3091
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0806349206349207e-05,
+      "loss": 0.3542,
+      "step": 3092
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0803174603174604e-05,
+      "loss": 0.3589,
+      "step": 3093
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 0.3753,
+      "step": 3094
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0796825396825396e-05,
+      "loss": 0.3376,
+      "step": 3095
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0793650793650794e-05,
+      "loss": 0.3649,
+      "step": 3096
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.079047619047619e-05,
+      "loss": 0.3425,
+      "step": 3097
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0787301587301589e-05,
+      "loss": 0.3648,
+      "step": 3098
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0784126984126987e-05,
+      "loss": 0.3481,
+      "step": 3099
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0780952380952381e-05,
+      "loss": 0.339,
+      "step": 3100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0777777777777778e-05,
+      "loss": 0.3483,
+      "step": 3101
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0774603174603176e-05,
+      "loss": 0.3491,
+      "step": 3102
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0771428571428572e-05,
+      "loss": 0.3573,
+      "step": 3103
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.076825396825397e-05,
+      "loss": 0.359,
+      "step": 3104
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0765079365079365e-05,
+      "loss": 0.3743,
+      "step": 3105
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0761904761904763e-05,
+      "loss": 0.3549,
+      "step": 3106
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0758730158730159e-05,
+      "loss": 0.339,
+      "step": 3107
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0755555555555557e-05,
+      "loss": 0.3285,
+      "step": 3108
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0752380952380955e-05,
+      "loss": 0.3582,
+      "step": 3109
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.074920634920635e-05,
+      "loss": 0.3627,
+      "step": 3110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0746031746031746e-05,
+      "loss": 0.3245,
+      "step": 3111
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0742857142857144e-05,
+      "loss": 0.3401,
+      "step": 3112
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0739682539682542e-05,
+      "loss": 0.3374,
+      "step": 3113
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0736507936507938e-05,
+      "loss": 0.3515,
+      "step": 3114
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0733333333333333e-05,
+      "loss": 0.333,
+      "step": 3115
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.073015873015873e-05,
+      "loss": 0.3642,
+      "step": 3116
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0726984126984127e-05,
+      "loss": 0.3682,
+      "step": 3117
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0723809523809525e-05,
+      "loss": 0.3271,
+      "step": 3118
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0720634920634923e-05,
+      "loss": 0.3663,
+      "step": 3119
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0717460317460318e-05,
+      "loss": 0.3401,
+      "step": 3120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.347,
+      "step": 3121
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0711111111111112e-05,
+      "loss": 0.3488,
+      "step": 3122
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.070793650793651e-05,
+      "loss": 0.3427,
+      "step": 3123
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0704761904761906e-05,
+      "loss": 0.3405,
+      "step": 3124
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0701587301587301e-05,
+      "loss": 0.356,
+      "step": 3125
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0698412698412699e-05,
+      "loss": 0.3259,
+      "step": 3126
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0695238095238097e-05,
+      "loss": 0.3381,
+      "step": 3127
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0692063492063493e-05,
+      "loss": 0.3538,
+      "step": 3128
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0688888888888891e-05,
+      "loss": 0.3803,
+      "step": 3129
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0685714285714286e-05,
+      "loss": 0.3421,
+      "step": 3130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0682539682539684e-05,
+      "loss": 0.3307,
+      "step": 3131
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.067936507936508e-05,
+      "loss": 0.359,
+      "step": 3132
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0676190476190478e-05,
+      "loss": 0.3691,
+      "step": 3133
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0673015873015873e-05,
+      "loss": 0.3637,
+      "step": 3134
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.066984126984127e-05,
+      "loss": 0.3235,
+      "step": 3135
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.3637,
+      "step": 3136
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0663492063492065e-05,
+      "loss": 0.3401,
+      "step": 3137
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0660317460317462e-05,
+      "loss": 0.3591,
+      "step": 3138
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0657142857142856e-05,
+      "loss": 0.3499,
+      "step": 3139
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0653968253968254e-05,
+      "loss": 0.3516,
+      "step": 3140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0650793650793652e-05,
+      "loss": 0.33,
+      "step": 3141
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0647619047619049e-05,
+      "loss": 0.3458,
+      "step": 3142
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0644444444444447e-05,
+      "loss": 0.3541,
+      "step": 3143
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0641269841269841e-05,
+      "loss": 0.332,
+      "step": 3144
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.063809523809524e-05,
+      "loss": 0.3415,
+      "step": 3145
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0634920634920636e-05,
+      "loss": 0.3412,
+      "step": 3146
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0631746031746034e-05,
+      "loss": 0.3768,
+      "step": 3147
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.062857142857143e-05,
+      "loss": 0.3416,
+      "step": 3148
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0625396825396825e-05,
+      "loss": 0.328,
+      "step": 3149
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0622222222222223e-05,
+      "loss": 0.3503,
+      "step": 3150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.061904761904762e-05,
+      "loss": 0.3434,
+      "step": 3151
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0615873015873017e-05,
+      "loss": 0.3253,
+      "step": 3152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0612698412698415e-05,
+      "loss": 0.3452,
+      "step": 3153
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.060952380952381e-05,
+      "loss": 0.3498,
+      "step": 3154
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0606349206349208e-05,
+      "loss": 0.3659,
+      "step": 3155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0603174603174604e-05,
+      "loss": 0.348,
+      "step": 3156
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 0.3342,
+      "step": 3157
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0596825396825398e-05,
+      "loss": 0.355,
+      "step": 3158
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0593650793650794e-05,
+      "loss": 0.3606,
+      "step": 3159
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.059047619047619e-05,
+      "loss": 0.3332,
+      "step": 3160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0587301587301589e-05,
+      "loss": 0.3596,
+      "step": 3161
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0584126984126985e-05,
+      "loss": 0.3462,
+      "step": 3162
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0580952380952383e-05,
+      "loss": 0.3548,
+      "step": 3163
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0577777777777778e-05,
+      "loss": 0.3146,
+      "step": 3164
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0574603174603176e-05,
+      "loss": 0.3596,
+      "step": 3165
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0571428571428572e-05,
+      "loss": 0.3528,
+      "step": 3166
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.056825396825397e-05,
+      "loss": 0.366,
+      "step": 3167
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0565079365079366e-05,
+      "loss": 0.3339,
+      "step": 3168
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0561904761904763e-05,
+      "loss": 0.3473,
+      "step": 3169
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0558730158730159e-05,
+      "loss": 0.3414,
+      "step": 3170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0555555555555557e-05,
+      "loss": 0.3456,
+      "step": 3171
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0552380952380953e-05,
+      "loss": 0.3526,
+      "step": 3172
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.054920634920635e-05,
+      "loss": 0.3817,
+      "step": 3173
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0546031746031746e-05,
+      "loss": 0.3206,
+      "step": 3174
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0542857142857144e-05,
+      "loss": 0.3388,
+      "step": 3175
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.053968253968254e-05,
+      "loss": 0.348,
+      "step": 3176
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0536507936507938e-05,
+      "loss": 0.3709,
+      "step": 3177
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0533333333333333e-05,
+      "loss": 0.336,
+      "step": 3178
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0530158730158731e-05,
+      "loss": 0.3234,
+      "step": 3179
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0526984126984127e-05,
+      "loss": 0.3703,
+      "step": 3180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0523809523809525e-05,
+      "loss": 0.3491,
+      "step": 3181
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0520634920634922e-05,
+      "loss": 0.3535,
+      "step": 3182
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0517460317460318e-05,
+      "loss": 0.3522,
+      "step": 3183
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0514285714285714e-05,
+      "loss": 0.3494,
+      "step": 3184
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0511111111111112e-05,
+      "loss": 0.3376,
+      "step": 3185
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0507936507936509e-05,
+      "loss": 0.3591,
+      "step": 3186
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0504761904761907e-05,
+      "loss": 0.3488,
+      "step": 3187
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0501587301587301e-05,
+      "loss": 0.3696,
+      "step": 3188
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.04984126984127e-05,
+      "loss": 0.3712,
+      "step": 3189
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0495238095238096e-05,
+      "loss": 0.339,
+      "step": 3190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0492063492063494e-05,
+      "loss": 0.3411,
+      "step": 3191
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.048888888888889e-05,
+      "loss": 0.3447,
+      "step": 3192
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0485714285714286e-05,
+      "loss": 0.3453,
+      "step": 3193
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0482539682539683e-05,
+      "loss": 0.3756,
+      "step": 3194
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.047936507936508e-05,
+      "loss": 0.3593,
+      "step": 3195
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 0.3248,
+      "step": 3196
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0473015873015875e-05,
+      "loss": 0.3631,
+      "step": 3197
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.046984126984127e-05,
+      "loss": 0.3429,
+      "step": 3198
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0466666666666668e-05,
+      "loss": 0.3366,
+      "step": 3199
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0463492063492064e-05,
+      "loss": 0.3586,
+      "step": 3200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0460317460317462e-05,
+      "loss": 0.3304,
+      "step": 3201
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.045714285714286e-05,
+      "loss": 0.3554,
+      "step": 3202
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0453968253968255e-05,
+      "loss": 0.3449,
+      "step": 3203
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.045079365079365e-05,
+      "loss": 0.3322,
+      "step": 3204
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0447619047619049e-05,
+      "loss": 0.3512,
+      "step": 3205
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0444444444444445e-05,
+      "loss": 0.3267,
+      "step": 3206
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0441269841269843e-05,
+      "loss": 0.3593,
+      "step": 3207
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0438095238095238e-05,
+      "loss": 0.3375,
+      "step": 3208
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0434920634920636e-05,
+      "loss": 0.3434,
+      "step": 3209
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0431746031746032e-05,
+      "loss": 0.3471,
+      "step": 3210
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.042857142857143e-05,
+      "loss": 0.3513,
+      "step": 3211
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0425396825396825e-05,
+      "loss": 0.3729,
+      "step": 3212
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0422222222222223e-05,
+      "loss": 0.3365,
+      "step": 3213
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0419047619047619e-05,
+      "loss": 0.3354,
+      "step": 3214
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0415873015873017e-05,
+      "loss": 0.3339,
+      "step": 3215
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0412698412698415e-05,
+      "loss": 0.3532,
+      "step": 3216
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.040952380952381e-05,
+      "loss": 0.3467,
+      "step": 3217
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0406349206349206e-05,
+      "loss": 0.3655,
+      "step": 3218
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0403174603174604e-05,
+      "loss": 0.3614,
+      "step": 3219
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.04e-05,
+      "loss": 0.3598,
+      "step": 3220
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0396825396825398e-05,
+      "loss": 0.3419,
+      "step": 3221
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0393650793650793e-05,
+      "loss": 0.3474,
+      "step": 3222
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0390476190476191e-05,
+      "loss": 0.3476,
+      "step": 3223
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0387301587301587e-05,
+      "loss": 0.3449,
+      "step": 3224
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0384126984126985e-05,
+      "loss": 0.3299,
+      "step": 3225
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0380952380952383e-05,
+      "loss": 0.3539,
+      "step": 3226
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0377777777777778e-05,
+      "loss": 0.3174,
+      "step": 3227
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0374603174603174e-05,
+      "loss": 0.3323,
+      "step": 3228
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0371428571428572e-05,
+      "loss": 0.3566,
+      "step": 3229
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.036825396825397e-05,
+      "loss": 0.3608,
+      "step": 3230
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0365079365079367e-05,
+      "loss": 0.3322,
+      "step": 3231
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0361904761904761e-05,
+      "loss": 0.3507,
+      "step": 3232
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.035873015873016e-05,
+      "loss": 0.3452,
+      "step": 3233
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0355555555555557e-05,
+      "loss": 0.3683,
+      "step": 3234
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0352380952380954e-05,
+      "loss": 0.3389,
+      "step": 3235
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0349206349206352e-05,
+      "loss": 0.341,
+      "step": 3236
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0346031746031746e-05,
+      "loss": 0.347,
+      "step": 3237
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0342857142857143e-05,
+      "loss": 0.3683,
+      "step": 3238
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.033968253968254e-05,
+      "loss": 0.3343,
+      "step": 3239
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0336507936507939e-05,
+      "loss": 0.3565,
+      "step": 3240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0333333333333335e-05,
+      "loss": 0.3586,
+      "step": 3241
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.033015873015873e-05,
+      "loss": 0.3161,
+      "step": 3242
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0326984126984128e-05,
+      "loss": 0.3243,
+      "step": 3243
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0323809523809526e-05,
+      "loss": 0.3573,
+      "step": 3244
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0320634920634922e-05,
+      "loss": 0.3536,
+      "step": 3245
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.031746031746032e-05,
+      "loss": 0.3537,
+      "step": 3246
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0314285714285715e-05,
+      "loss": 0.3566,
+      "step": 3247
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0311111111111113e-05,
+      "loss": 0.3527,
+      "step": 3248
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0307936507936509e-05,
+      "loss": 0.364,
+      "step": 3249
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0304761904761907e-05,
+      "loss": 0.3347,
+      "step": 3250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0301587301587301e-05,
+      "loss": 0.3402,
+      "step": 3251
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0298412698412698e-05,
+      "loss": 0.3496,
+      "step": 3252
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0295238095238096e-05,
+      "loss": 0.3422,
+      "step": 3253
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0292063492063494e-05,
+      "loss": 0.331,
+      "step": 3254
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.028888888888889e-05,
+      "loss": 0.3473,
+      "step": 3255
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0285714285714285e-05,
+      "loss": 0.3431,
+      "step": 3256
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0282539682539683e-05,
+      "loss": 0.3554,
+      "step": 3257
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.027936507936508e-05,
+      "loss": 0.343,
+      "step": 3258
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0276190476190477e-05,
+      "loss": 0.3562,
+      "step": 3259
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0273015873015875e-05,
+      "loss": 0.3439,
+      "step": 3260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.026984126984127e-05,
+      "loss": 0.3166,
+      "step": 3261
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0266666666666668e-05,
+      "loss": 0.3622,
+      "step": 3262
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0263492063492064e-05,
+      "loss": 0.3831,
+      "step": 3263
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0260317460317462e-05,
+      "loss": 0.3407,
+      "step": 3264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0257142857142858e-05,
+      "loss": 0.3355,
+      "step": 3265
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0253968253968255e-05,
+      "loss": 0.347,
+      "step": 3266
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0250793650793651e-05,
+      "loss": 0.3405,
+      "step": 3267
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0247619047619049e-05,
+      "loss": 0.3445,
+      "step": 3268
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0244444444444445e-05,
+      "loss": 0.3441,
+      "step": 3269
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0241269841269843e-05,
+      "loss": 0.3626,
+      "step": 3270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0238095238095238e-05,
+      "loss": 0.3528,
+      "step": 3271
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0234920634920636e-05,
+      "loss": 0.3329,
+      "step": 3272
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0231746031746032e-05,
+      "loss": 0.353,
+      "step": 3273
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.022857142857143e-05,
+      "loss": 0.3363,
+      "step": 3274
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0225396825396827e-05,
+      "loss": 0.3711,
+      "step": 3275
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0222222222222223e-05,
+      "loss": 0.3386,
+      "step": 3276
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.021904761904762e-05,
+      "loss": 0.3671,
+      "step": 3277
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0215873015873017e-05,
+      "loss": 0.3547,
+      "step": 3278
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0212698412698414e-05,
+      "loss": 0.3464,
+      "step": 3279
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0209523809523812e-05,
+      "loss": 0.3391,
+      "step": 3280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0206349206349206e-05,
+      "loss": 0.3385,
+      "step": 3281
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0203174603174604e-05,
+      "loss": 0.3412,
+      "step": 3282
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.02e-05,
+      "loss": 0.352,
+      "step": 3283
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0196825396825399e-05,
+      "loss": 0.3551,
+      "step": 3284
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0193650793650795e-05,
+      "loss": 0.3518,
+      "step": 3285
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0190476190476191e-05,
+      "loss": 0.3479,
+      "step": 3286
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0187301587301588e-05,
+      "loss": 0.3635,
+      "step": 3287
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0184126984126986e-05,
+      "loss": 0.3727,
+      "step": 3288
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0180952380952382e-05,
+      "loss": 0.3766,
+      "step": 3289
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0177777777777778e-05,
+      "loss": 0.3571,
+      "step": 3290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0174603174603175e-05,
+      "loss": 0.3348,
+      "step": 3291
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0171428571428573e-05,
+      "loss": 0.3644,
+      "step": 3292
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0168253968253969e-05,
+      "loss": 0.3682,
+      "step": 3293
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0165079365079367e-05,
+      "loss": 0.3419,
+      "step": 3294
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0161904761904761e-05,
+      "loss": 0.3475,
+      "step": 3295
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.015873015873016e-05,
+      "loss": 0.3687,
+      "step": 3296
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0155555555555556e-05,
+      "loss": 0.3325,
+      "step": 3297
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0152380952380954e-05,
+      "loss": 0.3475,
+      "step": 3298
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.014920634920635e-05,
+      "loss": 0.3212,
+      "step": 3299
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0146031746031746e-05,
+      "loss": 0.3252,
+      "step": 3300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0142857142857143e-05,
+      "loss": 0.332,
+      "step": 3301
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.013968253968254e-05,
+      "loss": 0.3439,
+      "step": 3302
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0136507936507937e-05,
+      "loss": 0.3511,
+      "step": 3303
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 0.3483,
+      "step": 3304
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013015873015873e-05,
+      "loss": 0.3326,
+      "step": 3305
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0126984126984128e-05,
+      "loss": 0.3748,
+      "step": 3306
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0123809523809524e-05,
+      "loss": 0.3524,
+      "step": 3307
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0120634920634922e-05,
+      "loss": 0.3429,
+      "step": 3308
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0117460317460318e-05,
+      "loss": 0.3588,
+      "step": 3309
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0114285714285715e-05,
+      "loss": 0.3577,
+      "step": 3310
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0111111111111111e-05,
+      "loss": 0.3538,
+      "step": 3311
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0107936507936509e-05,
+      "loss": 0.3496,
+      "step": 3312
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0104761904761905e-05,
+      "loss": 0.3148,
+      "step": 3313
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0101587301587303e-05,
+      "loss": 0.3113,
+      "step": 3314
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0098412698412698e-05,
+      "loss": 0.3361,
+      "step": 3315
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0095238095238096e-05,
+      "loss": 0.3678,
+      "step": 3316
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0092063492063492e-05,
+      "loss": 0.3505,
+      "step": 3317
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.008888888888889e-05,
+      "loss": 0.3512,
+      "step": 3318
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0085714285714288e-05,
+      "loss": 0.3354,
+      "step": 3319
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0082539682539683e-05,
+      "loss": 0.3699,
+      "step": 3320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.007936507936508e-05,
+      "loss": 0.3494,
+      "step": 3321
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0076190476190477e-05,
+      "loss": 0.3572,
+      "step": 3322
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0073015873015874e-05,
+      "loss": 0.3501,
+      "step": 3323
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0069841269841272e-05,
+      "loss": 0.3363,
+      "step": 3324
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0066666666666666e-05,
+      "loss": 0.3618,
+      "step": 3325
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0063492063492064e-05,
+      "loss": 0.3457,
+      "step": 3326
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006031746031746e-05,
+      "loss": 0.3733,
+      "step": 3327
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0057142857142859e-05,
+      "loss": 0.3509,
+      "step": 3328
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053968253968253e-05,
+      "loss": 0.3332,
+      "step": 3329
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050793650793651e-05,
+      "loss": 0.3231,
+      "step": 3330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0047619047619048e-05,
+      "loss": 0.3543,
+      "step": 3331
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044444444444446e-05,
+      "loss": 0.349,
+      "step": 3332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0041269841269844e-05,
+      "loss": 0.3634,
+      "step": 3333
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0038095238095238e-05,
+      "loss": 0.3345,
+      "step": 3334
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0034920634920635e-05,
+      "loss": 0.3501,
+      "step": 3335
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0031746031746033e-05,
+      "loss": 0.3526,
+      "step": 3336
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002857142857143e-05,
+      "loss": 0.3611,
+      "step": 3337
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0025396825396827e-05,
+      "loss": 0.3667,
+      "step": 3338
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0022222222222222e-05,
+      "loss": 0.3781,
+      "step": 3339
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001904761904762e-05,
+      "loss": 0.3504,
+      "step": 3340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0015873015873016e-05,
+      "loss": 0.3662,
+      "step": 3341
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0012698412698414e-05,
+      "loss": 0.343,
+      "step": 3342
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0009523809523812e-05,
+      "loss": 0.336,
+      "step": 3343
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0006349206349206e-05,
+      "loss": 0.352,
+      "step": 3344
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003174603174603e-05,
+      "loss": 0.3436,
+      "step": 3345
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.3551,
+      "step": 3346
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.996825396825399e-06,
+      "loss": 0.3409,
+      "step": 3347
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.993650793650793e-06,
+      "loss": 0.3549,
+      "step": 3348
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.990476190476191e-06,
+      "loss": 0.348,
+      "step": 3349
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987301587301588e-06,
+      "loss": 0.3528,
+      "step": 3350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984126984126986e-06,
+      "loss": 0.361,
+      "step": 3351
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980952380952382e-06,
+      "loss": 0.3552,
+      "step": 3352
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.977777777777778e-06,
+      "loss": 0.3691,
+      "step": 3353
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.974603174603176e-06,
+      "loss": 0.3318,
+      "step": 3354
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.971428571428571e-06,
+      "loss": 0.3308,
+      "step": 3355
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.968253968253969e-06,
+      "loss": 0.3341,
+      "step": 3356
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965079365079365e-06,
+      "loss": 0.3478,
+      "step": 3357
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.961904761904763e-06,
+      "loss": 0.3488,
+      "step": 3358
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95873015873016e-06,
+      "loss": 0.3655,
+      "step": 3359
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.955555555555556e-06,
+      "loss": 0.3564,
+      "step": 3360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.952380952380954e-06,
+      "loss": 0.3554,
+      "step": 3361
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94920634920635e-06,
+      "loss": 0.3643,
+      "step": 3362
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946031746031747e-06,
+      "loss": 0.3508,
+      "step": 3363
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.942857142857145e-06,
+      "loss": 0.3611,
+      "step": 3364
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.939682539682541e-06,
+      "loss": 0.374,
+      "step": 3365
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.936507936507937e-06,
+      "loss": 0.3622,
+      "step": 3366
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933333333333334e-06,
+      "loss": 0.3579,
+      "step": 3367
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.930158730158732e-06,
+      "loss": 0.3497,
+      "step": 3368
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.926984126984128e-06,
+      "loss": 0.3486,
+      "step": 3369
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.923809523809524e-06,
+      "loss": 0.3684,
+      "step": 3370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.920634920634922e-06,
+      "loss": 0.3769,
+      "step": 3371
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.917460317460319e-06,
+      "loss": 0.3317,
+      "step": 3372
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.914285714285715e-06,
+      "loss": 0.3233,
+      "step": 3373
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.911111111111113e-06,
+      "loss": 0.3477,
+      "step": 3374
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.90793650793651e-06,
+      "loss": 0.3321,
+      "step": 3375
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.904761904761906e-06,
+      "loss": 0.3453,
+      "step": 3376
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.901587301587302e-06,
+      "loss": 0.352,
+      "step": 3377
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8984126984127e-06,
+      "loss": 0.3532,
+      "step": 3378
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.895238095238096e-06,
+      "loss": 0.3316,
+      "step": 3379
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.892063492063493e-06,
+      "loss": 0.3677,
+      "step": 3380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88888888888889e-06,
+      "loss": 0.3307,
+      "step": 3381
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.885714285714287e-06,
+      "loss": 0.3327,
+      "step": 3382
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.882539682539683e-06,
+      "loss": 0.334,
+      "step": 3383
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87936507936508e-06,
+      "loss": 0.3445,
+      "step": 3384
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.876190476190478e-06,
+      "loss": 0.3576,
+      "step": 3385
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.873015873015874e-06,
+      "loss": 0.3639,
+      "step": 3386
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86984126984127e-06,
+      "loss": 0.34,
+      "step": 3387
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 0.3533,
+      "step": 3388
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.863492063492065e-06,
+      "loss": 0.3321,
+      "step": 3389
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.86031746031746e-06,
+      "loss": 0.3555,
+      "step": 3390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.857142857142859e-06,
+      "loss": 0.3309,
+      "step": 3391
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.853968253968255e-06,
+      "loss": 0.3749,
+      "step": 3392
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.850793650793651e-06,
+      "loss": 0.3388,
+      "step": 3393
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.847619047619048e-06,
+      "loss": 0.3367,
+      "step": 3394
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.844444444444446e-06,
+      "loss": 0.3479,
+      "step": 3395
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.841269841269842e-06,
+      "loss": 0.332,
+      "step": 3396
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.838095238095238e-06,
+      "loss": 0.3721,
+      "step": 3397
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.834920634920636e-06,
+      "loss": 0.3569,
+      "step": 3398
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.831746031746033e-06,
+      "loss": 0.3285,
+      "step": 3399
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.828571428571429e-06,
+      "loss": 0.3517,
+      "step": 3400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.825396825396825e-06,
+      "loss": 0.3495,
+      "step": 3401
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.822222222222223e-06,
+      "loss": 0.3587,
+      "step": 3402
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.81904761904762e-06,
+      "loss": 0.3334,
+      "step": 3403
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.815873015873016e-06,
+      "loss": 0.3529,
+      "step": 3404
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.812698412698414e-06,
+      "loss": 0.3496,
+      "step": 3405
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.80952380952381e-06,
+      "loss": 0.3408,
+      "step": 3406
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.806349206349207e-06,
+      "loss": 0.32,
+      "step": 3407
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.803174603174605e-06,
+      "loss": 0.3594,
+      "step": 3408
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.3528,
+      "step": 3409
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.796825396825397e-06,
+      "loss": 0.3453,
+      "step": 3410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.793650793650794e-06,
+      "loss": 0.3394,
+      "step": 3411
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.790476190476192e-06,
+      "loss": 0.3372,
+      "step": 3412
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.787301587301588e-06,
+      "loss": 0.3597,
+      "step": 3413
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.784126984126984e-06,
+      "loss": 0.3271,
+      "step": 3414
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.780952380952382e-06,
+      "loss": 0.3485,
+      "step": 3415
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 0.3684,
+      "step": 3416
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.774603174603175e-06,
+      "loss": 0.3565,
+      "step": 3417
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.771428571428571e-06,
+      "loss": 0.3394,
+      "step": 3418
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.76825396825397e-06,
+      "loss": 0.3481,
+      "step": 3419
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.765079365079366e-06,
+      "loss": 0.3322,
+      "step": 3420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.761904761904762e-06,
+      "loss": 0.3421,
+      "step": 3421
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.75873015873016e-06,
+      "loss": 0.3362,
+      "step": 3422
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.755555555555556e-06,
+      "loss": 0.3619,
+      "step": 3423
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.752380952380953e-06,
+      "loss": 0.368,
+      "step": 3424
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.74920634920635e-06,
+      "loss": 0.3474,
+      "step": 3425
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.746031746031747e-06,
+      "loss": 0.3567,
+      "step": 3426
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.742857142857143e-06,
+      "loss": 0.3535,
+      "step": 3427
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.73968253968254e-06,
+      "loss": 0.3288,
+      "step": 3428
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.736507936507938e-06,
+      "loss": 0.3507,
+      "step": 3429
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 0.3178,
+      "step": 3430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.73015873015873e-06,
+      "loss": 0.3473,
+      "step": 3431
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.726984126984128e-06,
+      "loss": 0.3436,
+      "step": 3432
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.723809523809525e-06,
+      "loss": 0.3404,
+      "step": 3433
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.720634920634921e-06,
+      "loss": 0.3464,
+      "step": 3434
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.717460317460317e-06,
+      "loss": 0.3911,
+      "step": 3435
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.3412,
+      "step": 3436
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.711111111111111e-06,
+      "loss": 0.3633,
+      "step": 3437
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.707936507936508e-06,
+      "loss": 0.3329,
+      "step": 3438
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.704761904761906e-06,
+      "loss": 0.3625,
+      "step": 3439
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.701587301587302e-06,
+      "loss": 0.3466,
+      "step": 3440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.698412698412698e-06,
+      "loss": 0.3492,
+      "step": 3441
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.695238095238096e-06,
+      "loss": 0.3443,
+      "step": 3442
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.692063492063493e-06,
+      "loss": 0.3268,
+      "step": 3443
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.688888888888889e-06,
+      "loss": 0.3287,
+      "step": 3444
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.685714285714285e-06,
+      "loss": 0.3554,
+      "step": 3445
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682539682539683e-06,
+      "loss": 0.3374,
+      "step": 3446
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.679365079365081e-06,
+      "loss": 0.3462,
+      "step": 3447
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.676190476190476e-06,
+      "loss": 0.3388,
+      "step": 3448
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.673015873015874e-06,
+      "loss": 0.3509,
+      "step": 3449
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.66984126984127e-06,
+      "loss": 0.3717,
+      "step": 3450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 0.3471,
+      "step": 3451
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.663492063492065e-06,
+      "loss": 0.3142,
+      "step": 3452
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.660317460317461e-06,
+      "loss": 0.358,
+      "step": 3453
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.657142857142859e-06,
+      "loss": 0.3416,
+      "step": 3454
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.653968253968254e-06,
+      "loss": 0.3516,
+      "step": 3455
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.650793650793652e-06,
+      "loss": 0.3311,
+      "step": 3456
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.647619047619048e-06,
+      "loss": 0.3524,
+      "step": 3457
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.644444444444444e-06,
+      "loss": 0.3488,
+      "step": 3458
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.641269841269842e-06,
+      "loss": 0.346,
+      "step": 3459
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.638095238095239e-06,
+      "loss": 0.3403,
+      "step": 3460
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.634920634920637e-06,
+      "loss": 0.3532,
+      "step": 3461
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.631746031746031e-06,
+      "loss": 0.3519,
+      "step": 3462
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.62857142857143e-06,
+      "loss": 0.3318,
+      "step": 3463
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.625396825396827e-06,
+      "loss": 0.3563,
+      "step": 3464
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.622222222222222e-06,
+      "loss": 0.3359,
+      "step": 3465
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.61904761904762e-06,
+      "loss": 0.3436,
+      "step": 3466
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.615873015873016e-06,
+      "loss": 0.342,
+      "step": 3467
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.612698412698414e-06,
+      "loss": 0.3461,
+      "step": 3468
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.60952380952381e-06,
+      "loss": 0.3713,
+      "step": 3469
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.606349206349207e-06,
+      "loss": 0.3898,
+      "step": 3470
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.603174603174605e-06,
+      "loss": 0.3656,
+      "step": 3471
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.3356,
+      "step": 3472
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.596825396825398e-06,
+      "loss": 0.343,
+      "step": 3473
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.593650793650794e-06,
+      "loss": 0.3316,
+      "step": 3474
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.590476190476192e-06,
+      "loss": 0.3345,
+      "step": 3475
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.587301587301588e-06,
+      "loss": 0.3403,
+      "step": 3476
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.584126984126985e-06,
+      "loss": 0.3312,
+      "step": 3477
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.580952380952383e-06,
+      "loss": 0.3327,
+      "step": 3478
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.577777777777779e-06,
+      "loss": 0.3972,
+      "step": 3479
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.574603174603175e-06,
+      "loss": 0.3462,
+      "step": 3480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.571428571428573e-06,
+      "loss": 0.3613,
+      "step": 3481
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.56825396825397e-06,
+      "loss": 0.3651,
+      "step": 3482
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.565079365079366e-06,
+      "loss": 0.3718,
+      "step": 3483
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.561904761904762e-06,
+      "loss": 0.3441,
+      "step": 3484
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.55873015873016e-06,
+      "loss": 0.3218,
+      "step": 3485
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.555555555555556e-06,
+      "loss": 0.3272,
+      "step": 3486
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.552380952380953e-06,
+      "loss": 0.3437,
+      "step": 3487
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.54920634920635e-06,
+      "loss": 0.3369,
+      "step": 3488
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.546031746031747e-06,
+      "loss": 0.3678,
+      "step": 3489
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.542857142857143e-06,
+      "loss": 0.3435,
+      "step": 3490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.539682539682541e-06,
+      "loss": 0.3645,
+      "step": 3491
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.536507936507938e-06,
+      "loss": 0.3432,
+      "step": 3492
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.533333333333334e-06,
+      "loss": 0.3416,
+      "step": 3493
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.53015873015873e-06,
+      "loss": 0.3274,
+      "step": 3494
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.526984126984128e-06,
+      "loss": 0.3225,
+      "step": 3495
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 0.3313,
+      "step": 3496
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.520634920634921e-06,
+      "loss": 0.3403,
+      "step": 3497
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.517460317460319e-06,
+      "loss": 0.3562,
+      "step": 3498
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.514285714285715e-06,
+      "loss": 0.3596,
+      "step": 3499
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.511111111111112e-06,
+      "loss": 0.3362,
+      "step": 3500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.507936507936508e-06,
+      "loss": 0.3766,
+      "step": 3501
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.504761904761906e-06,
+      "loss": 0.342,
+      "step": 3502
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.501587301587302e-06,
+      "loss": 0.3206,
+      "step": 3503
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.498412698412699e-06,
+      "loss": 0.3549,
+      "step": 3504
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.495238095238097e-06,
+      "loss": 0.3484,
+      "step": 3505
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.492063492063493e-06,
+      "loss": 0.3348,
+      "step": 3506
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.48888888888889e-06,
+      "loss": 0.3403,
+      "step": 3507
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.485714285714287e-06,
+      "loss": 0.3516,
+      "step": 3508
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.482539682539684e-06,
+      "loss": 0.3287,
+      "step": 3509
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.47936507936508e-06,
+      "loss": 0.363,
+      "step": 3510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.476190476190476e-06,
+      "loss": 0.3438,
+      "step": 3511
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.473015873015874e-06,
+      "loss": 0.3668,
+      "step": 3512
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.46984126984127e-06,
+      "loss": 0.3281,
+      "step": 3513
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 0.3365,
+      "step": 3514
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.463492063492065e-06,
+      "loss": 0.3263,
+      "step": 3515
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.460317460317461e-06,
+      "loss": 0.3693,
+      "step": 3516
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.457142857142858e-06,
+      "loss": 0.3329,
+      "step": 3517
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.453968253968254e-06,
+      "loss": 0.338,
+      "step": 3518
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.450793650793652e-06,
+      "loss": 0.3506,
+      "step": 3519
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.447619047619048e-06,
+      "loss": 0.3623,
+      "step": 3520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 0.357,
+      "step": 3521
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.441269841269843e-06,
+      "loss": 0.3589,
+      "step": 3522
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.438095238095239e-06,
+      "loss": 0.3374,
+      "step": 3523
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.434920634920635e-06,
+      "loss": 0.3428,
+      "step": 3524
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.431746031746033e-06,
+      "loss": 0.3376,
+      "step": 3525
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 0.3471,
+      "step": 3526
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.425396825396826e-06,
+      "loss": 0.3493,
+      "step": 3527
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.422222222222222e-06,
+      "loss": 0.3451,
+      "step": 3528
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.41904761904762e-06,
+      "loss": 0.3608,
+      "step": 3529
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.415873015873017e-06,
+      "loss": 0.3502,
+      "step": 3530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.412698412698413e-06,
+      "loss": 0.3551,
+      "step": 3531
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.40952380952381e-06,
+      "loss": 0.3297,
+      "step": 3532
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.406349206349207e-06,
+      "loss": 0.3412,
+      "step": 3533
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.403174603174603e-06,
+      "loss": 0.3706,
+      "step": 3534
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.4e-06,
+      "loss": 0.3319,
+      "step": 3535
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.396825396825398e-06,
+      "loss": 0.3465,
+      "step": 3536
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.393650793650794e-06,
+      "loss": 0.3535,
+      "step": 3537
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.39047619047619e-06,
+      "loss": 0.3523,
+      "step": 3538
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.387301587301588e-06,
+      "loss": 0.3219,
+      "step": 3539
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.384126984126985e-06,
+      "loss": 0.3556,
+      "step": 3540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.380952380952381e-06,
+      "loss": 0.3557,
+      "step": 3541
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.377777777777779e-06,
+      "loss": 0.3628,
+      "step": 3542
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.374603174603175e-06,
+      "loss": 0.3432,
+      "step": 3543
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.371428571428572e-06,
+      "loss": 0.3462,
+      "step": 3544
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.368253968253968e-06,
+      "loss": 0.3431,
+      "step": 3545
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.365079365079366e-06,
+      "loss": 0.3515,
+      "step": 3546
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.361904761904762e-06,
+      "loss": 0.3454,
+      "step": 3547
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.358730158730159e-06,
+      "loss": 0.3506,
+      "step": 3548
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.355555555555557e-06,
+      "loss": 0.3759,
+      "step": 3549
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.352380952380953e-06,
+      "loss": 0.3504,
+      "step": 3550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.34920634920635e-06,
+      "loss": 0.3292,
+      "step": 3551
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.346031746031747e-06,
+      "loss": 0.3049,
+      "step": 3552
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.342857142857144e-06,
+      "loss": 0.3683,
+      "step": 3553
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.33968253968254e-06,
+      "loss": 0.3381,
+      "step": 3554
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.336507936507936e-06,
+      "loss": 0.3523,
+      "step": 3555
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.3517,
+      "step": 3556
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.33015873015873e-06,
+      "loss": 0.3777,
+      "step": 3557
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.326984126984127e-06,
+      "loss": 0.3348,
+      "step": 3558
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.323809523809525e-06,
+      "loss": 0.3347,
+      "step": 3559
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.320634920634921e-06,
+      "loss": 0.3298,
+      "step": 3560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.317460317460318e-06,
+      "loss": 0.3319,
+      "step": 3561
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.314285714285714e-06,
+      "loss": 0.3548,
+      "step": 3562
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.311111111111112e-06,
+      "loss": 0.35,
+      "step": 3563
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.30793650793651e-06,
+      "loss": 0.3623,
+      "step": 3564
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.304761904761905e-06,
+      "loss": 0.3355,
+      "step": 3565
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.301587301587303e-06,
+      "loss": 0.3424,
+      "step": 3566
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.298412698412699e-06,
+      "loss": 0.3473,
+      "step": 3567
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.295238095238095e-06,
+      "loss": 0.3523,
+      "step": 3568
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.292063492063493e-06,
+      "loss": 0.3366,
+      "step": 3569
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.28888888888889e-06,
+      "loss": 0.3307,
+      "step": 3570
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 0.3262,
+      "step": 3571
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.282539682539682e-06,
+      "loss": 0.3425,
+      "step": 3572
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.27936507936508e-06,
+      "loss": 0.3336,
+      "step": 3573
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.276190476190477e-06,
+      "loss": 0.3208,
+      "step": 3574
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.273015873015875e-06,
+      "loss": 0.3278,
+      "step": 3575
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.26984126984127e-06,
+      "loss": 0.3524,
+      "step": 3576
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.266666666666667e-06,
+      "loss": 0.3339,
+      "step": 3577
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.263492063492065e-06,
+      "loss": 0.3336,
+      "step": 3578
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.26031746031746e-06,
+      "loss": 0.3327,
+      "step": 3579
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.257142857142858e-06,
+      "loss": 0.3441,
+      "step": 3580
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.253968253968256e-06,
+      "loss": 0.3526,
+      "step": 3581
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.250793650793652e-06,
+      "loss": 0.3401,
+      "step": 3582
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.247619047619048e-06,
+      "loss": 0.3639,
+      "step": 3583
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.244444444444445e-06,
+      "loss": 0.3559,
+      "step": 3584
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.241269841269843e-06,
+      "loss": 0.3595,
+      "step": 3585
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.238095238095239e-06,
+      "loss": 0.3523,
+      "step": 3586
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.234920634920635e-06,
+      "loss": 0.3326,
+      "step": 3587
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.231746031746033e-06,
+      "loss": 0.3467,
+      "step": 3588
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.22857142857143e-06,
+      "loss": 0.3659,
+      "step": 3589
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.225396825396826e-06,
+      "loss": 0.3345,
+      "step": 3590
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.222222222222224e-06,
+      "loss": 0.3713,
+      "step": 3591
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.21904761904762e-06,
+      "loss": 0.3435,
+      "step": 3592
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.215873015873017e-06,
+      "loss": 0.3294,
+      "step": 3593
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.212698412698413e-06,
+      "loss": 0.3577,
+      "step": 3594
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.209523809523811e-06,
+      "loss": 0.3548,
+      "step": 3595
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.206349206349207e-06,
+      "loss": 0.324,
+      "step": 3596
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.203174603174604e-06,
+      "loss": 0.3442,
+      "step": 3597
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.3655,
+      "step": 3598
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.196825396825398e-06,
+      "loss": 0.358,
+      "step": 3599
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.193650793650794e-06,
+      "loss": 0.3417,
+      "step": 3600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.19047619047619e-06,
+      "loss": 0.3454,
+      "step": 3601
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.187301587301589e-06,
+      "loss": 0.345,
+      "step": 3602
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.184126984126985e-06,
+      "loss": 0.3236,
+      "step": 3603
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.180952380952381e-06,
+      "loss": 0.3081,
+      "step": 3604
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.17777777777778e-06,
+      "loss": 0.375,
+      "step": 3605
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.174603174603176e-06,
+      "loss": 0.3347,
+      "step": 3606
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.171428571428572e-06,
+      "loss": 0.3485,
+      "step": 3607
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.16825396825397e-06,
+      "loss": 0.3562,
+      "step": 3608
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.165079365079366e-06,
+      "loss": 0.3312,
+      "step": 3609
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.161904761904763e-06,
+      "loss": 0.3677,
+      "step": 3610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.158730158730159e-06,
+      "loss": 0.3613,
+      "step": 3611
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.155555555555557e-06,
+      "loss": 0.352,
+      "step": 3612
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.152380952380953e-06,
+      "loss": 0.3561,
+      "step": 3613
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.14920634920635e-06,
+      "loss": 0.3652,
+      "step": 3614
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.146031746031748e-06,
+      "loss": 0.3345,
+      "step": 3615
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.3535,
+      "step": 3616
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.13968253968254e-06,
+      "loss": 0.3537,
+      "step": 3617
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.136507936507937e-06,
+      "loss": 0.3393,
+      "step": 3618
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.133333333333335e-06,
+      "loss": 0.339,
+      "step": 3619
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.130158730158731e-06,
+      "loss": 0.3383,
+      "step": 3620
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.126984126984127e-06,
+      "loss": 0.3467,
+      "step": 3621
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.123809523809525e-06,
+      "loss": 0.3355,
+      "step": 3622
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.120634920634922e-06,
+      "loss": 0.3541,
+      "step": 3623
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.117460317460318e-06,
+      "loss": 0.3365,
+      "step": 3624
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.114285714285716e-06,
+      "loss": 0.3573,
+      "step": 3625
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.111111111111112e-06,
+      "loss": 0.3455,
+      "step": 3626
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.107936507936508e-06,
+      "loss": 0.3296,
+      "step": 3627
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.104761904761905e-06,
+      "loss": 0.3666,
+      "step": 3628
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.101587301587303e-06,
+      "loss": 0.3067,
+      "step": 3629
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.098412698412699e-06,
+      "loss": 0.3426,
+      "step": 3630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.095238095238095e-06,
+      "loss": 0.3736,
+      "step": 3631
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.092063492063493e-06,
+      "loss": 0.3275,
+      "step": 3632
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.08888888888889e-06,
+      "loss": 0.3352,
+      "step": 3633
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.085714285714286e-06,
+      "loss": 0.3349,
+      "step": 3634
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.082539682539682e-06,
+      "loss": 0.3309,
+      "step": 3635
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.07936507936508e-06,
+      "loss": 0.3487,
+      "step": 3636
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.076190476190477e-06,
+      "loss": 0.3467,
+      "step": 3637
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.073015873015873e-06,
+      "loss": 0.3516,
+      "step": 3638
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.069841269841271e-06,
+      "loss": 0.3502,
+      "step": 3639
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 0.3467,
+      "step": 3640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.063492063492064e-06,
+      "loss": 0.3362,
+      "step": 3641
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.060317460317462e-06,
+      "loss": 0.3723,
+      "step": 3642
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.057142857142858e-06,
+      "loss": 0.324,
+      "step": 3643
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.053968253968254e-06,
+      "loss": 0.3351,
+      "step": 3644
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.05079365079365e-06,
+      "loss": 0.3407,
+      "step": 3645
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.047619047619049e-06,
+      "loss": 0.3666,
+      "step": 3646
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.044444444444445e-06,
+      "loss": 0.3487,
+      "step": 3647
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.041269841269841e-06,
+      "loss": 0.3694,
+      "step": 3648
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.03809523809524e-06,
+      "loss": 0.3674,
+      "step": 3649
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.034920634920636e-06,
+      "loss": 0.3449,
+      "step": 3650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.031746031746032e-06,
+      "loss": 0.3524,
+      "step": 3651
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.028571428571428e-06,
+      "loss": 0.3389,
+      "step": 3652
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.025396825396826e-06,
+      "loss": 0.3277,
+      "step": 3653
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.022222222222223e-06,
+      "loss": 0.3367,
+      "step": 3654
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.019047619047619e-06,
+      "loss": 0.3676,
+      "step": 3655
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.015873015873017e-06,
+      "loss": 0.3489,
+      "step": 3656
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.012698412698413e-06,
+      "loss": 0.3308,
+      "step": 3657
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.00952380952381e-06,
+      "loss": 0.3646,
+      "step": 3658
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.006349206349208e-06,
+      "loss": 0.3258,
+      "step": 3659
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.003174603174604e-06,
+      "loss": 0.3433,
+      "step": 3660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9e-06,
+      "loss": 0.3278,
+      "step": 3661
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.996825396825397e-06,
+      "loss": 0.3383,
+      "step": 3662
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.993650793650795e-06,
+      "loss": 0.356,
+      "step": 3663
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.990476190476191e-06,
+      "loss": 0.3585,
+      "step": 3664
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.987301587301587e-06,
+      "loss": 0.3397,
+      "step": 3665
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.984126984126985e-06,
+      "loss": 0.3381,
+      "step": 3666
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.980952380952382e-06,
+      "loss": 0.3403,
+      "step": 3667
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.977777777777778e-06,
+      "loss": 0.3407,
+      "step": 3668
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.974603174603176e-06,
+      "loss": 0.3406,
+      "step": 3669
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.971428571428572e-06,
+      "loss": 0.342,
+      "step": 3670
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.968253968253968e-06,
+      "loss": 0.3462,
+      "step": 3671
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.965079365079365e-06,
+      "loss": 0.3545,
+      "step": 3672
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.961904761904763e-06,
+      "loss": 0.3267,
+      "step": 3673
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.958730158730159e-06,
+      "loss": 0.3498,
+      "step": 3674
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.955555555555555e-06,
+      "loss": 0.3437,
+      "step": 3675
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.952380952380953e-06,
+      "loss": 0.3233,
+      "step": 3676
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.94920634920635e-06,
+      "loss": 0.3398,
+      "step": 3677
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.946031746031748e-06,
+      "loss": 0.3719,
+      "step": 3678
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.942857142857142e-06,
+      "loss": 0.3662,
+      "step": 3679
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.93968253968254e-06,
+      "loss": 0.3191,
+      "step": 3680
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.936507936507938e-06,
+      "loss": 0.3559,
+      "step": 3681
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 0.3383,
+      "step": 3682
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.930158730158731e-06,
+      "loss": 0.348,
+      "step": 3683
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.926984126984127e-06,
+      "loss": 0.3687,
+      "step": 3684
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.923809523809525e-06,
+      "loss": 0.3566,
+      "step": 3685
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.920634920634922e-06,
+      "loss": 0.3538,
+      "step": 3686
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.917460317460318e-06,
+      "loss": 0.3519,
+      "step": 3687
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.914285714285716e-06,
+      "loss": 0.3515,
+      "step": 3688
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.91111111111111e-06,
+      "loss": 0.3463,
+      "step": 3689
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.907936507936509e-06,
+      "loss": 0.3561,
+      "step": 3690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.904761904761905e-06,
+      "loss": 0.3663,
+      "step": 3691
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.901587301587303e-06,
+      "loss": 0.3558,
+      "step": 3692
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.8984126984127e-06,
+      "loss": 0.3643,
+      "step": 3693
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.895238095238096e-06,
+      "loss": 0.319,
+      "step": 3694
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.892063492063494e-06,
+      "loss": 0.3119,
+      "step": 3695
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.3431,
+      "step": 3696
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.885714285714286e-06,
+      "loss": 0.3563,
+      "step": 3697
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.882539682539684e-06,
+      "loss": 0.3476,
+      "step": 3698
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.87936507936508e-06,
+      "loss": 0.3443,
+      "step": 3699
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.876190476190477e-06,
+      "loss": 0.3299,
+      "step": 3700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.873015873015873e-06,
+      "loss": 0.3196,
+      "step": 3701
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.869841269841271e-06,
+      "loss": 0.3302,
+      "step": 3702
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.866666666666668e-06,
+      "loss": 0.3411,
+      "step": 3703
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.863492063492064e-06,
+      "loss": 0.3448,
+      "step": 3704
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.860317460317462e-06,
+      "loss": 0.3516,
+      "step": 3705
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.857142857142858e-06,
+      "loss": 0.3526,
+      "step": 3706
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.853968253968255e-06,
+      "loss": 0.3328,
+      "step": 3707
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.850793650793653e-06,
+      "loss": 0.3622,
+      "step": 3708
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.847619047619049e-06,
+      "loss": 0.3253,
+      "step": 3709
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.844444444444445e-06,
+      "loss": 0.3834,
+      "step": 3710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.841269841269842e-06,
+      "loss": 0.3272,
+      "step": 3711
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.83809523809524e-06,
+      "loss": 0.3618,
+      "step": 3712
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.834920634920636e-06,
+      "loss": 0.3388,
+      "step": 3713
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.831746031746032e-06,
+      "loss": 0.344,
+      "step": 3714
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.82857142857143e-06,
+      "loss": 0.3407,
+      "step": 3715
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.825396825396827e-06,
+      "loss": 0.3515,
+      "step": 3716
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.822222222222223e-06,
+      "loss": 0.3268,
+      "step": 3717
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.819047619047619e-06,
+      "loss": 0.3578,
+      "step": 3718
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.815873015873017e-06,
+      "loss": 0.3385,
+      "step": 3719
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.812698412698413e-06,
+      "loss": 0.3677,
+      "step": 3720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.80952380952381e-06,
+      "loss": 0.3256,
+      "step": 3721
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.806349206349208e-06,
+      "loss": 0.3387,
+      "step": 3722
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.803174603174604e-06,
+      "loss": 0.3471,
+      "step": 3723
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.8e-06,
+      "loss": 0.3688,
+      "step": 3724
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.796825396825398e-06,
+      "loss": 0.3749,
+      "step": 3725
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.793650793650795e-06,
+      "loss": 0.3342,
+      "step": 3726
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.790476190476191e-06,
+      "loss": 0.3466,
+      "step": 3727
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.787301587301587e-06,
+      "loss": 0.3351,
+      "step": 3728
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.784126984126985e-06,
+      "loss": 0.3492,
+      "step": 3729
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.780952380952382e-06,
+      "loss": 0.358,
+      "step": 3730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.777777777777778e-06,
+      "loss": 0.3677,
+      "step": 3731
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.774603174603176e-06,
+      "loss": 0.336,
+      "step": 3732
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.771428571428572e-06,
+      "loss": 0.3766,
+      "step": 3733
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.768253968253969e-06,
+      "loss": 0.3546,
+      "step": 3734
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.765079365079365e-06,
+      "loss": 0.3562,
+      "step": 3735
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.761904761904763e-06,
+      "loss": 0.332,
+      "step": 3736
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.75873015873016e-06,
+      "loss": 0.3399,
+      "step": 3737
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.755555555555556e-06,
+      "loss": 0.3406,
+      "step": 3738
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.752380952380954e-06,
+      "loss": 0.3472,
+      "step": 3739
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.74920634920635e-06,
+      "loss": 0.3413,
+      "step": 3740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.746031746031746e-06,
+      "loss": 0.3599,
+      "step": 3741
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.742857142857144e-06,
+      "loss": 0.3314,
+      "step": 3742
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.73968253968254e-06,
+      "loss": 0.3316,
+      "step": 3743
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.736507936507937e-06,
+      "loss": 0.3744,
+      "step": 3744
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.733333333333333e-06,
+      "loss": 0.3341,
+      "step": 3745
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.730158730158731e-06,
+      "loss": 0.3342,
+      "step": 3746
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.726984126984128e-06,
+      "loss": 0.3652,
+      "step": 3747
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.723809523809524e-06,
+      "loss": 0.3699,
+      "step": 3748
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.720634920634922e-06,
+      "loss": 0.3583,
+      "step": 3749
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.717460317460318e-06,
+      "loss": 0.3303,
+      "step": 3750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.714285714285715e-06,
+      "loss": 0.3459,
+      "step": 3751
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.711111111111111e-06,
+      "loss": 0.3312,
+      "step": 3752
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.707936507936509e-06,
+      "loss": 0.3608,
+      "step": 3753
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.704761904761905e-06,
+      "loss": 0.351,
+      "step": 3754
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.701587301587302e-06,
+      "loss": 0.3523,
+      "step": 3755
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.6984126984127e-06,
+      "loss": 0.3367,
+      "step": 3756
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.695238095238096e-06,
+      "loss": 0.3656,
+      "step": 3757
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.692063492063492e-06,
+      "loss": 0.3503,
+      "step": 3758
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.68888888888889e-06,
+      "loss": 0.3626,
+      "step": 3759
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.685714285714287e-06,
+      "loss": 0.3275,
+      "step": 3760
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.682539682539683e-06,
+      "loss": 0.3658,
+      "step": 3761
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.67936507936508e-06,
+      "loss": 0.3648,
+      "step": 3762
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.676190476190477e-06,
+      "loss": 0.3526,
+      "step": 3763
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.673015873015873e-06,
+      "loss": 0.3593,
+      "step": 3764
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.66984126984127e-06,
+      "loss": 0.3417,
+      "step": 3765
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.3469,
+      "step": 3766
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.663492063492064e-06,
+      "loss": 0.3181,
+      "step": 3767
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.66031746031746e-06,
+      "loss": 0.3169,
+      "step": 3768
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.657142857142858e-06,
+      "loss": 0.3417,
+      "step": 3769
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.653968253968255e-06,
+      "loss": 0.3627,
+      "step": 3770
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.650793650793651e-06,
+      "loss": 0.3283,
+      "step": 3771
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.647619047619047e-06,
+      "loss": 0.3389,
+      "step": 3772
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.644444444444445e-06,
+      "loss": 0.3461,
+      "step": 3773
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.641269841269842e-06,
+      "loss": 0.3485,
+      "step": 3774
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.638095238095238e-06,
+      "loss": 0.3503,
+      "step": 3775
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.634920634920636e-06,
+      "loss": 0.3359,
+      "step": 3776
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.631746031746032e-06,
+      "loss": 0.3399,
+      "step": 3777
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.628571428571429e-06,
+      "loss": 0.3191,
+      "step": 3778
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.625396825396825e-06,
+      "loss": 0.3515,
+      "step": 3779
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.622222222222223e-06,
+      "loss": 0.3407,
+      "step": 3780
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.61904761904762e-06,
+      "loss": 0.3384,
+      "step": 3781
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.615873015873016e-06,
+      "loss": 0.3504,
+      "step": 3782
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.612698412698414e-06,
+      "loss": 0.3582,
+      "step": 3783
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.60952380952381e-06,
+      "loss": 0.3494,
+      "step": 3784
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.606349206349206e-06,
+      "loss": 0.339,
+      "step": 3785
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.603174603174604e-06,
+      "loss": 0.3394,
+      "step": 3786
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.6e-06,
+      "loss": 0.3502,
+      "step": 3787
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.596825396825399e-06,
+      "loss": 0.3421,
+      "step": 3788
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.593650793650793e-06,
+      "loss": 0.3504,
+      "step": 3789
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.590476190476191e-06,
+      "loss": 0.3406,
+      "step": 3790
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.587301587301588e-06,
+      "loss": 0.3535,
+      "step": 3791
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.584126984126984e-06,
+      "loss": 0.3611,
+      "step": 3792
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.580952380952382e-06,
+      "loss": 0.3776,
+      "step": 3793
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.577777777777778e-06,
+      "loss": 0.3515,
+      "step": 3794
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.574603174603176e-06,
+      "loss": 0.3313,
+      "step": 3795
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.3323,
+      "step": 3796
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.568253968253969e-06,
+      "loss": 0.3391,
+      "step": 3797
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.565079365079367e-06,
+      "loss": 0.3585,
+      "step": 3798
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.561904761904762e-06,
+      "loss": 0.3568,
+      "step": 3799
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.55873015873016e-06,
+      "loss": 0.3387,
+      "step": 3800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.555555555555556e-06,
+      "loss": 0.3457,
+      "step": 3801
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.552380952380954e-06,
+      "loss": 0.3648,
+      "step": 3802
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.54920634920635e-06,
+      "loss": 0.3574,
+      "step": 3803
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.546031746031747e-06,
+      "loss": 0.3552,
+      "step": 3804
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.542857142857145e-06,
+      "loss": 0.3569,
+      "step": 3805
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.53968253968254e-06,
+      "loss": 0.3218,
+      "step": 3806
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.536507936507937e-06,
+      "loss": 0.3625,
+      "step": 3807
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 0.3327,
+      "step": 3808
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.530158730158732e-06,
+      "loss": 0.3335,
+      "step": 3809
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.526984126984128e-06,
+      "loss": 0.3315,
+      "step": 3810
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.523809523809524e-06,
+      "loss": 0.3375,
+      "step": 3811
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.520634920634922e-06,
+      "loss": 0.3378,
+      "step": 3812
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.517460317460318e-06,
+      "loss": 0.3331,
+      "step": 3813
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.514285714285715e-06,
+      "loss": 0.35,
+      "step": 3814
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.511111111111113e-06,
+      "loss": 0.3703,
+      "step": 3815
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.507936507936509e-06,
+      "loss": 0.3371,
+      "step": 3816
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.504761904761905e-06,
+      "loss": 0.3559,
+      "step": 3817
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.501587301587302e-06,
+      "loss": 0.3394,
+      "step": 3818
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.4984126984127e-06,
+      "loss": 0.365,
+      "step": 3819
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.495238095238096e-06,
+      "loss": 0.349,
+      "step": 3820
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.492063492063492e-06,
+      "loss": 0.3529,
+      "step": 3821
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.48888888888889e-06,
+      "loss": 0.3354,
+      "step": 3822
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.485714285714287e-06,
+      "loss": 0.3772,
+      "step": 3823
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.482539682539683e-06,
+      "loss": 0.3336,
+      "step": 3824
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.479365079365081e-06,
+      "loss": 0.3147,
+      "step": 3825
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.476190476190477e-06,
+      "loss": 0.3417,
+      "step": 3826
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.473015873015874e-06,
+      "loss": 0.3384,
+      "step": 3827
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.46984126984127e-06,
+      "loss": 0.3334,
+      "step": 3828
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.466666666666668e-06,
+      "loss": 0.3258,
+      "step": 3829
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.463492063492064e-06,
+      "loss": 0.3526,
+      "step": 3830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.46031746031746e-06,
+      "loss": 0.3487,
+      "step": 3831
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.457142857142859e-06,
+      "loss": 0.3304,
+      "step": 3832
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.453968253968255e-06,
+      "loss": 0.3688,
+      "step": 3833
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.450793650793651e-06,
+      "loss": 0.3376,
+      "step": 3834
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.447619047619048e-06,
+      "loss": 0.3466,
+      "step": 3835
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 0.3442,
+      "step": 3836
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.441269841269842e-06,
+      "loss": 0.3439,
+      "step": 3837
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.438095238095238e-06,
+      "loss": 0.3691,
+      "step": 3838
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.434920634920636e-06,
+      "loss": 0.3551,
+      "step": 3839
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.431746031746033e-06,
+      "loss": 0.3703,
+      "step": 3840
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.428571428571429e-06,
+      "loss": 0.357,
+      "step": 3841
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.425396825396827e-06,
+      "loss": 0.3665,
+      "step": 3842
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.422222222222223e-06,
+      "loss": 0.3386,
+      "step": 3843
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.41904761904762e-06,
+      "loss": 0.361,
+      "step": 3844
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.415873015873016e-06,
+      "loss": 0.377,
+      "step": 3845
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.412698412698414e-06,
+      "loss": 0.3396,
+      "step": 3846
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.40952380952381e-06,
+      "loss": 0.3426,
+      "step": 3847
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.406349206349207e-06,
+      "loss": 0.3546,
+      "step": 3848
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.403174603174605e-06,
+      "loss": 0.3617,
+      "step": 3849
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.3514,
+      "step": 3850
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.396825396825397e-06,
+      "loss": 0.354,
+      "step": 3851
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.393650793650794e-06,
+      "loss": 0.3367,
+      "step": 3852
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.390476190476192e-06,
+      "loss": 0.3499,
+      "step": 3853
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.387301587301588e-06,
+      "loss": 0.3409,
+      "step": 3854
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.384126984126984e-06,
+      "loss": 0.3364,
+      "step": 3855
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.380952380952382e-06,
+      "loss": 0.3215,
+      "step": 3856
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.377777777777779e-06,
+      "loss": 0.3273,
+      "step": 3857
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.374603174603175e-06,
+      "loss": 0.3431,
+      "step": 3858
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.371428571428573e-06,
+      "loss": 0.3244,
+      "step": 3859
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.368253968253969e-06,
+      "loss": 0.3292,
+      "step": 3860
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.365079365079365e-06,
+      "loss": 0.3236,
+      "step": 3861
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.361904761904762e-06,
+      "loss": 0.3545,
+      "step": 3862
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.35873015873016e-06,
+      "loss": 0.3491,
+      "step": 3863
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.355555555555556e-06,
+      "loss": 0.3606,
+      "step": 3864
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.352380952380952e-06,
+      "loss": 0.3472,
+      "step": 3865
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.34920634920635e-06,
+      "loss": 0.3434,
+      "step": 3866
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.346031746031747e-06,
+      "loss": 0.3319,
+      "step": 3867
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.342857142857143e-06,
+      "loss": 0.3558,
+      "step": 3868
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.33968253968254e-06,
+      "loss": 0.3557,
+      "step": 3869
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.336507936507937e-06,
+      "loss": 0.3409,
+      "step": 3870
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.3391,
+      "step": 3871
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.33015873015873e-06,
+      "loss": 0.3582,
+      "step": 3872
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.326984126984128e-06,
+      "loss": 0.3454,
+      "step": 3873
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.323809523809524e-06,
+      "loss": 0.325,
+      "step": 3874
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.32063492063492e-06,
+      "loss": 0.3353,
+      "step": 3875
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.317460317460319e-06,
+      "loss": 0.3357,
+      "step": 3876
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.314285714285715e-06,
+      "loss": 0.349,
+      "step": 3877
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.311111111111111e-06,
+      "loss": 0.3341,
+      "step": 3878
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.307936507936508e-06,
+      "loss": 0.363,
+      "step": 3879
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.304761904761906e-06,
+      "loss": 0.3616,
+      "step": 3880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.301587301587302e-06,
+      "loss": 0.3483,
+      "step": 3881
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.298412698412698e-06,
+      "loss": 0.3142,
+      "step": 3882
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.295238095238096e-06,
+      "loss": 0.3256,
+      "step": 3883
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.292063492063493e-06,
+      "loss": 0.3489,
+      "step": 3884
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.288888888888889e-06,
+      "loss": 0.3437,
+      "step": 3885
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.285714285714287e-06,
+      "loss": 0.3336,
+      "step": 3886
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.282539682539683e-06,
+      "loss": 0.3344,
+      "step": 3887
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.27936507936508e-06,
+      "loss": 0.3236,
+      "step": 3888
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.276190476190476e-06,
+      "loss": 0.3327,
+      "step": 3889
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.273015873015874e-06,
+      "loss": 0.3415,
+      "step": 3890
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.26984126984127e-06,
+      "loss": 0.3254,
+      "step": 3891
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 0.3403,
+      "step": 3892
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.263492063492065e-06,
+      "loss": 0.3806,
+      "step": 3893
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.260317460317461e-06,
+      "loss": 0.3438,
+      "step": 3894
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.257142857142857e-06,
+      "loss": 0.3364,
+      "step": 3895
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.253968253968254e-06,
+      "loss": 0.328,
+      "step": 3896
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.250793650793652e-06,
+      "loss": 0.3344,
+      "step": 3897
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.24761904761905e-06,
+      "loss": 0.3581,
+      "step": 3898
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.244444444444444e-06,
+      "loss": 0.3458,
+      "step": 3899
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.241269841269842e-06,
+      "loss": 0.3558,
+      "step": 3900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.238095238095239e-06,
+      "loss": 0.3167,
+      "step": 3901
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.234920634920635e-06,
+      "loss": 0.329,
+      "step": 3902
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.231746031746033e-06,
+      "loss": 0.3067,
+      "step": 3903
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.22857142857143e-06,
+      "loss": 0.3407,
+      "step": 3904
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.225396825396827e-06,
+      "loss": 0.3428,
+      "step": 3905
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.222222222222222e-06,
+      "loss": 0.3304,
+      "step": 3906
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.21904761904762e-06,
+      "loss": 0.3616,
+      "step": 3907
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.215873015873016e-06,
+      "loss": 0.3521,
+      "step": 3908
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.212698412698412e-06,
+      "loss": 0.3459,
+      "step": 3909
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.20952380952381e-06,
+      "loss": 0.3503,
+      "step": 3910
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.206349206349207e-06,
+      "loss": 0.3477,
+      "step": 3911
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.203174603174605e-06,
+      "loss": 0.3603,
+      "step": 3912
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.2e-06,
+      "loss": 0.3506,
+      "step": 3913
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.196825396825397e-06,
+      "loss": 0.3456,
+      "step": 3914
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.193650793650795e-06,
+      "loss": 0.3381,
+      "step": 3915
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.190476190476192e-06,
+      "loss": 0.3417,
+      "step": 3916
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.187301587301588e-06,
+      "loss": 0.3633,
+      "step": 3917
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.184126984126984e-06,
+      "loss": 0.328,
+      "step": 3918
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.180952380952382e-06,
+      "loss": 0.3363,
+      "step": 3919
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.177777777777779e-06,
+      "loss": 0.354,
+      "step": 3920
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.174603174603175e-06,
+      "loss": 0.3589,
+      "step": 3921
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.171428571428573e-06,
+      "loss": 0.3375,
+      "step": 3922
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.16825396825397e-06,
+      "loss": 0.3648,
+      "step": 3923
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.165079365079366e-06,
+      "loss": 0.3438,
+      "step": 3924
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.161904761904764e-06,
+      "loss": 0.3282,
+      "step": 3925
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.15873015873016e-06,
+      "loss": 0.3513,
+      "step": 3926
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.155555555555556e-06,
+      "loss": 0.3755,
+      "step": 3927
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.152380952380953e-06,
+      "loss": 0.3383,
+      "step": 3928
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.14920634920635e-06,
+      "loss": 0.3252,
+      "step": 3929
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.146031746031747e-06,
+      "loss": 0.3707,
+      "step": 3930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.142857142857143e-06,
+      "loss": 0.3509,
+      "step": 3931
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.139682539682541e-06,
+      "loss": 0.3451,
+      "step": 3932
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.136507936507938e-06,
+      "loss": 0.3436,
+      "step": 3933
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.133333333333334e-06,
+      "loss": 0.3251,
+      "step": 3934
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.13015873015873e-06,
+      "loss": 0.3635,
+      "step": 3935
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.126984126984128e-06,
+      "loss": 0.3506,
+      "step": 3936
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.123809523809525e-06,
+      "loss": 0.3701,
+      "step": 3937
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.120634920634921e-06,
+      "loss": 0.3331,
+      "step": 3938
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.117460317460319e-06,
+      "loss": 0.3328,
+      "step": 3939
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.114285714285715e-06,
+      "loss": 0.345,
+      "step": 3940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.111111111111112e-06,
+      "loss": 0.3322,
+      "step": 3941
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.10793650793651e-06,
+      "loss": 0.3547,
+      "step": 3942
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.104761904761906e-06,
+      "loss": 0.3512,
+      "step": 3943
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.101587301587302e-06,
+      "loss": 0.3354,
+      "step": 3944
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.098412698412699e-06,
+      "loss": 0.3451,
+      "step": 3945
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.095238095238097e-06,
+      "loss": 0.3454,
+      "step": 3946
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.092063492063493e-06,
+      "loss": 0.345,
+      "step": 3947
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.08888888888889e-06,
+      "loss": 0.3249,
+      "step": 3948
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.085714285714287e-06,
+      "loss": 0.3491,
+      "step": 3949
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.082539682539684e-06,
+      "loss": 0.3388,
+      "step": 3950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.07936507936508e-06,
+      "loss": 0.3674,
+      "step": 3951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.076190476190476e-06,
+      "loss": 0.3596,
+      "step": 3952
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.073015873015874e-06,
+      "loss": 0.3365,
+      "step": 3953
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.06984126984127e-06,
+      "loss": 0.3438,
+      "step": 3954
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.066666666666667e-06,
+      "loss": 0.3441,
+      "step": 3955
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.063492063492065e-06,
+      "loss": 0.3592,
+      "step": 3956
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.060317460317461e-06,
+      "loss": 0.3402,
+      "step": 3957
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.057142857142857e-06,
+      "loss": 0.3405,
+      "step": 3958
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.053968253968255e-06,
+      "loss": 0.3598,
+      "step": 3959
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.050793650793652e-06,
+      "loss": 0.3491,
+      "step": 3960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.047619047619048e-06,
+      "loss": 0.3403,
+      "step": 3961
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.044444444444444e-06,
+      "loss": 0.3348,
+      "step": 3962
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.041269841269842e-06,
+      "loss": 0.3467,
+      "step": 3963
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.038095238095239e-06,
+      "loss": 0.3626,
+      "step": 3964
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.034920634920635e-06,
+      "loss": 0.3353,
+      "step": 3965
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.031746031746033e-06,
+      "loss": 0.3577,
+      "step": 3966
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.02857142857143e-06,
+      "loss": 0.3585,
+      "step": 3967
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.025396825396826e-06,
+      "loss": 0.3392,
+      "step": 3968
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.022222222222222e-06,
+      "loss": 0.3184,
+      "step": 3969
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.01904761904762e-06,
+      "loss": 0.3658,
+      "step": 3970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.015873015873016e-06,
+      "loss": 0.3266,
+      "step": 3971
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.012698412698413e-06,
+      "loss": 0.3545,
+      "step": 3972
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.00952380952381e-06,
+      "loss": 0.3616,
+      "step": 3973
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.006349206349207e-06,
+      "loss": 0.3457,
+      "step": 3974
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.003174603174603e-06,
+      "loss": 0.3234,
+      "step": 3975
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.3392,
+      "step": 3976
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.996825396825398e-06,
+      "loss": 0.3563,
+      "step": 3977
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.993650793650794e-06,
+      "loss": 0.3288,
+      "step": 3978
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.99047619047619e-06,
+      "loss": 0.3548,
+      "step": 3979
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.987301587301588e-06,
+      "loss": 0.3321,
+      "step": 3980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.984126984126985e-06,
+      "loss": 0.3222,
+      "step": 3981
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.980952380952381e-06,
+      "loss": 0.3502,
+      "step": 3982
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.977777777777779e-06,
+      "loss": 0.3417,
+      "step": 3983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.974603174603175e-06,
+      "loss": 0.343,
+      "step": 3984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.971428571428572e-06,
+      "loss": 0.3432,
+      "step": 3985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.968253968253968e-06,
+      "loss": 0.328,
+      "step": 3986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.965079365079366e-06,
+      "loss": 0.3354,
+      "step": 3987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.961904761904762e-06,
+      "loss": 0.3366,
+      "step": 3988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.958730158730159e-06,
+      "loss": 0.3745,
+      "step": 3989
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.955555555555557e-06,
+      "loss": 0.329,
+      "step": 3990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.952380952380953e-06,
+      "loss": 0.3589,
+      "step": 3991
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.94920634920635e-06,
+      "loss": 0.3406,
+      "step": 3992
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.946031746031747e-06,
+      "loss": 0.3429,
+      "step": 3993
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.942857142857144e-06,
+      "loss": 0.3455,
+      "step": 3994
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.93968253968254e-06,
+      "loss": 0.3313,
+      "step": 3995
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.936507936507936e-06,
+      "loss": 0.3517,
+      "step": 3996
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.933333333333334e-06,
+      "loss": 0.3425,
+      "step": 3997
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.93015873015873e-06,
+      "loss": 0.3354,
+      "step": 3998
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.926984126984127e-06,
+      "loss": 0.3337,
+      "step": 3999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.923809523809525e-06,
+      "loss": 0.3414,
+      "step": 4000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.920634920634921e-06,
+      "loss": 0.3654,
+      "step": 4001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.917460317460317e-06,
+      "loss": 0.3304,
+      "step": 4002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.914285714285715e-06,
+      "loss": 0.3409,
+      "step": 4003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.911111111111112e-06,
+      "loss": 0.3576,
+      "step": 4004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.907936507936508e-06,
+      "loss": 0.3502,
+      "step": 4005
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.904761904761904e-06,
+      "loss": 0.3462,
+      "step": 4006
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.901587301587302e-06,
+      "loss": 0.3384,
+      "step": 4007
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.898412698412699e-06,
+      "loss": 0.3509,
+      "step": 4008
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.895238095238095e-06,
+      "loss": 0.3486,
+      "step": 4009
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.892063492063493e-06,
+      "loss": 0.3447,
+      "step": 4010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.88888888888889e-06,
+      "loss": 0.3544,
+      "step": 4011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.885714285714286e-06,
+      "loss": 0.3303,
+      "step": 4012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.882539682539682e-06,
+      "loss": 0.3527,
+      "step": 4013
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.87936507936508e-06,
+      "loss": 0.3279,
+      "step": 4014
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.876190476190478e-06,
+      "loss": 0.3355,
+      "step": 4015
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.873015873015873e-06,
+      "loss": 0.3489,
+      "step": 4016
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.86984126984127e-06,
+      "loss": 0.326,
+      "step": 4017
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 0.3681,
+      "step": 4018
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.863492063492063e-06,
+      "loss": 0.337,
+      "step": 4019
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.860317460317461e-06,
+      "loss": 0.3518,
+      "step": 4020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.3365,
+      "step": 4021
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.853968253968256e-06,
+      "loss": 0.3561,
+      "step": 4022
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.85079365079365e-06,
+      "loss": 0.3288,
+      "step": 4023
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.847619047619048e-06,
+      "loss": 0.3392,
+      "step": 4024
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.844444444444446e-06,
+      "loss": 0.3513,
+      "step": 4025
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.841269841269843e-06,
+      "loss": 0.3535,
+      "step": 4026
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.838095238095239e-06,
+      "loss": 0.3591,
+      "step": 4027
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.834920634920635e-06,
+      "loss": 0.3615,
+      "step": 4028
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.831746031746033e-06,
+      "loss": 0.3604,
+      "step": 4029
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.828571428571428e-06,
+      "loss": 0.3651,
+      "step": 4030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.825396825396826e-06,
+      "loss": 0.3555,
+      "step": 4031
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.822222222222224e-06,
+      "loss": 0.3407,
+      "step": 4032
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.81904761904762e-06,
+      "loss": 0.3516,
+      "step": 4033
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.815873015873017e-06,
+      "loss": 0.3576,
+      "step": 4034
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.812698412698413e-06,
+      "loss": 0.3114,
+      "step": 4035
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.809523809523811e-06,
+      "loss": 0.347,
+      "step": 4036
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.806349206349207e-06,
+      "loss": 0.341,
+      "step": 4037
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.803174603174604e-06,
+      "loss": 0.3688,
+      "step": 4038
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 0.3553,
+      "step": 4039
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.796825396825398e-06,
+      "loss": 0.332,
+      "step": 4040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.793650793650794e-06,
+      "loss": 0.3445,
+      "step": 4041
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.790476190476192e-06,
+      "loss": 0.3321,
+      "step": 4042
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.787301587301589e-06,
+      "loss": 0.3659,
+      "step": 4043
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.784126984126985e-06,
+      "loss": 0.3558,
+      "step": 4044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.780952380952381e-06,
+      "loss": 0.3376,
+      "step": 4045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 0.3349,
+      "step": 4046
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.774603174603175e-06,
+      "loss": 0.325,
+      "step": 4047
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.771428571428572e-06,
+      "loss": 0.3217,
+      "step": 4048
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.76825396825397e-06,
+      "loss": 0.3302,
+      "step": 4049
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.765079365079366e-06,
+      "loss": 0.3358,
+      "step": 4050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.761904761904762e-06,
+      "loss": 0.3858,
+      "step": 4051
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.758730158730159e-06,
+      "loss": 0.3643,
+      "step": 4052
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.755555555555557e-06,
+      "loss": 0.331,
+      "step": 4053
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.752380952380953e-06,
+      "loss": 0.3316,
+      "step": 4054
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.74920634920635e-06,
+      "loss": 0.3563,
+      "step": 4055
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.746031746031747e-06,
+      "loss": 0.3618,
+      "step": 4056
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.742857142857144e-06,
+      "loss": 0.3348,
+      "step": 4057
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.73968253968254e-06,
+      "loss": 0.363,
+      "step": 4058
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.736507936507938e-06,
+      "loss": 0.3466,
+      "step": 4059
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 0.3378,
+      "step": 4060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.73015873015873e-06,
+      "loss": 0.3216,
+      "step": 4061
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.726984126984127e-06,
+      "loss": 0.3294,
+      "step": 4062
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.723809523809525e-06,
+      "loss": 0.3444,
+      "step": 4063
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.720634920634921e-06,
+      "loss": 0.3293,
+      "step": 4064
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.717460317460318e-06,
+      "loss": 0.3223,
+      "step": 4065
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.714285714285716e-06,
+      "loss": 0.329,
+      "step": 4066
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.711111111111112e-06,
+      "loss": 0.3624,
+      "step": 4067
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.707936507936508e-06,
+      "loss": 0.3528,
+      "step": 4068
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.704761904761905e-06,
+      "loss": 0.3403,
+      "step": 4069
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.701587301587303e-06,
+      "loss": 0.3328,
+      "step": 4070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.698412698412699e-06,
+      "loss": 0.3544,
+      "step": 4071
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.695238095238095e-06,
+      "loss": 0.377,
+      "step": 4072
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.692063492063493e-06,
+      "loss": 0.3328,
+      "step": 4073
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.68888888888889e-06,
+      "loss": 0.3379,
+      "step": 4074
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.685714285714286e-06,
+      "loss": 0.3418,
+      "step": 4075
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.682539682539684e-06,
+      "loss": 0.3264,
+      "step": 4076
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.67936507936508e-06,
+      "loss": 0.3332,
+      "step": 4077
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.676190476190477e-06,
+      "loss": 0.3439,
+      "step": 4078
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.673015873015873e-06,
+      "loss": 0.3425,
+      "step": 4079
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.669841269841271e-06,
+      "loss": 0.3373,
+      "step": 4080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 0.3652,
+      "step": 4081
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.663492063492064e-06,
+      "loss": 0.3682,
+      "step": 4082
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.660317460317462e-06,
+      "loss": 0.3471,
+      "step": 4083
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.657142857142858e-06,
+      "loss": 0.3781,
+      "step": 4084
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.653968253968254e-06,
+      "loss": 0.348,
+      "step": 4085
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.65079365079365e-06,
+      "loss": 0.351,
+      "step": 4086
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.647619047619049e-06,
+      "loss": 0.3676,
+      "step": 4087
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.644444444444445e-06,
+      "loss": 0.3604,
+      "step": 4088
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.641269841269841e-06,
+      "loss": 0.349,
+      "step": 4089
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.63809523809524e-06,
+      "loss": 0.3192,
+      "step": 4090
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.634920634920635e-06,
+      "loss": 0.3649,
+      "step": 4091
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.631746031746032e-06,
+      "loss": 0.3332,
+      "step": 4092
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.62857142857143e-06,
+      "loss": 0.3379,
+      "step": 4093
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.625396825396826e-06,
+      "loss": 0.3446,
+      "step": 4094
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.622222222222223e-06,
+      "loss": 0.3628,
+      "step": 4095
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 0.3317,
+      "step": 4096
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.615873015873017e-06,
+      "loss": 0.3293,
+      "step": 4097
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.612698412698414e-06,
+      "loss": 0.3557,
+      "step": 4098
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.60952380952381e-06,
+      "loss": 0.3439,
+      "step": 4099
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.6063492063492074e-06,
+      "loss": 0.3514,
+      "step": 4100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.603174603174604e-06,
+      "loss": 0.3161,
+      "step": 4101
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.3512,
+      "step": 4102
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.596825396825398e-06,
+      "loss": 0.329,
+      "step": 4103
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.593650793650794e-06,
+      "loss": 0.3423,
+      "step": 4104
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.5904761904761916e-06,
+      "loss": 0.3479,
+      "step": 4105
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.587301587301588e-06,
+      "loss": 0.3195,
+      "step": 4106
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.584126984126985e-06,
+      "loss": 0.3616,
+      "step": 4107
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.580952380952381e-06,
+      "loss": 0.3188,
+      "step": 4108
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.5777777777777785e-06,
+      "loss": 0.3634,
+      "step": 4109
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.574603174603176e-06,
+      "loss": 0.3334,
+      "step": 4110
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.571428571428572e-06,
+      "loss": 0.3352,
+      "step": 4111
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.568253968253969e-06,
+      "loss": 0.3498,
+      "step": 4112
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.5650793650793655e-06,
+      "loss": 0.3445,
+      "step": 4113
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.561904761904763e-06,
+      "loss": 0.3416,
+      "step": 4114
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.55873015873016e-06,
+      "loss": 0.3427,
+      "step": 4115
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 0.3453,
+      "step": 4116
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.552380952380953e-06,
+      "loss": 0.3399,
+      "step": 4117
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.54920634920635e-06,
+      "loss": 0.3583,
+      "step": 4118
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.546031746031747e-06,
+      "loss": 0.3309,
+      "step": 4119
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.542857142857144e-06,
+      "loss": 0.3142,
+      "step": 4120
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.53968253968254e-06,
+      "loss": 0.3615,
+      "step": 4121
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.5365079365079375e-06,
+      "loss": 0.3862,
+      "step": 4122
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.533333333333334e-06,
+      "loss": 0.3397,
+      "step": 4123
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.530158730158731e-06,
+      "loss": 0.3642,
+      "step": 4124
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.526984126984127e-06,
+      "loss": 0.3384,
+      "step": 4125
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.523809523809524e-06,
+      "loss": 0.3489,
+      "step": 4126
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.520634920634922e-06,
+      "loss": 0.3869,
+      "step": 4127
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.517460317460318e-06,
+      "loss": 0.3372,
+      "step": 4128
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.514285714285715e-06,
+      "loss": 0.3205,
+      "step": 4129
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.511111111111111e-06,
+      "loss": 0.3568,
+      "step": 4130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.5079365079365085e-06,
+      "loss": 0.3718,
+      "step": 4131
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.504761904761906e-06,
+      "loss": 0.3173,
+      "step": 4132
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.501587301587302e-06,
+      "loss": 0.3483,
+      "step": 4133
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.498412698412699e-06,
+      "loss": 0.3362,
+      "step": 4134
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4952380952380955e-06,
+      "loss": 0.3267,
+      "step": 4135
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.492063492063493e-06,
+      "loss": 0.3627,
+      "step": 4136
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.48888888888889e-06,
+      "loss": 0.3249,
+      "step": 4137
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.485714285714286e-06,
+      "loss": 0.3715,
+      "step": 4138
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.482539682539683e-06,
+      "loss": 0.327,
+      "step": 4139
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.47936507936508e-06,
+      "loss": 0.3507,
+      "step": 4140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.476190476190477e-06,
+      "loss": 0.3202,
+      "step": 4141
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.473015873015874e-06,
+      "loss": 0.3567,
+      "step": 4142
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.46984126984127e-06,
+      "loss": 0.3295,
+      "step": 4143
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 0.3261,
+      "step": 4144
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.463492063492064e-06,
+      "loss": 0.335,
+      "step": 4145
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.460317460317461e-06,
+      "loss": 0.3436,
+      "step": 4146
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.457142857142857e-06,
+      "loss": 0.3607,
+      "step": 4147
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4539682539682544e-06,
+      "loss": 0.3277,
+      "step": 4148
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.450793650793652e-06,
+      "loss": 0.3419,
+      "step": 4149
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.447619047619048e-06,
+      "loss": 0.3254,
+      "step": 4150
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.444444444444445e-06,
+      "loss": 0.3801,
+      "step": 4151
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.441269841269841e-06,
+      "loss": 0.3405,
+      "step": 4152
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4380952380952386e-06,
+      "loss": 0.3611,
+      "step": 4153
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.434920634920636e-06,
+      "loss": 0.3648,
+      "step": 4154
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.431746031746032e-06,
+      "loss": 0.3344,
+      "step": 4155
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 0.3507,
+      "step": 4156
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4253968253968255e-06,
+      "loss": 0.3584,
+      "step": 4157
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.422222222222223e-06,
+      "loss": 0.3496,
+      "step": 4158
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.41904761904762e-06,
+      "loss": 0.3261,
+      "step": 4159
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.415873015873016e-06,
+      "loss": 0.365,
+      "step": 4160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.412698412698413e-06,
+      "loss": 0.3419,
+      "step": 4161
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.40952380952381e-06,
+      "loss": 0.3426,
+      "step": 4162
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.406349206349207e-06,
+      "loss": 0.3357,
+      "step": 4163
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.403174603174603e-06,
+      "loss": 0.3536,
+      "step": 4164
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4e-06,
+      "loss": 0.3459,
+      "step": 4165
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.3968253968253975e-06,
+      "loss": 0.3372,
+      "step": 4166
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.393650793650794e-06,
+      "loss": 0.3608,
+      "step": 4167
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.390476190476191e-06,
+      "loss": 0.3548,
+      "step": 4168
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.387301587301587e-06,
+      "loss": 0.3339,
+      "step": 4169
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3841269841269844e-06,
+      "loss": 0.3533,
+      "step": 4170
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.380952380952382e-06,
+      "loss": 0.3504,
+      "step": 4171
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.377777777777778e-06,
+      "loss": 0.3445,
+      "step": 4172
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.374603174603175e-06,
+      "loss": 0.3524,
+      "step": 4173
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.371428571428571e-06,
+      "loss": 0.3353,
+      "step": 4174
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3682539682539686e-06,
+      "loss": 0.3382,
+      "step": 4175
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3650793650793666e-06,
+      "loss": 0.3369,
+      "step": 4176
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.361904761904762e-06,
+      "loss": 0.3369,
+      "step": 4177
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.358730158730159e-06,
+      "loss": 0.314,
+      "step": 4178
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3555555555555555e-06,
+      "loss": 0.3301,
+      "step": 4179
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.352380952380953e-06,
+      "loss": 0.3518,
+      "step": 4180
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.349206349206351e-06,
+      "loss": 0.3381,
+      "step": 4181
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.346031746031746e-06,
+      "loss": 0.3585,
+      "step": 4182
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.342857142857144e-06,
+      "loss": 0.3098,
+      "step": 4183
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.33968253968254e-06,
+      "loss": 0.3483,
+      "step": 4184
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.336507936507937e-06,
+      "loss": 0.3506,
+      "step": 4185
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.3204,
+      "step": 4186
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.33015873015873e-06,
+      "loss": 0.3226,
+      "step": 4187
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.326984126984128e-06,
+      "loss": 0.3586,
+      "step": 4188
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.323809523809524e-06,
+      "loss": 0.3466,
+      "step": 4189
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.320634920634922e-06,
+      "loss": 0.3377,
+      "step": 4190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.317460317460317e-06,
+      "loss": 0.3311,
+      "step": 4191
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.314285714285715e-06,
+      "loss": 0.3356,
+      "step": 4192
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.3111111111111125e-06,
+      "loss": 0.3593,
+      "step": 4193
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.307936507936508e-06,
+      "loss": 0.3727,
+      "step": 4194
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.304761904761906e-06,
+      "loss": 0.2914,
+      "step": 4195
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.301587301587301e-06,
+      "loss": 0.3383,
+      "step": 4196
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.298412698412699e-06,
+      "loss": 0.3484,
+      "step": 4197
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.295238095238097e-06,
+      "loss": 0.3337,
+      "step": 4198
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.292063492063493e-06,
+      "loss": 0.3283,
+      "step": 4199
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.28888888888889e-06,
+      "loss": 0.3188,
+      "step": 4200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.285714285714286e-06,
+      "loss": 0.339,
+      "step": 4201
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.2825396825396835e-06,
+      "loss": 0.3605,
+      "step": 4202
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.279365079365079e-06,
+      "loss": 0.3479,
+      "step": 4203
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.276190476190477e-06,
+      "loss": 0.3131,
+      "step": 4204
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.273015873015874e-06,
+      "loss": 0.3335,
+      "step": 4205
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.2698412698412705e-06,
+      "loss": 0.3326,
+      "step": 4206
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.266666666666668e-06,
+      "loss": 0.3476,
+      "step": 4207
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.263492063492064e-06,
+      "loss": 0.3407,
+      "step": 4208
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.260317460317461e-06,
+      "loss": 0.3194,
+      "step": 4209
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.257142857142858e-06,
+      "loss": 0.3574,
+      "step": 4210
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.253968253968255e-06,
+      "loss": 0.3312,
+      "step": 4211
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.250793650793652e-06,
+      "loss": 0.3312,
+      "step": 4212
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.247619047619048e-06,
+      "loss": 0.3325,
+      "step": 4213
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.244444444444445e-06,
+      "loss": 0.3554,
+      "step": 4214
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.2412698412698425e-06,
+      "loss": 0.3385,
+      "step": 4215
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.238095238095239e-06,
+      "loss": 0.3182,
+      "step": 4216
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.234920634920636e-06,
+      "loss": 0.3618,
+      "step": 4217
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.231746031746032e-06,
+      "loss": 0.3534,
+      "step": 4218
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.2285714285714294e-06,
+      "loss": 0.327,
+      "step": 4219
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.225396825396827e-06,
+      "loss": 0.3496,
+      "step": 4220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 0.3389,
+      "step": 4221
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.21904761904762e-06,
+      "loss": 0.351,
+      "step": 4222
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.215873015873016e-06,
+      "loss": 0.3532,
+      "step": 4223
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.2126984126984136e-06,
+      "loss": 0.3284,
+      "step": 4224
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.20952380952381e-06,
+      "loss": 0.3536,
+      "step": 4225
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.206349206349207e-06,
+      "loss": 0.3576,
+      "step": 4226
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.203174603174604e-06,
+      "loss": 0.3236,
+      "step": 4227
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.3593,
+      "step": 4228
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.196825396825398e-06,
+      "loss": 0.3433,
+      "step": 4229
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.193650793650794e-06,
+      "loss": 0.3544,
+      "step": 4230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.190476190476191e-06,
+      "loss": 0.3433,
+      "step": 4231
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.187301587301588e-06,
+      "loss": 0.3246,
+      "step": 4232
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.184126984126985e-06,
+      "loss": 0.3415,
+      "step": 4233
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.180952380952382e-06,
+      "loss": 0.3632,
+      "step": 4234
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.177777777777778e-06,
+      "loss": 0.3351,
+      "step": 4235
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.174603174603175e-06,
+      "loss": 0.3525,
+      "step": 4236
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1714285714285725e-06,
+      "loss": 0.3593,
+      "step": 4237
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.168253968253969e-06,
+      "loss": 0.3562,
+      "step": 4238
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.165079365079366e-06,
+      "loss": 0.3388,
+      "step": 4239
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.161904761904762e-06,
+      "loss": 0.3383,
+      "step": 4240
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1587301587301594e-06,
+      "loss": 0.3545,
+      "step": 4241
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.155555555555556e-06,
+      "loss": 0.3525,
+      "step": 4242
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.152380952380953e-06,
+      "loss": 0.3073,
+      "step": 4243
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.14920634920635e-06,
+      "loss": 0.3656,
+      "step": 4244
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.146031746031746e-06,
+      "loss": 0.3425,
+      "step": 4245
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.3978,
+      "step": 4246
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.13968253968254e-06,
+      "loss": 0.3361,
+      "step": 4247
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.136507936507937e-06,
+      "loss": 0.3409,
+      "step": 4248
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.133333333333334e-06,
+      "loss": 0.3281,
+      "step": 4249
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1301587301587305e-06,
+      "loss": 0.324,
+      "step": 4250
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.126984126984128e-06,
+      "loss": 0.3539,
+      "step": 4251
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.123809523809524e-06,
+      "loss": 0.3375,
+      "step": 4252
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.120634920634921e-06,
+      "loss": 0.3266,
+      "step": 4253
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.117460317460318e-06,
+      "loss": 0.3461,
+      "step": 4254
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.114285714285715e-06,
+      "loss": 0.3269,
+      "step": 4255
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 0.3521,
+      "step": 4256
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.107936507936508e-06,
+      "loss": 0.3592,
+      "step": 4257
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.104761904761905e-06,
+      "loss": 0.3437,
+      "step": 4258
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.1015873015873025e-06,
+      "loss": 0.341,
+      "step": 4259
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.098412698412699e-06,
+      "loss": 0.3168,
+      "step": 4260
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.095238095238096e-06,
+      "loss": 0.3404,
+      "step": 4261
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.092063492063492e-06,
+      "loss": 0.3344,
+      "step": 4262
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0888888888888894e-06,
+      "loss": 0.345,
+      "step": 4263
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.085714285714286e-06,
+      "loss": 0.3625,
+      "step": 4264
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.082539682539683e-06,
+      "loss": 0.3416,
+      "step": 4265
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.07936507936508e-06,
+      "loss": 0.3538,
+      "step": 4266
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.076190476190476e-06,
+      "loss": 0.3306,
+      "step": 4267
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0730158730158736e-06,
+      "loss": 0.3348,
+      "step": 4268
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.06984126984127e-06,
+      "loss": 0.342,
+      "step": 4269
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.066666666666667e-06,
+      "loss": 0.3789,
+      "step": 4270
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.063492063492064e-06,
+      "loss": 0.3467,
+      "step": 4271
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0603174603174605e-06,
+      "loss": 0.311,
+      "step": 4272
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.057142857142858e-06,
+      "loss": 0.3326,
+      "step": 4273
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.053968253968254e-06,
+      "loss": 0.3534,
+      "step": 4274
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.050793650793651e-06,
+      "loss": 0.3371,
+      "step": 4275
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.047619047619048e-06,
+      "loss": 0.3463,
+      "step": 4276
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.044444444444445e-06,
+      "loss": 0.3313,
+      "step": 4277
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.041269841269842e-06,
+      "loss": 0.3327,
+      "step": 4278
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.038095238095238e-06,
+      "loss": 0.3334,
+      "step": 4279
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.034920634920635e-06,
+      "loss": 0.3379,
+      "step": 4280
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.0317460317460325e-06,
+      "loss": 0.3449,
+      "step": 4281
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.028571428571429e-06,
+      "loss": 0.3235,
+      "step": 4282
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.025396825396826e-06,
+      "loss": 0.3317,
+      "step": 4283
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.022222222222222e-06,
+      "loss": 0.3435,
+      "step": 4284
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.0190476190476194e-06,
+      "loss": 0.3419,
+      "step": 4285
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.015873015873016e-06,
+      "loss": 0.3449,
+      "step": 4286
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.012698412698413e-06,
+      "loss": 0.2982,
+      "step": 4287
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.00952380952381e-06,
+      "loss": 0.3454,
+      "step": 4288
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.006349206349206e-06,
+      "loss": 0.3475,
+      "step": 4289
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.003174603174604e-06,
+      "loss": 0.3219,
+      "step": 4290
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7e-06,
+      "loss": 0.3377,
+      "step": 4291
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.996825396825397e-06,
+      "loss": 0.36,
+      "step": 4292
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.993650793650795e-06,
+      "loss": 0.3273,
+      "step": 4293
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.9904761904761905e-06,
+      "loss": 0.3352,
+      "step": 4294
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.9873015873015886e-06,
+      "loss": 0.3615,
+      "step": 4295
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.984126984126984e-06,
+      "loss": 0.3424,
+      "step": 4296
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.980952380952381e-06,
+      "loss": 0.3516,
+      "step": 4297
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.977777777777779e-06,
+      "loss": 0.339,
+      "step": 4298
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.974603174603175e-06,
+      "loss": 0.3422,
+      "step": 4299
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.971428571428573e-06,
+      "loss": 0.3365,
+      "step": 4300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.968253968253968e-06,
+      "loss": 0.3413,
+      "step": 4301
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.965079365079366e-06,
+      "loss": 0.3502,
+      "step": 4302
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.961904761904762e-06,
+      "loss": 0.3297,
+      "step": 4303
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.958730158730159e-06,
+      "loss": 0.3258,
+      "step": 4304
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.955555555555557e-06,
+      "loss": 0.3128,
+      "step": 4305
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.952380952380952e-06,
+      "loss": 0.3224,
+      "step": 4306
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.94920634920635e-06,
+      "loss": 0.3208,
+      "step": 4307
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.946031746031746e-06,
+      "loss": 0.3305,
+      "step": 4308
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.942857142857144e-06,
+      "loss": 0.3471,
+      "step": 4309
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.939682539682541e-06,
+      "loss": 0.3795,
+      "step": 4310
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.936507936507937e-06,
+      "loss": 0.3459,
+      "step": 4311
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 0.3495,
+      "step": 4312
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.93015873015873e-06,
+      "loss": 0.3387,
+      "step": 4313
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.926984126984128e-06,
+      "loss": 0.3224,
+      "step": 4314
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.923809523809525e-06,
+      "loss": 0.3523,
+      "step": 4315
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.920634920634921e-06,
+      "loss": 0.3449,
+      "step": 4316
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9174603174603186e-06,
+      "loss": 0.3354,
+      "step": 4317
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.914285714285715e-06,
+      "loss": 0.336,
+      "step": 4318
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.911111111111112e-06,
+      "loss": 0.3597,
+      "step": 4319
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.907936507936509e-06,
+      "loss": 0.3286,
+      "step": 4320
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.9047619047619055e-06,
+      "loss": 0.3428,
+      "step": 4321
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.901587301587303e-06,
+      "loss": 0.3419,
+      "step": 4322
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.898412698412699e-06,
+      "loss": 0.3497,
+      "step": 4323
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.895238095238096e-06,
+      "loss": 0.3764,
+      "step": 4324
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8920634920634925e-06,
+      "loss": 0.3362,
+      "step": 4325
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.88888888888889e-06,
+      "loss": 0.3388,
+      "step": 4326
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.885714285714287e-06,
+      "loss": 0.3663,
+      "step": 4327
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.882539682539683e-06,
+      "loss": 0.3487,
+      "step": 4328
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.87936507936508e-06,
+      "loss": 0.3586,
+      "step": 4329
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.876190476190477e-06,
+      "loss": 0.3572,
+      "step": 4330
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.873015873015874e-06,
+      "loss": 0.346,
+      "step": 4331
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.869841269841271e-06,
+      "loss": 0.1869,
+      "step": 4332
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.866666666666667e-06,
+      "loss": 0.1878,
+      "step": 4333
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8634920634920644e-06,
+      "loss": 0.1768,
+      "step": 4334
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.860317460317461e-06,
+      "loss": 0.1596,
+      "step": 4335
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 0.1686,
+      "step": 4336
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.853968253968255e-06,
+      "loss": 0.1409,
+      "step": 4337
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.850793650793651e-06,
+      "loss": 0.1561,
+      "step": 4338
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8476190476190486e-06,
+      "loss": 0.1454,
+      "step": 4339
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.844444444444445e-06,
+      "loss": 0.159,
+      "step": 4340
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.841269841269842e-06,
+      "loss": 0.1473,
+      "step": 4341
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.838095238095238e-06,
+      "loss": 0.1584,
+      "step": 4342
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.8349206349206355e-06,
+      "loss": 0.1446,
+      "step": 4343
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.831746031746033e-06,
+      "loss": 0.1466,
+      "step": 4344
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.828571428571429e-06,
+      "loss": 0.1526,
+      "step": 4345
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.825396825396826e-06,
+      "loss": 0.1524,
+      "step": 4346
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.8222222222222225e-06,
+      "loss": 0.149,
+      "step": 4347
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.81904761904762e-06,
+      "loss": 0.1411,
+      "step": 4348
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.815873015873017e-06,
+      "loss": 0.1403,
+      "step": 4349
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.812698412698413e-06,
+      "loss": 0.1454,
+      "step": 4350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.80952380952381e-06,
+      "loss": 0.1515,
+      "step": 4351
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.806349206349207e-06,
+      "loss": 0.1439,
+      "step": 4352
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.803174603174604e-06,
+      "loss": 0.1413,
+      "step": 4353
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.1427,
+      "step": 4354
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.796825396825397e-06,
+      "loss": 0.1466,
+      "step": 4355
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.7936507936507944e-06,
+      "loss": 0.1361,
+      "step": 4356
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.790476190476191e-06,
+      "loss": 0.1496,
+      "step": 4357
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.787301587301588e-06,
+      "loss": 0.1493,
+      "step": 4358
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.784126984126985e-06,
+      "loss": 0.1482,
+      "step": 4359
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.780952380952381e-06,
+      "loss": 0.15,
+      "step": 4360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.777777777777779e-06,
+      "loss": 0.1441,
+      "step": 4361
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.774603174603175e-06,
+      "loss": 0.1454,
+      "step": 4362
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.771428571428572e-06,
+      "loss": 0.1418,
+      "step": 4363
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.768253968253968e-06,
+      "loss": 0.1397,
+      "step": 4364
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7650793650793655e-06,
+      "loss": 0.1394,
+      "step": 4365
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.761904761904763e-06,
+      "loss": 0.1372,
+      "step": 4366
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.758730158730159e-06,
+      "loss": 0.1508,
+      "step": 4367
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.755555555555556e-06,
+      "loss": 0.1467,
+      "step": 4368
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7523809523809525e-06,
+      "loss": 0.1514,
+      "step": 4369
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.74920634920635e-06,
+      "loss": 0.1409,
+      "step": 4370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.746031746031747e-06,
+      "loss": 0.1465,
+      "step": 4371
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.742857142857143e-06,
+      "loss": 0.1414,
+      "step": 4372
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.73968253968254e-06,
+      "loss": 0.1456,
+      "step": 4373
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.736507936507937e-06,
+      "loss": 0.1421,
+      "step": 4374
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.733333333333334e-06,
+      "loss": 0.1498,
+      "step": 4375
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.730158730158731e-06,
+      "loss": 0.1403,
+      "step": 4376
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.726984126984127e-06,
+      "loss": 0.1422,
+      "step": 4377
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7238095238095245e-06,
+      "loss": 0.1407,
+      "step": 4378
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.720634920634921e-06,
+      "loss": 0.155,
+      "step": 4379
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.717460317460318e-06,
+      "loss": 0.1405,
+      "step": 4380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.714285714285714e-06,
+      "loss": 0.1509,
+      "step": 4381
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.711111111111111e-06,
+      "loss": 0.1537,
+      "step": 4382
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.707936507936509e-06,
+      "loss": 0.14,
+      "step": 4383
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.704761904761905e-06,
+      "loss": 0.1366,
+      "step": 4384
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.701587301587302e-06,
+      "loss": 0.1437,
+      "step": 4385
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.698412698412698e-06,
+      "loss": 0.1499,
+      "step": 4386
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6952380952380956e-06,
+      "loss": 0.1398,
+      "step": 4387
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.692063492063493e-06,
+      "loss": 0.1382,
+      "step": 4388
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.688888888888889e-06,
+      "loss": 0.1553,
+      "step": 4389
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.685714285714286e-06,
+      "loss": 0.1476,
+      "step": 4390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6825396825396825e-06,
+      "loss": 0.1447,
+      "step": 4391
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.67936507936508e-06,
+      "loss": 0.1417,
+      "step": 4392
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.676190476190477e-06,
+      "loss": 0.134,
+      "step": 4393
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.673015873015873e-06,
+      "loss": 0.1513,
+      "step": 4394
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.66984126984127e-06,
+      "loss": 0.1328,
+      "step": 4395
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.1444,
+      "step": 4396
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.663492063492064e-06,
+      "loss": 0.148,
+      "step": 4397
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.660317460317462e-06,
+      "loss": 0.1504,
+      "step": 4398
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.657142857142857e-06,
+      "loss": 0.1447,
+      "step": 4399
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6539682539682545e-06,
+      "loss": 0.1472,
+      "step": 4400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.650793650793651e-06,
+      "loss": 0.1462,
+      "step": 4401
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.647619047619048e-06,
+      "loss": 0.149,
+      "step": 4402
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.644444444444444e-06,
+      "loss": 0.14,
+      "step": 4403
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6412698412698414e-06,
+      "loss": 0.1272,
+      "step": 4404
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6380952380952394e-06,
+      "loss": 0.1435,
+      "step": 4405
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.634920634920635e-06,
+      "loss": 0.1513,
+      "step": 4406
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.631746031746032e-06,
+      "loss": 0.1427,
+      "step": 4407
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.628571428571428e-06,
+      "loss": 0.1421,
+      "step": 4408
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6253968253968256e-06,
+      "loss": 0.1411,
+      "step": 4409
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6222222222222236e-06,
+      "loss": 0.1473,
+      "step": 4410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.619047619047619e-06,
+      "loss": 0.1444,
+      "step": 4411
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.615873015873017e-06,
+      "loss": 0.1516,
+      "step": 4412
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6126984126984125e-06,
+      "loss": 0.1503,
+      "step": 4413
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6095238095238105e-06,
+      "loss": 0.1411,
+      "step": 4414
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.606349206349208e-06,
+      "loss": 0.1485,
+      "step": 4415
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.603174603174603e-06,
+      "loss": 0.1443,
+      "step": 4416
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.1369,
+      "step": 4417
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.596825396825397e-06,
+      "loss": 0.145,
+      "step": 4418
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.593650793650795e-06,
+      "loss": 0.1358,
+      "step": 4419
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.59047619047619e-06,
+      "loss": 0.1438,
+      "step": 4420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.587301587301588e-06,
+      "loss": 0.1503,
+      "step": 4421
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.584126984126985e-06,
+      "loss": 0.1425,
+      "step": 4422
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.580952380952381e-06,
+      "loss": 0.1376,
+      "step": 4423
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.577777777777779e-06,
+      "loss": 0.1412,
+      "step": 4424
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.574603174603174e-06,
+      "loss": 0.1386,
+      "step": 4425
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.571428571428572e-06,
+      "loss": 0.1488,
+      "step": 4426
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.5682539682539694e-06,
+      "loss": 0.1446,
+      "step": 4427
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.565079365079366e-06,
+      "loss": 0.1504,
+      "step": 4428
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.561904761904763e-06,
+      "loss": 0.1483,
+      "step": 4429
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.558730158730159e-06,
+      "loss": 0.1471,
+      "step": 4430
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.555555555555556e-06,
+      "loss": 0.1336,
+      "step": 4431
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.552380952380954e-06,
+      "loss": 0.1438,
+      "step": 4432
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.54920634920635e-06,
+      "loss": 0.1417,
+      "step": 4433
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.546031746031747e-06,
+      "loss": 0.1435,
+      "step": 4434
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.542857142857143e-06,
+      "loss": 0.1343,
+      "step": 4435
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5396825396825405e-06,
+      "loss": 0.1392,
+      "step": 4436
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.536507936507938e-06,
+      "loss": 0.1339,
+      "step": 4437
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.533333333333334e-06,
+      "loss": 0.1478,
+      "step": 4438
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.530158730158731e-06,
+      "loss": 0.1416,
+      "step": 4439
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5269841269841275e-06,
+      "loss": 0.1358,
+      "step": 4440
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.523809523809525e-06,
+      "loss": 0.1441,
+      "step": 4441
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.520634920634921e-06,
+      "loss": 0.1274,
+      "step": 4442
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.517460317460318e-06,
+      "loss": 0.1342,
+      "step": 4443
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.514285714285715e-06,
+      "loss": 0.1428,
+      "step": 4444
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.511111111111112e-06,
+      "loss": 0.1526,
+      "step": 4445
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.507936507936509e-06,
+      "loss": 0.1447,
+      "step": 4446
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.504761904761905e-06,
+      "loss": 0.1524,
+      "step": 4447
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.501587301587302e-06,
+      "loss": 0.1363,
+      "step": 4448
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.4984126984126995e-06,
+      "loss": 0.1417,
+      "step": 4449
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.495238095238096e-06,
+      "loss": 0.1443,
+      "step": 4450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.492063492063493e-06,
+      "loss": 0.1434,
+      "step": 4451
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.488888888888889e-06,
+      "loss": 0.1472,
+      "step": 4452
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.485714285714286e-06,
+      "loss": 0.1513,
+      "step": 4453
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.482539682539684e-06,
+      "loss": 0.145,
+      "step": 4454
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.47936507936508e-06,
+      "loss": 0.1485,
+      "step": 4455
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.476190476190477e-06,
+      "loss": 0.1497,
+      "step": 4456
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.473015873015873e-06,
+      "loss": 0.1536,
+      "step": 4457
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4698412698412706e-06,
+      "loss": 0.1471,
+      "step": 4458
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.466666666666667e-06,
+      "loss": 0.1352,
+      "step": 4459
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.463492063492064e-06,
+      "loss": 0.1381,
+      "step": 4460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.460317460317461e-06,
+      "loss": 0.1455,
+      "step": 4461
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4571428571428575e-06,
+      "loss": 0.1384,
+      "step": 4462
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.453968253968255e-06,
+      "loss": 0.1389,
+      "step": 4463
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.450793650793651e-06,
+      "loss": 0.141,
+      "step": 4464
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.447619047619048e-06,
+      "loss": 0.1404,
+      "step": 4465
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.444444444444445e-06,
+      "loss": 0.1461,
+      "step": 4466
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.441269841269842e-06,
+      "loss": 0.1443,
+      "step": 4467
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.438095238095239e-06,
+      "loss": 0.1427,
+      "step": 4468
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.434920634920635e-06,
+      "loss": 0.1516,
+      "step": 4469
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.431746031746032e-06,
+      "loss": 0.1461,
+      "step": 4470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.143,
+      "step": 4471
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.425396825396826e-06,
+      "loss": 0.1365,
+      "step": 4472
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.422222222222223e-06,
+      "loss": 0.1369,
+      "step": 4473
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.419047619047619e-06,
+      "loss": 0.1379,
+      "step": 4474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.4158730158730164e-06,
+      "loss": 0.1453,
+      "step": 4475
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.412698412698414e-06,
+      "loss": 0.1359,
+      "step": 4476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.40952380952381e-06,
+      "loss": 0.1437,
+      "step": 4477
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.406349206349207e-06,
+      "loss": 0.1312,
+      "step": 4478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.403174603174603e-06,
+      "loss": 0.1491,
+      "step": 4479
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.129,
+      "step": 4480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.396825396825397e-06,
+      "loss": 0.1308,
+      "step": 4481
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.393650793650794e-06,
+      "loss": 0.1424,
+      "step": 4482
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.390476190476191e-06,
+      "loss": 0.1553,
+      "step": 4483
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3873015873015875e-06,
+      "loss": 0.1448,
+      "step": 4484
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.384126984126985e-06,
+      "loss": 0.152,
+      "step": 4485
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.380952380952381e-06,
+      "loss": 0.1425,
+      "step": 4486
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.377777777777778e-06,
+      "loss": 0.1421,
+      "step": 4487
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.374603174603175e-06,
+      "loss": 0.1471,
+      "step": 4488
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.371428571428572e-06,
+      "loss": 0.1465,
+      "step": 4489
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.368253968253969e-06,
+      "loss": 0.1398,
+      "step": 4490
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.365079365079365e-06,
+      "loss": 0.1544,
+      "step": 4491
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.361904761904762e-06,
+      "loss": 0.1383,
+      "step": 4492
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3587301587301595e-06,
+      "loss": 0.1408,
+      "step": 4493
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.355555555555556e-06,
+      "loss": 0.1427,
+      "step": 4494
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.352380952380953e-06,
+      "loss": 0.139,
+      "step": 4495
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.349206349206349e-06,
+      "loss": 0.1436,
+      "step": 4496
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3460317460317464e-06,
+      "loss": 0.142,
+      "step": 4497
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.342857142857143e-06,
+      "loss": 0.1368,
+      "step": 4498
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.33968253968254e-06,
+      "loss": 0.1356,
+      "step": 4499
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.336507936507937e-06,
+      "loss": 0.1408,
+      "step": 4500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 0.1373,
+      "step": 4501
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3301587301587306e-06,
+      "loss": 0.1375,
+      "step": 4502
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.326984126984127e-06,
+      "loss": 0.1411,
+      "step": 4503
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.323809523809524e-06,
+      "loss": 0.1404,
+      "step": 4504
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.320634920634921e-06,
+      "loss": 0.1424,
+      "step": 4505
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3174603174603175e-06,
+      "loss": 0.1477,
+      "step": 4506
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.314285714285715e-06,
+      "loss": 0.1391,
+      "step": 4507
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.311111111111111e-06,
+      "loss": 0.1488,
+      "step": 4508
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.307936507936508e-06,
+      "loss": 0.1413,
+      "step": 4509
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.304761904761905e-06,
+      "loss": 0.1591,
+      "step": 4510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.301587301587302e-06,
+      "loss": 0.1333,
+      "step": 4511
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.298412698412699e-06,
+      "loss": 0.1556,
+      "step": 4512
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.295238095238095e-06,
+      "loss": 0.1535,
+      "step": 4513
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.292063492063492e-06,
+      "loss": 0.1435,
+      "step": 4514
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.28888888888889e-06,
+      "loss": 0.149,
+      "step": 4515
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 0.1458,
+      "step": 4516
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.282539682539684e-06,
+      "loss": 0.1354,
+      "step": 4517
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.279365079365079e-06,
+      "loss": 0.1475,
+      "step": 4518
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2761904761904764e-06,
+      "loss": 0.1253,
+      "step": 4519
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.273015873015873e-06,
+      "loss": 0.156,
+      "step": 4520
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.26984126984127e-06,
+      "loss": 0.1475,
+      "step": 4521
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.266666666666668e-06,
+      "loss": 0.1438,
+      "step": 4522
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.263492063492063e-06,
+      "loss": 0.1355,
+      "step": 4523
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.260317460317461e-06,
+      "loss": 0.1454,
+      "step": 4524
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.257142857142857e-06,
+      "loss": 0.1422,
+      "step": 4525
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.253968253968254e-06,
+      "loss": 0.142,
+      "step": 4526
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.250793650793652e-06,
+      "loss": 0.1477,
+      "step": 4527
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2476190476190475e-06,
+      "loss": 0.1404,
+      "step": 4528
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2444444444444456e-06,
+      "loss": 0.1316,
+      "step": 4529
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.241269841269841e-06,
+      "loss": 0.1354,
+      "step": 4530
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.238095238095239e-06,
+      "loss": 0.1528,
+      "step": 4531
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.234920634920636e-06,
+      "loss": 0.1483,
+      "step": 4532
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2317460317460325e-06,
+      "loss": 0.1464,
+      "step": 4533
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.22857142857143e-06,
+      "loss": 0.149,
+      "step": 4534
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.225396825396825e-06,
+      "loss": 0.1358,
+      "step": 4535
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 0.145,
+      "step": 4536
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.21904761904762e-06,
+      "loss": 0.1384,
+      "step": 4537
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.215873015873017e-06,
+      "loss": 0.1381,
+      "step": 4538
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.212698412698414e-06,
+      "loss": 0.1472,
+      "step": 4539
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.20952380952381e-06,
+      "loss": 0.1454,
+      "step": 4540
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.206349206349207e-06,
+      "loss": 0.1501,
+      "step": 4541
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.203174603174603e-06,
+      "loss": 0.1471,
+      "step": 4542
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.1498,
+      "step": 4543
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.196825396825398e-06,
+      "loss": 0.1522,
+      "step": 4544
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.193650793650794e-06,
+      "loss": 0.1397,
+      "step": 4545
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1904761904761914e-06,
+      "loss": 0.1364,
+      "step": 4546
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.187301587301588e-06,
+      "loss": 0.1489,
+      "step": 4547
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.184126984126985e-06,
+      "loss": 0.1452,
+      "step": 4548
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.180952380952382e-06,
+      "loss": 0.1355,
+      "step": 4549
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.177777777777778e-06,
+      "loss": 0.1421,
+      "step": 4550
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1746031746031756e-06,
+      "loss": 0.1388,
+      "step": 4551
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.171428571428572e-06,
+      "loss": 0.1521,
+      "step": 4552
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.168253968253969e-06,
+      "loss": 0.1393,
+      "step": 4553
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.165079365079366e-06,
+      "loss": 0.1448,
+      "step": 4554
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1619047619047625e-06,
+      "loss": 0.1478,
+      "step": 4555
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.15873015873016e-06,
+      "loss": 0.1333,
+      "step": 4556
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.155555555555556e-06,
+      "loss": 0.1429,
+      "step": 4557
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.152380952380953e-06,
+      "loss": 0.1394,
+      "step": 4558
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1492063492063495e-06,
+      "loss": 0.1313,
+      "step": 4559
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.146031746031747e-06,
+      "loss": 0.143,
+      "step": 4560
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.142857142857144e-06,
+      "loss": 0.1484,
+      "step": 4561
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.13968253968254e-06,
+      "loss": 0.1328,
+      "step": 4562
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.136507936507937e-06,
+      "loss": 0.1427,
+      "step": 4563
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 0.1438,
+      "step": 4564
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.130158730158731e-06,
+      "loss": 0.1376,
+      "step": 4565
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.126984126984128e-06,
+      "loss": 0.144,
+      "step": 4566
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.123809523809524e-06,
+      "loss": 0.1537,
+      "step": 4567
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1206349206349214e-06,
+      "loss": 0.1469,
+      "step": 4568
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.117460317460318e-06,
+      "loss": 0.1384,
+      "step": 4569
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.114285714285715e-06,
+      "loss": 0.1322,
+      "step": 4570
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 0.144,
+      "step": 4571
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.107936507936508e-06,
+      "loss": 0.1398,
+      "step": 4572
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1047619047619056e-06,
+      "loss": 0.1336,
+      "step": 4573
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.101587301587302e-06,
+      "loss": 0.1473,
+      "step": 4574
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.098412698412699e-06,
+      "loss": 0.1441,
+      "step": 4575
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.095238095238096e-06,
+      "loss": 0.1429,
+      "step": 4576
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.0920634920634925e-06,
+      "loss": 0.1517,
+      "step": 4577
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.08888888888889e-06,
+      "loss": 0.1382,
+      "step": 4578
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.085714285714286e-06,
+      "loss": 0.1564,
+      "step": 4579
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.082539682539683e-06,
+      "loss": 0.133,
+      "step": 4580
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0793650793650795e-06,
+      "loss": 0.147,
+      "step": 4581
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.076190476190477e-06,
+      "loss": 0.1427,
+      "step": 4582
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.073015873015874e-06,
+      "loss": 0.1424,
+      "step": 4583
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.06984126984127e-06,
+      "loss": 0.1482,
+      "step": 4584
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.066666666666667e-06,
+      "loss": 0.1511,
+      "step": 4585
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.063492063492064e-06,
+      "loss": 0.1468,
+      "step": 4586
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.060317460317461e-06,
+      "loss": 0.1401,
+      "step": 4587
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.057142857142858e-06,
+      "loss": 0.1385,
+      "step": 4588
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.053968253968254e-06,
+      "loss": 0.1355,
+      "step": 4589
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0507936507936514e-06,
+      "loss": 0.1468,
+      "step": 4590
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.047619047619048e-06,
+      "loss": 0.1447,
+      "step": 4591
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.044444444444445e-06,
+      "loss": 0.1414,
+      "step": 4592
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.041269841269842e-06,
+      "loss": 0.1441,
+      "step": 4593
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.038095238095238e-06,
+      "loss": 0.1329,
+      "step": 4594
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0349206349206356e-06,
+      "loss": 0.1383,
+      "step": 4595
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.031746031746032e-06,
+      "loss": 0.1471,
+      "step": 4596
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.028571428571429e-06,
+      "loss": 0.1369,
+      "step": 4597
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.025396825396825e-06,
+      "loss": 0.1494,
+      "step": 4598
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0222222222222225e-06,
+      "loss": 0.1335,
+      "step": 4599
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.01904761904762e-06,
+      "loss": 0.1425,
+      "step": 4600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.015873015873016e-06,
+      "loss": 0.1482,
+      "step": 4601
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.012698412698413e-06,
+      "loss": 0.1376,
+      "step": 4602
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.0095238095238095e-06,
+      "loss": 0.1408,
+      "step": 4603
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.006349206349207e-06,
+      "loss": 0.1375,
+      "step": 4604
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.003174603174604e-06,
+      "loss": 0.1484,
+      "step": 4605
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6e-06,
+      "loss": 0.1469,
+      "step": 4606
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.996825396825397e-06,
+      "loss": 0.1438,
+      "step": 4607
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.993650793650794e-06,
+      "loss": 0.1475,
+      "step": 4608
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.990476190476191e-06,
+      "loss": 0.1502,
+      "step": 4609
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.987301587301588e-06,
+      "loss": 0.1433,
+      "step": 4610
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.984126984126984e-06,
+      "loss": 0.1508,
+      "step": 4611
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9809523809523814e-06,
+      "loss": 0.1412,
+      "step": 4612
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.977777777777778e-06,
+      "loss": 0.1369,
+      "step": 4613
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.974603174603175e-06,
+      "loss": 0.1395,
+      "step": 4614
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.971428571428572e-06,
+      "loss": 0.1399,
+      "step": 4615
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.968253968253968e-06,
+      "loss": 0.1401,
+      "step": 4616
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.965079365079366e-06,
+      "loss": 0.1375,
+      "step": 4617
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.961904761904762e-06,
+      "loss": 0.134,
+      "step": 4618
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.958730158730159e-06,
+      "loss": 0.1531,
+      "step": 4619
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.955555555555555e-06,
+      "loss": 0.1471,
+      "step": 4620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9523809523809525e-06,
+      "loss": 0.1443,
+      "step": 4621
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.94920634920635e-06,
+      "loss": 0.1445,
+      "step": 4622
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.946031746031746e-06,
+      "loss": 0.1409,
+      "step": 4623
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.942857142857143e-06,
+      "loss": 0.154,
+      "step": 4624
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.9396825396825395e-06,
+      "loss": 0.1483,
+      "step": 4625
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.936507936507937e-06,
+      "loss": 0.1489,
+      "step": 4626
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.933333333333335e-06,
+      "loss": 0.1281,
+      "step": 4627
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.93015873015873e-06,
+      "loss": 0.1406,
+      "step": 4628
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.926984126984127e-06,
+      "loss": 0.1401,
+      "step": 4629
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.923809523809524e-06,
+      "loss": 0.1455,
+      "step": 4630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.920634920634921e-06,
+      "loss": 0.1285,
+      "step": 4631
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.917460317460319e-06,
+      "loss": 0.1414,
+      "step": 4632
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.914285714285714e-06,
+      "loss": 0.1335,
+      "step": 4633
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.911111111111112e-06,
+      "loss": 0.1512,
+      "step": 4634
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.907936507936508e-06,
+      "loss": 0.1407,
+      "step": 4635
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.904761904761905e-06,
+      "loss": 0.1438,
+      "step": 4636
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.901587301587301e-06,
+      "loss": 0.1487,
+      "step": 4637
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.898412698412698e-06,
+      "loss": 0.1381,
+      "step": 4638
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.8952380952380964e-06,
+      "loss": 0.1601,
+      "step": 4639
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.892063492063492e-06,
+      "loss": 0.136,
+      "step": 4640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.88888888888889e-06,
+      "loss": 0.1418,
+      "step": 4641
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.885714285714285e-06,
+      "loss": 0.1482,
+      "step": 4642
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.882539682539683e-06,
+      "loss": 0.1339,
+      "step": 4643
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.8793650793650806e-06,
+      "loss": 0.1407,
+      "step": 4644
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.876190476190476e-06,
+      "loss": 0.153,
+      "step": 4645
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.873015873015874e-06,
+      "loss": 0.1455,
+      "step": 4646
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8698412698412695e-06,
+      "loss": 0.1367,
+      "step": 4647
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 0.1392,
+      "step": 4648
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.863492063492065e-06,
+      "loss": 0.1305,
+      "step": 4649
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.860317460317461e-06,
+      "loss": 0.1438,
+      "step": 4650
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.857142857142858e-06,
+      "loss": 0.1446,
+      "step": 4651
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8539682539682545e-06,
+      "loss": 0.1509,
+      "step": 4652
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.850793650793652e-06,
+      "loss": 0.1454,
+      "step": 4653
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.847619047619049e-06,
+      "loss": 0.1342,
+      "step": 4654
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.844444444444445e-06,
+      "loss": 0.1354,
+      "step": 4655
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.841269841269842e-06,
+      "loss": 0.1405,
+      "step": 4656
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.838095238095239e-06,
+      "loss": 0.1413,
+      "step": 4657
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.834920634920636e-06,
+      "loss": 0.1437,
+      "step": 4658
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.831746031746032e-06,
+      "loss": 0.1398,
+      "step": 4659
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.828571428571429e-06,
+      "loss": 0.1469,
+      "step": 4660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8253968253968264e-06,
+      "loss": 0.142,
+      "step": 4661
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.822222222222223e-06,
+      "loss": 0.1456,
+      "step": 4662
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.81904761904762e-06,
+      "loss": 0.1447,
+      "step": 4663
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.815873015873016e-06,
+      "loss": 0.1346,
+      "step": 4664
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.812698412698413e-06,
+      "loss": 0.1482,
+      "step": 4665
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8095238095238106e-06,
+      "loss": 0.1424,
+      "step": 4666
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.806349206349207e-06,
+      "loss": 0.1371,
+      "step": 4667
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.803174603174604e-06,
+      "loss": 0.1339,
+      "step": 4668
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.8e-06,
+      "loss": 0.1455,
+      "step": 4669
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7968253968253975e-06,
+      "loss": 0.1429,
+      "step": 4670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.793650793650795e-06,
+      "loss": 0.1435,
+      "step": 4671
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.790476190476191e-06,
+      "loss": 0.1439,
+      "step": 4672
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.787301587301588e-06,
+      "loss": 0.1449,
+      "step": 4673
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7841269841269845e-06,
+      "loss": 0.1465,
+      "step": 4674
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.780952380952382e-06,
+      "loss": 0.1287,
+      "step": 4675
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 0.1395,
+      "step": 4676
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.774603174603175e-06,
+      "loss": 0.1426,
+      "step": 4677
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.771428571428572e-06,
+      "loss": 0.1491,
+      "step": 4678
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.768253968253969e-06,
+      "loss": 0.1305,
+      "step": 4679
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.765079365079366e-06,
+      "loss": 0.1471,
+      "step": 4680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.761904761904762e-06,
+      "loss": 0.143,
+      "step": 4681
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.758730158730159e-06,
+      "loss": 0.1416,
+      "step": 4682
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7555555555555564e-06,
+      "loss": 0.1432,
+      "step": 4683
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.752380952380953e-06,
+      "loss": 0.1474,
+      "step": 4684
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.74920634920635e-06,
+      "loss": 0.1458,
+      "step": 4685
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.746031746031746e-06,
+      "loss": 0.1396,
+      "step": 4686
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.742857142857143e-06,
+      "loss": 0.1366,
+      "step": 4687
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.739682539682541e-06,
+      "loss": 0.1372,
+      "step": 4688
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.736507936507937e-06,
+      "loss": 0.1363,
+      "step": 4689
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.733333333333334e-06,
+      "loss": 0.1519,
+      "step": 4690
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.73015873015873e-06,
+      "loss": 0.1439,
+      "step": 4691
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7269841269841275e-06,
+      "loss": 0.1379,
+      "step": 4692
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.723809523809525e-06,
+      "loss": 0.1381,
+      "step": 4693
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.720634920634921e-06,
+      "loss": 0.1454,
+      "step": 4694
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.717460317460318e-06,
+      "loss": 0.1434,
+      "step": 4695
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.1431,
+      "step": 4696
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.711111111111112e-06,
+      "loss": 0.1485,
+      "step": 4697
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.707936507936508e-06,
+      "loss": 0.1533,
+      "step": 4698
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.704761904761905e-06,
+      "loss": 0.1364,
+      "step": 4699
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.701587301587302e-06,
+      "loss": 0.1475,
+      "step": 4700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.698412698412699e-06,
+      "loss": 0.138,
+      "step": 4701
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.695238095238096e-06,
+      "loss": 0.1379,
+      "step": 4702
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.692063492063492e-06,
+      "loss": 0.1389,
+      "step": 4703
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.688888888888889e-06,
+      "loss": 0.1388,
+      "step": 4704
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.6857142857142865e-06,
+      "loss": 0.1445,
+      "step": 4705
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.682539682539683e-06,
+      "loss": 0.1462,
+      "step": 4706
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.67936507936508e-06,
+      "loss": 0.1327,
+      "step": 4707
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.676190476190476e-06,
+      "loss": 0.1507,
+      "step": 4708
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.673015873015873e-06,
+      "loss": 0.1499,
+      "step": 4709
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.669841269841271e-06,
+      "loss": 0.1369,
+      "step": 4710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 0.1405,
+      "step": 4711
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.663492063492064e-06,
+      "loss": 0.1425,
+      "step": 4712
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.66031746031746e-06,
+      "loss": 0.1396,
+      "step": 4713
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6571428571428576e-06,
+      "loss": 0.1418,
+      "step": 4714
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.653968253968254e-06,
+      "loss": 0.1545,
+      "step": 4715
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.650793650793651e-06,
+      "loss": 0.1376,
+      "step": 4716
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.647619047619048e-06,
+      "loss": 0.1454,
+      "step": 4717
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6444444444444445e-06,
+      "loss": 0.1441,
+      "step": 4718
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.641269841269842e-06,
+      "loss": 0.1479,
+      "step": 4719
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.638095238095238e-06,
+      "loss": 0.151,
+      "step": 4720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.634920634920635e-06,
+      "loss": 0.1499,
+      "step": 4721
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.631746031746032e-06,
+      "loss": 0.1569,
+      "step": 4722
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.628571428571429e-06,
+      "loss": 0.1408,
+      "step": 4723
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.625396825396826e-06,
+      "loss": 0.1501,
+      "step": 4724
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.622222222222222e-06,
+      "loss": 0.1395,
+      "step": 4725
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.619047619047619e-06,
+      "loss": 0.1474,
+      "step": 4726
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6158730158730165e-06,
+      "loss": 0.138,
+      "step": 4727
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.612698412698413e-06,
+      "loss": 0.1401,
+      "step": 4728
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.60952380952381e-06,
+      "loss": 0.1484,
+      "step": 4729
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.606349206349206e-06,
+      "loss": 0.1429,
+      "step": 4730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6031746031746034e-06,
+      "loss": 0.1445,
+      "step": 4731
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.1401,
+      "step": 4732
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.596825396825397e-06,
+      "loss": 0.1393,
+      "step": 4733
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.593650793650794e-06,
+      "loss": 0.1484,
+      "step": 4734
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.59047619047619e-06,
+      "loss": 0.1521,
+      "step": 4735
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.5873015873015876e-06,
+      "loss": 0.1337,
+      "step": 4736
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.584126984126984e-06,
+      "loss": 0.1408,
+      "step": 4737
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.580952380952381e-06,
+      "loss": 0.1338,
+      "step": 4738
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.577777777777778e-06,
+      "loss": 0.1414,
+      "step": 4739
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.5746031746031745e-06,
+      "loss": 0.1514,
+      "step": 4740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.571428571428572e-06,
+      "loss": 0.1428,
+      "step": 4741
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.568253968253968e-06,
+      "loss": 0.1366,
+      "step": 4742
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.565079365079365e-06,
+      "loss": 0.134,
+      "step": 4743
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.561904761904763e-06,
+      "loss": 0.1373,
+      "step": 4744
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.558730158730159e-06,
+      "loss": 0.1393,
+      "step": 4745
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 0.1326,
+      "step": 4746
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.552380952380952e-06,
+      "loss": 0.139,
+      "step": 4747
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.549206349206349e-06,
+      "loss": 0.1429,
+      "step": 4748
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.546031746031747e-06,
+      "loss": 0.1485,
+      "step": 4749
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.542857142857143e-06,
+      "loss": 0.1547,
+      "step": 4750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.539682539682541e-06,
+      "loss": 0.1344,
+      "step": 4751
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.536507936507936e-06,
+      "loss": 0.1428,
+      "step": 4752
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.533333333333334e-06,
+      "loss": 0.1409,
+      "step": 4753
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.53015873015873e-06,
+      "loss": 0.1441,
+      "step": 4754
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.526984126984127e-06,
+      "loss": 0.1412,
+      "step": 4755
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.523809523809525e-06,
+      "loss": 0.1392,
+      "step": 4756
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.52063492063492e-06,
+      "loss": 0.1457,
+      "step": 4757
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.517460317460318e-06,
+      "loss": 0.1426,
+      "step": 4758
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.514285714285714e-06,
+      "loss": 0.1472,
+      "step": 4759
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.511111111111112e-06,
+      "loss": 0.147,
+      "step": 4760
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.507936507936509e-06,
+      "loss": 0.1273,
+      "step": 4761
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.504761904761905e-06,
+      "loss": 0.1479,
+      "step": 4762
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.5015873015873025e-06,
+      "loss": 0.1477,
+      "step": 4763
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.498412698412698e-06,
+      "loss": 0.1499,
+      "step": 4764
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.495238095238096e-06,
+      "loss": 0.1426,
+      "step": 4765
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.492063492063493e-06,
+      "loss": 0.1411,
+      "step": 4766
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.4888888888888895e-06,
+      "loss": 0.1491,
+      "step": 4767
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.485714285714287e-06,
+      "loss": 0.1418,
+      "step": 4768
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.482539682539683e-06,
+      "loss": 0.1445,
+      "step": 4769
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.47936507936508e-06,
+      "loss": 0.1383,
+      "step": 4770
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.476190476190477e-06,
+      "loss": 0.1465,
+      "step": 4771
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.473015873015874e-06,
+      "loss": 0.1408,
+      "step": 4772
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.469841269841271e-06,
+      "loss": 0.1377,
+      "step": 4773
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.466666666666667e-06,
+      "loss": 0.1381,
+      "step": 4774
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.463492063492064e-06,
+      "loss": 0.1357,
+      "step": 4775
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.460317460317461e-06,
+      "loss": 0.1421,
+      "step": 4776
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.457142857142858e-06,
+      "loss": 0.1515,
+      "step": 4777
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.453968253968255e-06,
+      "loss": 0.1358,
+      "step": 4778
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.450793650793651e-06,
+      "loss": 0.1431,
+      "step": 4779
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.447619047619048e-06,
+      "loss": 0.1534,
+      "step": 4780
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.444444444444445e-06,
+      "loss": 0.1493,
+      "step": 4781
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.441269841269842e-06,
+      "loss": 0.1463,
+      "step": 4782
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.438095238095239e-06,
+      "loss": 0.1394,
+      "step": 4783
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.434920634920635e-06,
+      "loss": 0.1561,
+      "step": 4784
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4317460317460326e-06,
+      "loss": 0.1469,
+      "step": 4785
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.428571428571429e-06,
+      "loss": 0.1356,
+      "step": 4786
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.425396825396826e-06,
+      "loss": 0.1541,
+      "step": 4787
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.422222222222223e-06,
+      "loss": 0.1361,
+      "step": 4788
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4190476190476195e-06,
+      "loss": 0.1444,
+      "step": 4789
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.415873015873017e-06,
+      "loss": 0.1372,
+      "step": 4790
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.412698412698413e-06,
+      "loss": 0.1494,
+      "step": 4791
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.40952380952381e-06,
+      "loss": 0.1298,
+      "step": 4792
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.406349206349207e-06,
+      "loss": 0.1373,
+      "step": 4793
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.403174603174604e-06,
+      "loss": 0.1385,
+      "step": 4794
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.1342,
+      "step": 4795
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.396825396825397e-06,
+      "loss": 0.1572,
+      "step": 4796
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.393650793650794e-06,
+      "loss": 0.1406,
+      "step": 4797
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.390476190476191e-06,
+      "loss": 0.1421,
+      "step": 4798
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.387301587301588e-06,
+      "loss": 0.1431,
+      "step": 4799
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.384126984126985e-06,
+      "loss": 0.1426,
+      "step": 4800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.380952380952381e-06,
+      "loss": 0.1468,
+      "step": 4801
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3777777777777784e-06,
+      "loss": 0.1503,
+      "step": 4802
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.374603174603175e-06,
+      "loss": 0.1353,
+      "step": 4803
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.371428571428572e-06,
+      "loss": 0.1389,
+      "step": 4804
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.368253968253969e-06,
+      "loss": 0.1351,
+      "step": 4805
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.365079365079365e-06,
+      "loss": 0.1349,
+      "step": 4806
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3619047619047626e-06,
+      "loss": 0.1317,
+      "step": 4807
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.358730158730159e-06,
+      "loss": 0.1459,
+      "step": 4808
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.355555555555556e-06,
+      "loss": 0.1447,
+      "step": 4809
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.352380952380953e-06,
+      "loss": 0.1401,
+      "step": 4810
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3492063492063495e-06,
+      "loss": 0.1361,
+      "step": 4811
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.346031746031747e-06,
+      "loss": 0.1533,
+      "step": 4812
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.342857142857143e-06,
+      "loss": 0.1369,
+      "step": 4813
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.33968253968254e-06,
+      "loss": 0.141,
+      "step": 4814
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3365079365079365e-06,
+      "loss": 0.1332,
+      "step": 4815
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.1419,
+      "step": 4816
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.330158730158731e-06,
+      "loss": 0.1435,
+      "step": 4817
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.326984126984127e-06,
+      "loss": 0.1496,
+      "step": 4818
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.323809523809524e-06,
+      "loss": 0.1446,
+      "step": 4819
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.320634920634921e-06,
+      "loss": 0.1383,
+      "step": 4820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.317460317460318e-06,
+      "loss": 0.1485,
+      "step": 4821
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.314285714285715e-06,
+      "loss": 0.1415,
+      "step": 4822
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.311111111111111e-06,
+      "loss": 0.1406,
+      "step": 4823
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.3079365079365084e-06,
+      "loss": 0.1515,
+      "step": 4824
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.304761904761905e-06,
+      "loss": 0.1443,
+      "step": 4825
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.301587301587302e-06,
+      "loss": 0.1478,
+      "step": 4826
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.298412698412699e-06,
+      "loss": 0.1541,
+      "step": 4827
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.295238095238095e-06,
+      "loss": 0.1392,
+      "step": 4828
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.2920634920634926e-06,
+      "loss": 0.1435,
+      "step": 4829
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.288888888888889e-06,
+      "loss": 0.1405,
+      "step": 4830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.285714285714286e-06,
+      "loss": 0.1507,
+      "step": 4831
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.282539682539683e-06,
+      "loss": 0.152,
+      "step": 4832
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.2793650793650795e-06,
+      "loss": 0.1507,
+      "step": 4833
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.276190476190477e-06,
+      "loss": 0.1379,
+      "step": 4834
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.273015873015873e-06,
+      "loss": 0.1422,
+      "step": 4835
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.26984126984127e-06,
+      "loss": 0.1446,
+      "step": 4836
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.2666666666666665e-06,
+      "loss": 0.1415,
+      "step": 4837
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.263492063492064e-06,
+      "loss": 0.1413,
+      "step": 4838
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.260317460317461e-06,
+      "loss": 0.1478,
+      "step": 4839
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.257142857142857e-06,
+      "loss": 0.1472,
+      "step": 4840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.253968253968254e-06,
+      "loss": 0.1412,
+      "step": 4841
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.250793650793651e-06,
+      "loss": 0.1536,
+      "step": 4842
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.247619047619048e-06,
+      "loss": 0.1392,
+      "step": 4843
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.244444444444445e-06,
+      "loss": 0.1431,
+      "step": 4844
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.241269841269841e-06,
+      "loss": 0.1481,
+      "step": 4845
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2380952380952384e-06,
+      "loss": 0.1243,
+      "step": 4846
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.234920634920635e-06,
+      "loss": 0.1497,
+      "step": 4847
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.231746031746032e-06,
+      "loss": 0.136,
+      "step": 4848
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.22857142857143e-06,
+      "loss": 0.1385,
+      "step": 4849
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.225396825396825e-06,
+      "loss": 0.144,
+      "step": 4850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2222222222222226e-06,
+      "loss": 0.1366,
+      "step": 4851
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.219047619047619e-06,
+      "loss": 0.1399,
+      "step": 4852
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.215873015873016e-06,
+      "loss": 0.1456,
+      "step": 4853
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.212698412698412e-06,
+      "loss": 0.1535,
+      "step": 4854
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2095238095238095e-06,
+      "loss": 0.1424,
+      "step": 4855
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2063492063492076e-06,
+      "loss": 0.1438,
+      "step": 4856
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.203174603174603e-06,
+      "loss": 0.1425,
+      "step": 4857
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2e-06,
+      "loss": 0.147,
+      "step": 4858
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.1968253968253965e-06,
+      "loss": 0.1318,
+      "step": 4859
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.193650793650794e-06,
+      "loss": 0.1343,
+      "step": 4860
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.190476190476192e-06,
+      "loss": 0.1484,
+      "step": 4861
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.187301587301587e-06,
+      "loss": 0.1462,
+      "step": 4862
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.184126984126985e-06,
+      "loss": 0.1435,
+      "step": 4863
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.180952380952381e-06,
+      "loss": 0.148,
+      "step": 4864
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.177777777777779e-06,
+      "loss": 0.137,
+      "step": 4865
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.174603174603176e-06,
+      "loss": 0.1438,
+      "step": 4866
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.171428571428571e-06,
+      "loss": 0.1472,
+      "step": 4867
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.168253968253969e-06,
+      "loss": 0.1332,
+      "step": 4868
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.165079365079365e-06,
+      "loss": 0.1494,
+      "step": 4869
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.161904761904763e-06,
+      "loss": 0.1491,
+      "step": 4870
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.15873015873016e-06,
+      "loss": 0.1415,
+      "step": 4871
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.155555555555556e-06,
+      "loss": 0.1407,
+      "step": 4872
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.1523809523809534e-06,
+      "loss": 0.1445,
+      "step": 4873
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.149206349206349e-06,
+      "loss": 0.1415,
+      "step": 4874
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.146031746031747e-06,
+      "loss": 0.1529,
+      "step": 4875
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 0.1419,
+      "step": 4876
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.13968253968254e-06,
+      "loss": 0.1526,
+      "step": 4877
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.1365079365079376e-06,
+      "loss": 0.1461,
+      "step": 4878
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.133333333333334e-06,
+      "loss": 0.1535,
+      "step": 4879
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.130158730158731e-06,
+      "loss": 0.1342,
+      "step": 4880
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.126984126984127e-06,
+      "loss": 0.1426,
+      "step": 4881
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.1238095238095245e-06,
+      "loss": 0.1378,
+      "step": 4882
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.120634920634922e-06,
+      "loss": 0.1557,
+      "step": 4883
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.117460317460318e-06,
+      "loss": 0.1493,
+      "step": 4884
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.114285714285715e-06,
+      "loss": 0.1494,
+      "step": 4885
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.1111111111111115e-06,
+      "loss": 0.1534,
+      "step": 4886
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.107936507936509e-06,
+      "loss": 0.1415,
+      "step": 4887
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.104761904761906e-06,
+      "loss": 0.1407,
+      "step": 4888
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.101587301587302e-06,
+      "loss": 0.1349,
+      "step": 4889
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.098412698412699e-06,
+      "loss": 0.1394,
+      "step": 4890
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.095238095238096e-06,
+      "loss": 0.1458,
+      "step": 4891
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.092063492063493e-06,
+      "loss": 0.1338,
+      "step": 4892
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.088888888888889e-06,
+      "loss": 0.153,
+      "step": 4893
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.085714285714286e-06,
+      "loss": 0.1397,
+      "step": 4894
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0825396825396834e-06,
+      "loss": 0.1396,
+      "step": 4895
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.07936507936508e-06,
+      "loss": 0.138,
+      "step": 4896
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.076190476190477e-06,
+      "loss": 0.1456,
+      "step": 4897
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.073015873015873e-06,
+      "loss": 0.1531,
+      "step": 4898
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.06984126984127e-06,
+      "loss": 0.1428,
+      "step": 4899
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 0.1341,
+      "step": 4900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.063492063492064e-06,
+      "loss": 0.1556,
+      "step": 4901
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.060317460317461e-06,
+      "loss": 0.1287,
+      "step": 4902
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.057142857142857e-06,
+      "loss": 0.153,
+      "step": 4903
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0539682539682545e-06,
+      "loss": 0.1558,
+      "step": 4904
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.050793650793652e-06,
+      "loss": 0.1526,
+      "step": 4905
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.047619047619048e-06,
+      "loss": 0.1371,
+      "step": 4906
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.044444444444445e-06,
+      "loss": 0.1295,
+      "step": 4907
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.0412698412698415e-06,
+      "loss": 0.1373,
+      "step": 4908
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.038095238095239e-06,
+      "loss": 0.1334,
+      "step": 4909
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.034920634920636e-06,
+      "loss": 0.133,
+      "step": 4910
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.031746031746032e-06,
+      "loss": 0.1392,
+      "step": 4911
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.028571428571429e-06,
+      "loss": 0.1421,
+      "step": 4912
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.025396825396826e-06,
+      "loss": 0.1577,
+      "step": 4913
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.022222222222223e-06,
+      "loss": 0.1428,
+      "step": 4914
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.019047619047619e-06,
+      "loss": 0.1391,
+      "step": 4915
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.015873015873016e-06,
+      "loss": 0.1421,
+      "step": 4916
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.0126984126984134e-06,
+      "loss": 0.1359,
+      "step": 4917
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.00952380952381e-06,
+      "loss": 0.1436,
+      "step": 4918
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.006349206349207e-06,
+      "loss": 0.1442,
+      "step": 4919
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.003174603174603e-06,
+      "loss": 0.1409,
+      "step": 4920
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "step": 4921
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.996825396825397e-06,
+      "loss": 0.1384,
+      "step": 4922
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.993650793650794e-06,
+      "loss": 0.1497,
+      "step": 4923
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.990476190476191e-06,
+      "loss": 0.1442,
+      "step": 4924
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.987301587301588e-06,
+      "loss": 0.1414,
+      "step": 4925
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.9841269841269845e-06,
+      "loss": 0.1336,
+      "step": 4926
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.980952380952382e-06,
+      "loss": 0.1366,
+      "step": 4927
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.977777777777778e-06,
+      "loss": 0.1305,
+      "step": 4928
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.974603174603175e-06,
+      "loss": 0.1461,
+      "step": 4929
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.971428571428572e-06,
+      "loss": 0.1466,
+      "step": 4930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.968253968253969e-06,
+      "loss": 0.1401,
+      "step": 4931
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.965079365079366e-06,
+      "loss": 0.1424,
+      "step": 4932
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.961904761904762e-06,
+      "loss": 0.1448,
+      "step": 4933
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.958730158730159e-06,
+      "loss": 0.1358,
+      "step": 4934
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9555555555555565e-06,
+      "loss": 0.1309,
+      "step": 4935
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.952380952380953e-06,
+      "loss": 0.1442,
+      "step": 4936
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.94920634920635e-06,
+      "loss": 0.1394,
+      "step": 4937
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.946031746031746e-06,
+      "loss": 0.1455,
+      "step": 4938
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9428571428571435e-06,
+      "loss": 0.1388,
+      "step": 4939
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.93968253968254e-06,
+      "loss": 0.136,
+      "step": 4940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.936507936507937e-06,
+      "loss": 0.1455,
+      "step": 4941
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.933333333333334e-06,
+      "loss": 0.1416,
+      "step": 4942
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.93015873015873e-06,
+      "loss": 0.1549,
+      "step": 4943
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.926984126984128e-06,
+      "loss": 0.1364,
+      "step": 4944
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.923809523809524e-06,
+      "loss": 0.145,
+      "step": 4945
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.920634920634921e-06,
+      "loss": 0.1408,
+      "step": 4946
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.917460317460318e-06,
+      "loss": 0.1361,
+      "step": 4947
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9142857142857145e-06,
+      "loss": 0.1409,
+      "step": 4948
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.911111111111112e-06,
+      "loss": 0.1352,
+      "step": 4949
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.907936507936508e-06,
+      "loss": 0.1541,
+      "step": 4950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.904761904761905e-06,
+      "loss": 0.1352,
+      "step": 4951
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.901587301587302e-06,
+      "loss": 0.1448,
+      "step": 4952
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.898412698412699e-06,
+      "loss": 0.1335,
+      "step": 4953
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.895238095238096e-06,
+      "loss": 0.1405,
+      "step": 4954
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.892063492063492e-06,
+      "loss": 0.1356,
+      "step": 4955
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 0.1317,
+      "step": 4956
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.885714285714286e-06,
+      "loss": 0.1419,
+      "step": 4957
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.882539682539683e-06,
+      "loss": 0.143,
+      "step": 4958
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.87936507936508e-06,
+      "loss": 0.1426,
+      "step": 4959
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.876190476190476e-06,
+      "loss": 0.1241,
+      "step": 4960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8730158730158735e-06,
+      "loss": 0.1417,
+      "step": 4961
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.86984126984127e-06,
+      "loss": 0.1405,
+      "step": 4962
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.866666666666667e-06,
+      "loss": 0.1253,
+      "step": 4963
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.863492063492064e-06,
+      "loss": 0.127,
+      "step": 4964
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8603174603174604e-06,
+      "loss": 0.1464,
+      "step": 4965
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.857142857142858e-06,
+      "loss": 0.1461,
+      "step": 4966
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.853968253968254e-06,
+      "loss": 0.1416,
+      "step": 4967
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.850793650793651e-06,
+      "loss": 0.1422,
+      "step": 4968
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.847619047619048e-06,
+      "loss": 0.1483,
+      "step": 4969
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8444444444444446e-06,
+      "loss": 0.1417,
+      "step": 4970
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.841269841269842e-06,
+      "loss": 0.1463,
+      "step": 4971
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.838095238095238e-06,
+      "loss": 0.1492,
+      "step": 4972
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.834920634920635e-06,
+      "loss": 0.1465,
+      "step": 4973
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.831746031746032e-06,
+      "loss": 0.1428,
+      "step": 4974
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8285714285714295e-06,
+      "loss": 0.1379,
+      "step": 4975
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.825396825396826e-06,
+      "loss": 0.1368,
+      "step": 4976
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.822222222222222e-06,
+      "loss": 0.1429,
+      "step": 4977
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.819047619047619e-06,
+      "loss": 0.1284,
+      "step": 4978
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.815873015873016e-06,
+      "loss": 0.1444,
+      "step": 4979
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.812698412698414e-06,
+      "loss": 0.1353,
+      "step": 4980
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.80952380952381e-06,
+      "loss": 0.141,
+      "step": 4981
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.806349206349207e-06,
+      "loss": 0.1425,
+      "step": 4982
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8031746031746035e-06,
+      "loss": 0.1443,
+      "step": 4983
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.1448,
+      "step": 4984
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.796825396825397e-06,
+      "loss": 0.1431,
+      "step": 4985
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.793650793650794e-06,
+      "loss": 0.1465,
+      "step": 4986
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.790476190476191e-06,
+      "loss": 0.1442,
+      "step": 4987
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.787301587301588e-06,
+      "loss": 0.1475,
+      "step": 4988
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.784126984126985e-06,
+      "loss": 0.1307,
+      "step": 4989
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.780952380952381e-06,
+      "loss": 0.1402,
+      "step": 4990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.777777777777778e-06,
+      "loss": 0.1405,
+      "step": 4991
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.774603174603175e-06,
+      "loss": 0.1425,
+      "step": 4992
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.771428571428572e-06,
+      "loss": 0.1341,
+      "step": 4993
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.768253968253969e-06,
+      "loss": 0.1369,
+      "step": 4994
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.765079365079365e-06,
+      "loss": 0.1486,
+      "step": 4995
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 0.1391,
+      "step": 4996
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7587301587301595e-06,
+      "loss": 0.1399,
+      "step": 4997
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.755555555555556e-06,
+      "loss": 0.148,
+      "step": 4998
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.752380952380953e-06,
+      "loss": 0.1463,
+      "step": 4999
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.749206349206349e-06,
+      "loss": 0.1437,
+      "step": 5000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7460317460317465e-06,
+      "loss": 0.141,
+      "step": 5001
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.742857142857144e-06,
+      "loss": 0.1414,
+      "step": 5002
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.73968253968254e-06,
+      "loss": 0.1339,
+      "step": 5003
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.736507936507937e-06,
+      "loss": 0.1516,
+      "step": 5004
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7333333333333335e-06,
+      "loss": 0.1395,
+      "step": 5005
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.730158730158731e-06,
+      "loss": 0.1321,
+      "step": 5006
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.726984126984127e-06,
+      "loss": 0.1412,
+      "step": 5007
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.723809523809524e-06,
+      "loss": 0.1416,
+      "step": 5008
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.720634920634921e-06,
+      "loss": 0.141,
+      "step": 5009
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.717460317460318e-06,
+      "loss": 0.146,
+      "step": 5010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.714285714285715e-06,
+      "loss": 0.1372,
+      "step": 5011
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.711111111111111e-06,
+      "loss": 0.1412,
+      "step": 5012
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.707936507936508e-06,
+      "loss": 0.1401,
+      "step": 5013
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.704761904761905e-06,
+      "loss": 0.1343,
+      "step": 5014
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.701587301587302e-06,
+      "loss": 0.148,
+      "step": 5015
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.698412698412699e-06,
+      "loss": 0.1434,
+      "step": 5016
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.695238095238095e-06,
+      "loss": 0.1443,
+      "step": 5017
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.692063492063492e-06,
+      "loss": 0.1338,
+      "step": 5018
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6888888888888895e-06,
+      "loss": 0.1374,
+      "step": 5019
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.685714285714286e-06,
+      "loss": 0.1426,
+      "step": 5020
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.682539682539683e-06,
+      "loss": 0.1505,
+      "step": 5021
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.679365079365079e-06,
+      "loss": 0.1422,
+      "step": 5022
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6761904761904765e-06,
+      "loss": 0.148,
+      "step": 5023
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.673015873015874e-06,
+      "loss": 0.1381,
+      "step": 5024
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.66984126984127e-06,
+      "loss": 0.1413,
+      "step": 5025
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.147,
+      "step": 5026
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6634920634920635e-06,
+      "loss": 0.1418,
+      "step": 5027
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.660317460317461e-06,
+      "loss": 0.134,
+      "step": 5028
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.657142857142857e-06,
+      "loss": 0.1381,
+      "step": 5029
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.653968253968255e-06,
+      "loss": 0.1352,
+      "step": 5030
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.650793650793651e-06,
+      "loss": 0.1367,
+      "step": 5031
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.647619047619048e-06,
+      "loss": 0.1349,
+      "step": 5032
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.644444444444445e-06,
+      "loss": 0.1405,
+      "step": 5033
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.641269841269841e-06,
+      "loss": 0.1303,
+      "step": 5034
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.638095238095238e-06,
+      "loss": 0.1432,
+      "step": 5035
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.634920634920635e-06,
+      "loss": 0.1438,
+      "step": 5036
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.631746031746033e-06,
+      "loss": 0.1408,
+      "step": 5037
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.628571428571429e-06,
+      "loss": 0.1392,
+      "step": 5038
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.625396825396826e-06,
+      "loss": 0.14,
+      "step": 5039
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.622222222222222e-06,
+      "loss": 0.1369,
+      "step": 5040
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6190476190476196e-06,
+      "loss": 0.133,
+      "step": 5041
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.615873015873017e-06,
+      "loss": 0.1246,
+      "step": 5042
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.612698412698413e-06,
+      "loss": 0.1369,
+      "step": 5043
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.60952380952381e-06,
+      "loss": 0.1336,
+      "step": 5044
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6063492063492065e-06,
+      "loss": 0.139,
+      "step": 5045
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.603174603174604e-06,
+      "loss": 0.1369,
+      "step": 5046
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.1441,
+      "step": 5047
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.596825396825397e-06,
+      "loss": 0.1365,
+      "step": 5048
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.593650793650794e-06,
+      "loss": 0.144,
+      "step": 5049
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.590476190476191e-06,
+      "loss": 0.1354,
+      "step": 5050
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.587301587301588e-06,
+      "loss": 0.1506,
+      "step": 5051
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.584126984126985e-06,
+      "loss": 0.1423,
+      "step": 5052
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.580952380952381e-06,
+      "loss": 0.1365,
+      "step": 5053
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.5777777777777785e-06,
+      "loss": 0.1357,
+      "step": 5054
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.574603174603175e-06,
+      "loss": 0.131,
+      "step": 5055
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 0.1384,
+      "step": 5056
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.568253968253968e-06,
+      "loss": 0.1424,
+      "step": 5057
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5650793650793654e-06,
+      "loss": 0.1463,
+      "step": 5058
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.561904761904763e-06,
+      "loss": 0.1532,
+      "step": 5059
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.558730158730159e-06,
+      "loss": 0.1491,
+      "step": 5060
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.555555555555556e-06,
+      "loss": 0.1311,
+      "step": 5061
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.552380952380952e-06,
+      "loss": 0.1433,
+      "step": 5062
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5492063492063496e-06,
+      "loss": 0.1404,
+      "step": 5063
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.546031746031747e-06,
+      "loss": 0.1345,
+      "step": 5064
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.542857142857143e-06,
+      "loss": 0.1395,
+      "step": 5065
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.53968253968254e-06,
+      "loss": 0.143,
+      "step": 5066
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5365079365079365e-06,
+      "loss": 0.1405,
+      "step": 5067
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 0.1414,
+      "step": 5068
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.530158730158731e-06,
+      "loss": 0.1428,
+      "step": 5069
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.526984126984127e-06,
+      "loss": 0.1375,
+      "step": 5070
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.523809523809524e-06,
+      "loss": 0.1337,
+      "step": 5071
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.520634920634921e-06,
+      "loss": 0.1336,
+      "step": 5072
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.517460317460318e-06,
+      "loss": 0.1432,
+      "step": 5073
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.514285714285714e-06,
+      "loss": 0.1463,
+      "step": 5074
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.511111111111111e-06,
+      "loss": 0.144,
+      "step": 5075
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5079365079365085e-06,
+      "loss": 0.1473,
+      "step": 5076
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.504761904761905e-06,
+      "loss": 0.14,
+      "step": 5077
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.501587301587302e-06,
+      "loss": 0.1362,
+      "step": 5078
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.498412698412698e-06,
+      "loss": 0.1361,
+      "step": 5079
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4952380952380954e-06,
+      "loss": 0.1372,
+      "step": 5080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.492063492063493e-06,
+      "loss": 0.1483,
+      "step": 5081
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.488888888888889e-06,
+      "loss": 0.1361,
+      "step": 5082
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.485714285714286e-06,
+      "loss": 0.1381,
+      "step": 5083
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.482539682539682e-06,
+      "loss": 0.1406,
+      "step": 5084
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4793650793650796e-06,
+      "loss": 0.1482,
+      "step": 5085
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.476190476190477e-06,
+      "loss": 0.1415,
+      "step": 5086
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.473015873015874e-06,
+      "loss": 0.1383,
+      "step": 5087
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.46984126984127e-06,
+      "loss": 0.1381,
+      "step": 5088
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4666666666666665e-06,
+      "loss": 0.1311,
+      "step": 5089
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.463492063492064e-06,
+      "loss": 0.1403,
+      "step": 5090
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.460317460317461e-06,
+      "loss": 0.1494,
+      "step": 5091
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.457142857142858e-06,
+      "loss": 0.1421,
+      "step": 5092
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.453968253968254e-06,
+      "loss": 0.1466,
+      "step": 5093
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4507936507936515e-06,
+      "loss": 0.1438,
+      "step": 5094
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.447619047619048e-06,
+      "loss": 0.1364,
+      "step": 5095
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.1372,
+      "step": 5096
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.441269841269842e-06,
+      "loss": 0.146,
+      "step": 5097
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4380952380952385e-06,
+      "loss": 0.1314,
+      "step": 5098
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.434920634920636e-06,
+      "loss": 0.1418,
+      "step": 5099
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.431746031746032e-06,
+      "loss": 0.1499,
+      "step": 5100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.428571428571429e-06,
+      "loss": 0.135,
+      "step": 5101
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.425396825396826e-06,
+      "loss": 0.1381,
+      "step": 5102
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.422222222222223e-06,
+      "loss": 0.148,
+      "step": 5103
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.41904761904762e-06,
+      "loss": 0.1482,
+      "step": 5104
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.415873015873016e-06,
+      "loss": 0.1438,
+      "step": 5105
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.412698412698413e-06,
+      "loss": 0.139,
+      "step": 5106
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4095238095238096e-06,
+      "loss": 0.1454,
+      "step": 5107
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.406349206349207e-06,
+      "loss": 0.1393,
+      "step": 5108
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.403174603174604e-06,
+      "loss": 0.1401,
+      "step": 5109
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4e-06,
+      "loss": 0.1418,
+      "step": 5110
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.396825396825397e-06,
+      "loss": 0.1434,
+      "step": 5111
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.393650793650794e-06,
+      "loss": 0.1366,
+      "step": 5112
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.390476190476191e-06,
+      "loss": 0.1359,
+      "step": 5113
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.387301587301588e-06,
+      "loss": 0.1392,
+      "step": 5114
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.384126984126984e-06,
+      "loss": 0.1397,
+      "step": 5115
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.3809523809523815e-06,
+      "loss": 0.144,
+      "step": 5116
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.377777777777778e-06,
+      "loss": 0.1396,
+      "step": 5117
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.374603174603175e-06,
+      "loss": 0.1408,
+      "step": 5118
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.371428571428572e-06,
+      "loss": 0.1413,
+      "step": 5119
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.3682539682539685e-06,
+      "loss": 0.1368,
+      "step": 5120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.365079365079366e-06,
+      "loss": 0.1546,
+      "step": 5121
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.361904761904762e-06,
+      "loss": 0.133,
+      "step": 5122
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.358730158730159e-06,
+      "loss": 0.1431,
+      "step": 5123
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3555555555555555e-06,
+      "loss": 0.1376,
+      "step": 5124
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.352380952380953e-06,
+      "loss": 0.132,
+      "step": 5125
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.34920634920635e-06,
+      "loss": 0.1405,
+      "step": 5126
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.346031746031746e-06,
+      "loss": 0.1434,
+      "step": 5127
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.342857142857143e-06,
+      "loss": 0.1391,
+      "step": 5128
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.33968253968254e-06,
+      "loss": 0.1301,
+      "step": 5129
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.336507936507937e-06,
+      "loss": 0.1454,
+      "step": 5130
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 0.1365,
+      "step": 5131
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.33015873015873e-06,
+      "loss": 0.1443,
+      "step": 5132
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.326984126984127e-06,
+      "loss": 0.14,
+      "step": 5133
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.323809523809524e-06,
+      "loss": 0.1385,
+      "step": 5134
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.320634920634921e-06,
+      "loss": 0.1292,
+      "step": 5135
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.317460317460318e-06,
+      "loss": 0.1365,
+      "step": 5136
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.314285714285714e-06,
+      "loss": 0.141,
+      "step": 5137
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3111111111111115e-06,
+      "loss": 0.135,
+      "step": 5138
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.307936507936508e-06,
+      "loss": 0.143,
+      "step": 5139
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.304761904761905e-06,
+      "loss": 0.1271,
+      "step": 5140
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.301587301587302e-06,
+      "loss": 0.1417,
+      "step": 5141
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.298412698412699e-06,
+      "loss": 0.1345,
+      "step": 5142
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.295238095238096e-06,
+      "loss": 0.1363,
+      "step": 5143
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.292063492063492e-06,
+      "loss": 0.1329,
+      "step": 5144
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.288888888888889e-06,
+      "loss": 0.1414,
+      "step": 5145
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.1442,
+      "step": 5146
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2825396825396835e-06,
+      "loss": 0.1366,
+      "step": 5147
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.27936507936508e-06,
+      "loss": 0.1379,
+      "step": 5148
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.276190476190477e-06,
+      "loss": 0.1363,
+      "step": 5149
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.273015873015873e-06,
+      "loss": 0.1357,
+      "step": 5150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.26984126984127e-06,
+      "loss": 0.1346,
+      "step": 5151
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 0.1313,
+      "step": 5152
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.263492063492064e-06,
+      "loss": 0.1545,
+      "step": 5153
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.260317460317461e-06,
+      "loss": 0.1461,
+      "step": 5154
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.257142857142857e-06,
+      "loss": 0.1358,
+      "step": 5155
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2539682539682546e-06,
+      "loss": 0.1419,
+      "step": 5156
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.250793650793651e-06,
+      "loss": 0.1464,
+      "step": 5157
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.247619047619048e-06,
+      "loss": 0.1421,
+      "step": 5158
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.244444444444445e-06,
+      "loss": 0.1455,
+      "step": 5159
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2412698412698415e-06,
+      "loss": 0.1371,
+      "step": 5160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.238095238095239e-06,
+      "loss": 0.137,
+      "step": 5161
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.234920634920635e-06,
+      "loss": 0.1396,
+      "step": 5162
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.231746031746032e-06,
+      "loss": 0.1465,
+      "step": 5163
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.228571428571429e-06,
+      "loss": 0.1309,
+      "step": 5164
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.225396825396826e-06,
+      "loss": 0.1281,
+      "step": 5165
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.222222222222223e-06,
+      "loss": 0.155,
+      "step": 5166
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.219047619047619e-06,
+      "loss": 0.1364,
+      "step": 5167
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.215873015873016e-06,
+      "loss": 0.1331,
+      "step": 5168
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2126984126984135e-06,
+      "loss": 0.1423,
+      "step": 5169
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.20952380952381e-06,
+      "loss": 0.1502,
+      "step": 5170
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.206349206349207e-06,
+      "loss": 0.1404,
+      "step": 5171
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.203174603174603e-06,
+      "loss": 0.1245,
+      "step": 5172
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.1465,
+      "step": 5173
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.196825396825397e-06,
+      "loss": 0.1382,
+      "step": 5174
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.193650793650794e-06,
+      "loss": 0.138,
+      "step": 5175
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.190476190476191e-06,
+      "loss": 0.138,
+      "step": 5176
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.187301587301587e-06,
+      "loss": 0.1364,
+      "step": 5177
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1841269841269846e-06,
+      "loss": 0.1357,
+      "step": 5178
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.180952380952381e-06,
+      "loss": 0.132,
+      "step": 5179
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.177777777777778e-06,
+      "loss": 0.1365,
+      "step": 5180
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.174603174603175e-06,
+      "loss": 0.1337,
+      "step": 5181
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1714285714285715e-06,
+      "loss": 0.1442,
+      "step": 5182
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.168253968253969e-06,
+      "loss": 0.1348,
+      "step": 5183
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.165079365079365e-06,
+      "loss": 0.1354,
+      "step": 5184
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.161904761904762e-06,
+      "loss": 0.1441,
+      "step": 5185
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.158730158730159e-06,
+      "loss": 0.129,
+      "step": 5186
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.155555555555556e-06,
+      "loss": 0.1403,
+      "step": 5187
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.152380952380953e-06,
+      "loss": 0.137,
+      "step": 5188
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.149206349206349e-06,
+      "loss": 0.138,
+      "step": 5189
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.146031746031746e-06,
+      "loss": 0.1425,
+      "step": 5190
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1428571428571435e-06,
+      "loss": 0.1454,
+      "step": 5191
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.13968253968254e-06,
+      "loss": 0.1438,
+      "step": 5192
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.136507936507937e-06,
+      "loss": 0.1442,
+      "step": 5193
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 0.1425,
+      "step": 5194
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1301587301587305e-06,
+      "loss": 0.1384,
+      "step": 5195
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.126984126984127e-06,
+      "loss": 0.1419,
+      "step": 5196
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.123809523809525e-06,
+      "loss": 0.1328,
+      "step": 5197
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.120634920634921e-06,
+      "loss": 0.1432,
+      "step": 5198
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.117460317460317e-06,
+      "loss": 0.1425,
+      "step": 5199
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.114285714285715e-06,
+      "loss": 0.1445,
+      "step": 5200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.111111111111111e-06,
+      "loss": 0.1466,
+      "step": 5201
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.107936507936508e-06,
+      "loss": 0.1505,
+      "step": 5202
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.104761904761905e-06,
+      "loss": 0.1359,
+      "step": 5203
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.101587301587302e-06,
+      "loss": 0.1481,
+      "step": 5204
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.098412698412699e-06,
+      "loss": 0.1271,
+      "step": 5205
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.095238095238096e-06,
+      "loss": 0.1299,
+      "step": 5206
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.092063492063492e-06,
+      "loss": 0.1343,
+      "step": 5207
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.088888888888889e-06,
+      "loss": 0.1402,
+      "step": 5208
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0857142857142865e-06,
+      "loss": 0.1481,
+      "step": 5209
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.082539682539683e-06,
+      "loss": 0.1307,
+      "step": 5210
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.07936507936508e-06,
+      "loss": 0.136,
+      "step": 5211
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.076190476190476e-06,
+      "loss": 0.1359,
+      "step": 5212
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0730158730158735e-06,
+      "loss": 0.1381,
+      "step": 5213
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.069841269841271e-06,
+      "loss": 0.1306,
+      "step": 5214
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.066666666666667e-06,
+      "loss": 0.1372,
+      "step": 5215
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.063492063492064e-06,
+      "loss": 0.1418,
+      "step": 5216
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0603174603174605e-06,
+      "loss": 0.1274,
+      "step": 5217
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.057142857142858e-06,
+      "loss": 0.1348,
+      "step": 5218
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.053968253968255e-06,
+      "loss": 0.143,
+      "step": 5219
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.050793650793651e-06,
+      "loss": 0.1398,
+      "step": 5220
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.047619047619048e-06,
+      "loss": 0.1442,
+      "step": 5221
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.044444444444445e-06,
+      "loss": 0.1285,
+      "step": 5222
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.041269841269842e-06,
+      "loss": 0.1309,
+      "step": 5223
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.038095238095238e-06,
+      "loss": 0.144,
+      "step": 5224
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.034920634920635e-06,
+      "loss": 0.1365,
+      "step": 5225
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.031746031746032e-06,
+      "loss": 0.134,
+      "step": 5226
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.028571428571429e-06,
+      "loss": 0.1363,
+      "step": 5227
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.025396825396826e-06,
+      "loss": 0.1372,
+      "step": 5228
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.022222222222222e-06,
+      "loss": 0.1455,
+      "step": 5229
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.019047619047619e-06,
+      "loss": 0.1342,
+      "step": 5230
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.0158730158730165e-06,
+      "loss": 0.1382,
+      "step": 5231
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.012698412698413e-06,
+      "loss": 0.1361,
+      "step": 5232
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.00952380952381e-06,
+      "loss": 0.1322,
+      "step": 5233
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.006349206349206e-06,
+      "loss": 0.1351,
+      "step": 5234
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.0031746031746035e-06,
+      "loss": 0.146,
+      "step": 5235
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.1312,
+      "step": 5236
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.996825396825397e-06,
+      "loss": 0.1408,
+      "step": 5237
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.993650793650794e-06,
+      "loss": 0.1291,
+      "step": 5238
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9904761904761905e-06,
+      "loss": 0.1384,
+      "step": 5239
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.987301587301588e-06,
+      "loss": 0.1461,
+      "step": 5240
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.984126984126984e-06,
+      "loss": 0.1329,
+      "step": 5241
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.980952380952381e-06,
+      "loss": 0.1375,
+      "step": 5242
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.977777777777778e-06,
+      "loss": 0.1311,
+      "step": 5243
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.974603174603175e-06,
+      "loss": 0.1444,
+      "step": 5244
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.971428571428572e-06,
+      "loss": 0.1365,
+      "step": 5245
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.968253968253968e-06,
+      "loss": 0.1301,
+      "step": 5246
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.965079365079365e-06,
+      "loss": 0.1351,
+      "step": 5247
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.961904761904762e-06,
+      "loss": 0.1544,
+      "step": 5248
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.958730158730159e-06,
+      "loss": 0.1204,
+      "step": 5249
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.955555555555556e-06,
+      "loss": 0.1339,
+      "step": 5250
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.952380952380952e-06,
+      "loss": 0.1402,
+      "step": 5251
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.949206349206349e-06,
+      "loss": 0.1438,
+      "step": 5252
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9460317460317465e-06,
+      "loss": 0.1418,
+      "step": 5253
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.942857142857143e-06,
+      "loss": 0.1369,
+      "step": 5254
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.93968253968254e-06,
+      "loss": 0.1431,
+      "step": 5255
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.936507936507936e-06,
+      "loss": 0.1466,
+      "step": 5256
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9333333333333335e-06,
+      "loss": 0.1455,
+      "step": 5257
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.930158730158731e-06,
+      "loss": 0.1418,
+      "step": 5258
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.926984126984128e-06,
+      "loss": 0.1338,
+      "step": 5259
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.923809523809524e-06,
+      "loss": 0.1311,
+      "step": 5260
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.920634920634921e-06,
+      "loss": 0.1322,
+      "step": 5261
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.917460317460318e-06,
+      "loss": 0.13,
+      "step": 5262
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.914285714285714e-06,
+      "loss": 0.1386,
+      "step": 5263
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.911111111111112e-06,
+      "loss": 0.1458,
+      "step": 5264
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.907936507936508e-06,
+      "loss": 0.1345,
+      "step": 5265
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9047619047619055e-06,
+      "loss": 0.1539,
+      "step": 5266
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.901587301587302e-06,
+      "loss": 0.1314,
+      "step": 5267
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.898412698412699e-06,
+      "loss": 0.1435,
+      "step": 5268
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.895238095238096e-06,
+      "loss": 0.1474,
+      "step": 5269
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.892063492063492e-06,
+      "loss": 0.1405,
+      "step": 5270
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 0.1356,
+      "step": 5271
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.885714285714286e-06,
+      "loss": 0.1464,
+      "step": 5272
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.882539682539683e-06,
+      "loss": 0.1373,
+      "step": 5273
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.879365079365079e-06,
+      "loss": 0.1429,
+      "step": 5274
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8761904761904765e-06,
+      "loss": 0.1454,
+      "step": 5275
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.873015873015874e-06,
+      "loss": 0.1357,
+      "step": 5276
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.86984126984127e-06,
+      "loss": 0.132,
+      "step": 5277
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.866666666666667e-06,
+      "loss": 0.1443,
+      "step": 5278
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8634920634920635e-06,
+      "loss": 0.1414,
+      "step": 5279
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.860317460317461e-06,
+      "loss": 0.1307,
+      "step": 5280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.857142857142858e-06,
+      "loss": 0.1375,
+      "step": 5281
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.853968253968254e-06,
+      "loss": 0.1457,
+      "step": 5282
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.850793650793651e-06,
+      "loss": 0.1342,
+      "step": 5283
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.847619047619048e-06,
+      "loss": 0.1414,
+      "step": 5284
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.844444444444445e-06,
+      "loss": 0.1353,
+      "step": 5285
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.841269841269842e-06,
+      "loss": 0.1352,
+      "step": 5286
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.838095238095238e-06,
+      "loss": 0.1356,
+      "step": 5287
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8349206349206355e-06,
+      "loss": 0.1447,
+      "step": 5288
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.831746031746032e-06,
+      "loss": 0.1325,
+      "step": 5289
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.828571428571429e-06,
+      "loss": 0.1365,
+      "step": 5290
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.825396825396825e-06,
+      "loss": 0.1289,
+      "step": 5291
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8222222222222224e-06,
+      "loss": 0.135,
+      "step": 5292
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.81904761904762e-06,
+      "loss": 0.1417,
+      "step": 5293
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.815873015873016e-06,
+      "loss": 0.1418,
+      "step": 5294
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.812698412698413e-06,
+      "loss": 0.1439,
+      "step": 5295
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 0.1444,
+      "step": 5296
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.806349206349207e-06,
+      "loss": 0.1354,
+      "step": 5297
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8031746031746037e-06,
+      "loss": 0.144,
+      "step": 5298
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.1244,
+      "step": 5299
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.796825396825397e-06,
+      "loss": 0.1375,
+      "step": 5300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.793650793650794e-06,
+      "loss": 0.1472,
+      "step": 5301
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7904761904761907e-06,
+      "loss": 0.1402,
+      "step": 5302
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.787301587301588e-06,
+      "loss": 0.1336,
+      "step": 5303
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7841269841269846e-06,
+      "loss": 0.1386,
+      "step": 5304
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7809523809523813e-06,
+      "loss": 0.1373,
+      "step": 5305
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.777777777777778e-06,
+      "loss": 0.1435,
+      "step": 5306
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.774603174603175e-06,
+      "loss": 0.1334,
+      "step": 5307
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.771428571428572e-06,
+      "loss": 0.1448,
+      "step": 5308
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7682539682539687e-06,
+      "loss": 0.1383,
+      "step": 5309
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7650793650793655e-06,
+      "loss": 0.1313,
+      "step": 5310
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.761904761904762e-06,
+      "loss": 0.1395,
+      "step": 5311
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.758730158730159e-06,
+      "loss": 0.1431,
+      "step": 5312
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7555555555555557e-06,
+      "loss": 0.1288,
+      "step": 5313
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.752380952380953e-06,
+      "loss": 0.1412,
+      "step": 5314
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7492063492063496e-06,
+      "loss": 0.1445,
+      "step": 5315
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7460317460317463e-06,
+      "loss": 0.135,
+      "step": 5316
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.742857142857143e-06,
+      "loss": 0.1249,
+      "step": 5317
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.73968253968254e-06,
+      "loss": 0.1448,
+      "step": 5318
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.736507936507937e-06,
+      "loss": 0.1371,
+      "step": 5319
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 0.1387,
+      "step": 5320
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7301587301587305e-06,
+      "loss": 0.1378,
+      "step": 5321
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7269841269841272e-06,
+      "loss": 0.1389,
+      "step": 5322
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.723809523809524e-06,
+      "loss": 0.1396,
+      "step": 5323
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7206349206349207e-06,
+      "loss": 0.1427,
+      "step": 5324
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.717460317460318e-06,
+      "loss": 0.1447,
+      "step": 5325
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7142857142857146e-06,
+      "loss": 0.144,
+      "step": 5326
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7111111111111113e-06,
+      "loss": 0.1391,
+      "step": 5327
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.707936507936508e-06,
+      "loss": 0.1395,
+      "step": 5328
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.704761904761905e-06,
+      "loss": 0.1447,
+      "step": 5329
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7015873015873016e-06,
+      "loss": 0.132,
+      "step": 5330
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6984126984126987e-06,
+      "loss": 0.1351,
+      "step": 5331
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6952380952380955e-06,
+      "loss": 0.1328,
+      "step": 5332
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6920634920634922e-06,
+      "loss": 0.1496,
+      "step": 5333
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.688888888888889e-06,
+      "loss": 0.132,
+      "step": 5334
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6857142857142857e-06,
+      "loss": 0.1419,
+      "step": 5335
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6825396825396833e-06,
+      "loss": 0.1402,
+      "step": 5336
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6793650793650796e-06,
+      "loss": 0.1331,
+      "step": 5337
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6761904761904763e-06,
+      "loss": 0.1478,
+      "step": 5338
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.673015873015873e-06,
+      "loss": 0.1384,
+      "step": 5339
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.66984126984127e-06,
+      "loss": 0.1383,
+      "step": 5340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 0.1349,
+      "step": 5341
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.663492063492064e-06,
+      "loss": 0.133,
+      "step": 5342
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.660317460317461e-06,
+      "loss": 0.1437,
+      "step": 5343
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6571428571428576e-06,
+      "loss": 0.138,
+      "step": 5344
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.653968253968254e-06,
+      "loss": 0.1262,
+      "step": 5345
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6507936507936507e-06,
+      "loss": 0.1453,
+      "step": 5346
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6476190476190483e-06,
+      "loss": 0.1335,
+      "step": 5347
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.644444444444445e-06,
+      "loss": 0.142,
+      "step": 5348
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6412698412698418e-06,
+      "loss": 0.1334,
+      "step": 5349
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6380952380952385e-06,
+      "loss": 0.1378,
+      "step": 5350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6349206349206353e-06,
+      "loss": 0.1458,
+      "step": 5351
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.631746031746032e-06,
+      "loss": 0.1449,
+      "step": 5352
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.628571428571429e-06,
+      "loss": 0.1328,
+      "step": 5353
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.625396825396826e-06,
+      "loss": 0.1351,
+      "step": 5354
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6222222222222226e-06,
+      "loss": 0.1419,
+      "step": 5355
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6190476190476194e-06,
+      "loss": 0.1361,
+      "step": 5356
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.615873015873016e-06,
+      "loss": 0.1436,
+      "step": 5357
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6126984126984133e-06,
+      "loss": 0.1482,
+      "step": 5358
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.60952380952381e-06,
+      "loss": 0.1363,
+      "step": 5359
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6063492063492068e-06,
+      "loss": 0.1401,
+      "step": 5360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6031746031746035e-06,
+      "loss": 0.1434,
+      "step": 5361
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.1331,
+      "step": 5362
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.596825396825397e-06,
+      "loss": 0.1218,
+      "step": 5363
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.593650793650794e-06,
+      "loss": 0.1394,
+      "step": 5364
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.590476190476191e-06,
+      "loss": 0.1428,
+      "step": 5365
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5873015873015877e-06,
+      "loss": 0.149,
+      "step": 5366
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5841269841269844e-06,
+      "loss": 0.1272,
+      "step": 5367
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.580952380952381e-06,
+      "loss": 0.1322,
+      "step": 5368
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.577777777777778e-06,
+      "loss": 0.1339,
+      "step": 5369
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.574603174603175e-06,
+      "loss": 0.1372,
+      "step": 5370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 0.1369,
+      "step": 5371
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5682539682539685e-06,
+      "loss": 0.1335,
+      "step": 5372
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5650793650793653e-06,
+      "loss": 0.1344,
+      "step": 5373
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.561904761904762e-06,
+      "loss": 0.1424,
+      "step": 5374
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.558730158730159e-06,
+      "loss": 0.1412,
+      "step": 5375
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 0.1324,
+      "step": 5376
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5523809523809527e-06,
+      "loss": 0.1419,
+      "step": 5377
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5492063492063494e-06,
+      "loss": 0.1428,
+      "step": 5378
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.546031746031746e-06,
+      "loss": 0.1333,
+      "step": 5379
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.542857142857143e-06,
+      "loss": 0.1384,
+      "step": 5380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.53968253968254e-06,
+      "loss": 0.1342,
+      "step": 5381
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5365079365079368e-06,
+      "loss": 0.1345,
+      "step": 5382
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5333333333333335e-06,
+      "loss": 0.1594,
+      "step": 5383
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5301587301587303e-06,
+      "loss": 0.1316,
+      "step": 5384
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.526984126984127e-06,
+      "loss": 0.1386,
+      "step": 5385
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.523809523809524e-06,
+      "loss": 0.135,
+      "step": 5386
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.520634920634921e-06,
+      "loss": 0.1321,
+      "step": 5387
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5174603174603177e-06,
+      "loss": 0.1436,
+      "step": 5388
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5142857142857144e-06,
+      "loss": 0.1476,
+      "step": 5389
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.511111111111111e-06,
+      "loss": 0.1421,
+      "step": 5390
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.507936507936508e-06,
+      "loss": 0.147,
+      "step": 5391
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.504761904761905e-06,
+      "loss": 0.1346,
+      "step": 5392
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.501587301587302e-06,
+      "loss": 0.1461,
+      "step": 5393
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4984126984126985e-06,
+      "loss": 0.1456,
+      "step": 5394
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4952380952380953e-06,
+      "loss": 0.1446,
+      "step": 5395
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.492063492063492e-06,
+      "loss": 0.13,
+      "step": 5396
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4888888888888896e-06,
+      "loss": 0.1303,
+      "step": 5397
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4857142857142863e-06,
+      "loss": 0.1439,
+      "step": 5398
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.482539682539683e-06,
+      "loss": 0.1337,
+      "step": 5399
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4793650793650794e-06,
+      "loss": 0.1409,
+      "step": 5400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.476190476190476e-06,
+      "loss": 0.1378,
+      "step": 5401
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.473015873015873e-06,
+      "loss": 0.1374,
+      "step": 5402
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4698412698412705e-06,
+      "loss": 0.1401,
+      "step": 5403
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 0.1464,
+      "step": 5404
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.463492063492064e-06,
+      "loss": 0.1357,
+      "step": 5405
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4603174603174607e-06,
+      "loss": 0.1314,
+      "step": 5406
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4571428571428574e-06,
+      "loss": 0.1337,
+      "step": 5407
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4539682539682546e-06,
+      "loss": 0.1236,
+      "step": 5408
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4507936507936513e-06,
+      "loss": 0.1505,
+      "step": 5409
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.447619047619048e-06,
+      "loss": 0.1336,
+      "step": 5410
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.444444444444445e-06,
+      "loss": 0.1364,
+      "step": 5411
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4412698412698416e-06,
+      "loss": 0.1405,
+      "step": 5412
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4380952380952383e-06,
+      "loss": 0.1461,
+      "step": 5413
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4349206349206355e-06,
+      "loss": 0.1415,
+      "step": 5414
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4317460317460322e-06,
+      "loss": 0.1354,
+      "step": 5415
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 0.1434,
+      "step": 5416
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4253968253968257e-06,
+      "loss": 0.1372,
+      "step": 5417
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4222222222222224e-06,
+      "loss": 0.1406,
+      "step": 5418
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.419047619047619e-06,
+      "loss": 0.141,
+      "step": 5419
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4158730158730164e-06,
+      "loss": 0.1336,
+      "step": 5420
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.412698412698413e-06,
+      "loss": 0.1321,
+      "step": 5421
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.40952380952381e-06,
+      "loss": 0.1417,
+      "step": 5422
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4063492063492066e-06,
+      "loss": 0.135,
+      "step": 5423
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4031746031746033e-06,
+      "loss": 0.1294,
+      "step": 5424
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.135,
+      "step": 5425
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3968253968253972e-06,
+      "loss": 0.1438,
+      "step": 5426
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.393650793650794e-06,
+      "loss": 0.1421,
+      "step": 5427
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3904761904761907e-06,
+      "loss": 0.1419,
+      "step": 5428
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3873015873015874e-06,
+      "loss": 0.1354,
+      "step": 5429
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.384126984126984e-06,
+      "loss": 0.1321,
+      "step": 5430
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3809523809523814e-06,
+      "loss": 0.1419,
+      "step": 5431
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.377777777777778e-06,
+      "loss": 0.1349,
+      "step": 5432
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.374603174603175e-06,
+      "loss": 0.1386,
+      "step": 5433
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3714285714285716e-06,
+      "loss": 0.1281,
+      "step": 5434
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3682539682539683e-06,
+      "loss": 0.14,
+      "step": 5435
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3650793650793655e-06,
+      "loss": 0.1341,
+      "step": 5436
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3619047619047622e-06,
+      "loss": 0.1402,
+      "step": 5437
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.358730158730159e-06,
+      "loss": 0.1317,
+      "step": 5438
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3555555555555557e-06,
+      "loss": 0.1287,
+      "step": 5439
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3523809523809525e-06,
+      "loss": 0.1395,
+      "step": 5440
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.349206349206349e-06,
+      "loss": 0.1296,
+      "step": 5441
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3460317460317464e-06,
+      "loss": 0.143,
+      "step": 5442
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.342857142857143e-06,
+      "loss": 0.1381,
+      "step": 5443
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.33968253968254e-06,
+      "loss": 0.1263,
+      "step": 5444
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3365079365079366e-06,
+      "loss": 0.141,
+      "step": 5445
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.1278,
+      "step": 5446
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.330158730158731e-06,
+      "loss": 0.1358,
+      "step": 5447
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3269841269841272e-06,
+      "loss": 0.1431,
+      "step": 5448
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.323809523809524e-06,
+      "loss": 0.1326,
+      "step": 5449
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3206349206349207e-06,
+      "loss": 0.1359,
+      "step": 5450
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3174603174603175e-06,
+      "loss": 0.1356,
+      "step": 5451
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.314285714285714e-06,
+      "loss": 0.13,
+      "step": 5452
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3111111111111118e-06,
+      "loss": 0.1397,
+      "step": 5453
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3079365079365085e-06,
+      "loss": 0.1365,
+      "step": 5454
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3047619047619053e-06,
+      "loss": 0.1445,
+      "step": 5455
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3015873015873016e-06,
+      "loss": 0.1229,
+      "step": 5456
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2984126984126983e-06,
+      "loss": 0.1295,
+      "step": 5457
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.295238095238095e-06,
+      "loss": 0.1271,
+      "step": 5458
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2920634920634927e-06,
+      "loss": 0.1424,
+      "step": 5459
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2888888888888894e-06,
+      "loss": 0.1257,
+      "step": 5460
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.285714285714286e-06,
+      "loss": 0.1302,
+      "step": 5461
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.282539682539683e-06,
+      "loss": 0.13,
+      "step": 5462
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2793650793650796e-06,
+      "loss": 0.1268,
+      "step": 5463
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.276190476190477e-06,
+      "loss": 0.1377,
+      "step": 5464
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2730158730158735e-06,
+      "loss": 0.1356,
+      "step": 5465
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2698412698412703e-06,
+      "loss": 0.1257,
+      "step": 5466
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.266666666666667e-06,
+      "loss": 0.1425,
+      "step": 5467
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2634920634920638e-06,
+      "loss": 0.1373,
+      "step": 5468
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2603174603174605e-06,
+      "loss": 0.137,
+      "step": 5469
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2571428571428577e-06,
+      "loss": 0.1419,
+      "step": 5470
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2539682539682544e-06,
+      "loss": 0.1421,
+      "step": 5471
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.250793650793651e-06,
+      "loss": 0.1292,
+      "step": 5472
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.247619047619048e-06,
+      "loss": 0.1277,
+      "step": 5473
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2444444444444446e-06,
+      "loss": 0.1486,
+      "step": 5474
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.241269841269842e-06,
+      "loss": 0.1329,
+      "step": 5475
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2380952380952385e-06,
+      "loss": 0.1427,
+      "step": 5476
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2349206349206353e-06,
+      "loss": 0.137,
+      "step": 5477
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.231746031746032e-06,
+      "loss": 0.1317,
+      "step": 5478
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2285714285714288e-06,
+      "loss": 0.1318,
+      "step": 5479
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2253968253968255e-06,
+      "loss": 0.1343,
+      "step": 5480
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2222222222222227e-06,
+      "loss": 0.1314,
+      "step": 5481
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2190476190476194e-06,
+      "loss": 0.1379,
+      "step": 5482
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.215873015873016e-06,
+      "loss": 0.143,
+      "step": 5483
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.212698412698413e-06,
+      "loss": 0.1392,
+      "step": 5484
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2095238095238096e-06,
+      "loss": 0.1405,
+      "step": 5485
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.206349206349207e-06,
+      "loss": 0.1394,
+      "step": 5486
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2031746031746035e-06,
+      "loss": 0.1421,
+      "step": 5487
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.136,
+      "step": 5488
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.196825396825397e-06,
+      "loss": 0.136,
+      "step": 5489
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1936507936507938e-06,
+      "loss": 0.1469,
+      "step": 5490
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1904761904761905e-06,
+      "loss": 0.1431,
+      "step": 5491
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1873015873015877e-06,
+      "loss": 0.138,
+      "step": 5492
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1841269841269844e-06,
+      "loss": 0.1319,
+      "step": 5493
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.180952380952381e-06,
+      "loss": 0.1382,
+      "step": 5494
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.177777777777778e-06,
+      "loss": 0.1365,
+      "step": 5495
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1746031746031746e-06,
+      "loss": 0.1325,
+      "step": 5496
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1714285714285714e-06,
+      "loss": 0.1296,
+      "step": 5497
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1682539682539685e-06,
+      "loss": 0.143,
+      "step": 5498
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1650793650793653e-06,
+      "loss": 0.1346,
+      "step": 5499
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.161904761904762e-06,
+      "loss": 0.1461,
+      "step": 5500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1587301587301588e-06,
+      "loss": 0.1378,
+      "step": 5501
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1555555555555555e-06,
+      "loss": 0.1351,
+      "step": 5502
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1523809523809527e-06,
+      "loss": 0.1385,
+      "step": 5503
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1492063492063494e-06,
+      "loss": 0.1405,
+      "step": 5504
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.146031746031746e-06,
+      "loss": 0.1415,
+      "step": 5505
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.142857142857143e-06,
+      "loss": 0.1413,
+      "step": 5506
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1396825396825396e-06,
+      "loss": 0.1399,
+      "step": 5507
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1365079365079364e-06,
+      "loss": 0.1291,
+      "step": 5508
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.133333333333334e-06,
+      "loss": 0.1359,
+      "step": 5509
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1301587301587307e-06,
+      "loss": 0.1347,
+      "step": 5510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.126984126984127e-06,
+      "loss": 0.1396,
+      "step": 5511
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1238095238095238e-06,
+      "loss": 0.1349,
+      "step": 5512
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1206349206349205e-06,
+      "loss": 0.1478,
+      "step": 5513
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.117460317460318e-06,
+      "loss": 0.1365,
+      "step": 5514
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.114285714285715e-06,
+      "loss": 0.1422,
+      "step": 5515
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1111111111111116e-06,
+      "loss": 0.1372,
+      "step": 5516
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1079365079365083e-06,
+      "loss": 0.1339,
+      "step": 5517
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.104761904761905e-06,
+      "loss": 0.14,
+      "step": 5518
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1015873015873014e-06,
+      "loss": 0.14,
+      "step": 5519
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.098412698412699e-06,
+      "loss": 0.1324,
+      "step": 5520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0952380952380957e-06,
+      "loss": 0.135,
+      "step": 5521
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0920634920634925e-06,
+      "loss": 0.1355,
+      "step": 5522
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.088888888888889e-06,
+      "loss": 0.1413,
+      "step": 5523
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.085714285714286e-06,
+      "loss": 0.133,
+      "step": 5524
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.082539682539683e-06,
+      "loss": 0.1458,
+      "step": 5525
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.07936507936508e-06,
+      "loss": 0.1469,
+      "step": 5526
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0761904761904766e-06,
+      "loss": 0.1333,
+      "step": 5527
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0730158730158733e-06,
+      "loss": 0.1409,
+      "step": 5528
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.06984126984127e-06,
+      "loss": 0.1362,
+      "step": 5529
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.066666666666667e-06,
+      "loss": 0.1308,
+      "step": 5530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.063492063492064e-06,
+      "loss": 0.1407,
+      "step": 5531
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0603174603174607e-06,
+      "loss": 0.131,
+      "step": 5532
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0571428571428575e-06,
+      "loss": 0.1402,
+      "step": 5533
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.053968253968254e-06,
+      "loss": 0.1366,
+      "step": 5534
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.050793650793651e-06,
+      "loss": 0.1331,
+      "step": 5535
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.047619047619048e-06,
+      "loss": 0.1294,
+      "step": 5536
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.044444444444445e-06,
+      "loss": 0.1353,
+      "step": 5537
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0412698412698416e-06,
+      "loss": 0.1346,
+      "step": 5538
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0380952380952383e-06,
+      "loss": 0.1378,
+      "step": 5539
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.034920634920635e-06,
+      "loss": 0.1533,
+      "step": 5540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.031746031746032e-06,
+      "loss": 0.1449,
+      "step": 5541
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.028571428571429e-06,
+      "loss": 0.1549,
+      "step": 5542
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0253968253968257e-06,
+      "loss": 0.1279,
+      "step": 5543
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0222222222222225e-06,
+      "loss": 0.1397,
+      "step": 5544
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.019047619047619e-06,
+      "loss": 0.144,
+      "step": 5545
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.015873015873016e-06,
+      "loss": 0.1358,
+      "step": 5546
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0126984126984127e-06,
+      "loss": 0.1377,
+      "step": 5547
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.00952380952381e-06,
+      "loss": 0.137,
+      "step": 5548
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0063492063492066e-06,
+      "loss": 0.1326,
+      "step": 5549
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0031746031746033e-06,
+      "loss": 0.1485,
+      "step": 5550
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3e-06,
+      "loss": 0.131,
+      "step": 5551
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.996825396825397e-06,
+      "loss": 0.1301,
+      "step": 5552
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.993650793650794e-06,
+      "loss": 0.143,
+      "step": 5553
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9904761904761907e-06,
+      "loss": 0.1291,
+      "step": 5554
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9873015873015875e-06,
+      "loss": 0.1322,
+      "step": 5555
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.984126984126984e-06,
+      "loss": 0.1366,
+      "step": 5556
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.980952380952381e-06,
+      "loss": 0.1371,
+      "step": 5557
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9777777777777777e-06,
+      "loss": 0.1408,
+      "step": 5558
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.974603174603175e-06,
+      "loss": 0.134,
+      "step": 5559
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9714285714285716e-06,
+      "loss": 0.1367,
+      "step": 5560
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9682539682539683e-06,
+      "loss": 0.1318,
+      "step": 5561
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.965079365079365e-06,
+      "loss": 0.1332,
+      "step": 5562
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.961904761904762e-06,
+      "loss": 0.138,
+      "step": 5563
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9587301587301594e-06,
+      "loss": 0.1471,
+      "step": 5564
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.955555555555556e-06,
+      "loss": 0.1376,
+      "step": 5565
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9523809523809525e-06,
+      "loss": 0.1369,
+      "step": 5566
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.949206349206349e-06,
+      "loss": 0.1472,
+      "step": 5567
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.946031746031746e-06,
+      "loss": 0.1327,
+      "step": 5568
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9428571428571427e-06,
+      "loss": 0.1458,
+      "step": 5569
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9396825396825403e-06,
+      "loss": 0.1362,
+      "step": 5570
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.936507936507937e-06,
+      "loss": 0.1475,
+      "step": 5571
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 0.1328,
+      "step": 5572
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9301587301587305e-06,
+      "loss": 0.1535,
+      "step": 5573
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9269841269841272e-06,
+      "loss": 0.1379,
+      "step": 5574
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9238095238095244e-06,
+      "loss": 0.1371,
+      "step": 5575
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.920634920634921e-06,
+      "loss": 0.132,
+      "step": 5576
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.917460317460318e-06,
+      "loss": 0.1356,
+      "step": 5577
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9142857142857146e-06,
+      "loss": 0.137,
+      "step": 5578
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9111111111111114e-06,
+      "loss": 0.1441,
+      "step": 5579
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.907936507936508e-06,
+      "loss": 0.1455,
+      "step": 5580
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9047619047619053e-06,
+      "loss": 0.1281,
+      "step": 5581
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.901587301587302e-06,
+      "loss": 0.1311,
+      "step": 5582
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8984126984126988e-06,
+      "loss": 0.1424,
+      "step": 5583
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8952380952380955e-06,
+      "loss": 0.144,
+      "step": 5584
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8920634920634923e-06,
+      "loss": 0.1257,
+      "step": 5585
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.888888888888889e-06,
+      "loss": 0.1398,
+      "step": 5586
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.885714285714286e-06,
+      "loss": 0.1355,
+      "step": 5587
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.882539682539683e-06,
+      "loss": 0.1273,
+      "step": 5588
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8793650793650796e-06,
+      "loss": 0.1313,
+      "step": 5589
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8761904761904764e-06,
+      "loss": 0.1443,
+      "step": 5590
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.873015873015873e-06,
+      "loss": 0.1306,
+      "step": 5591
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8698412698412703e-06,
+      "loss": 0.1328,
+      "step": 5592
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.866666666666667e-06,
+      "loss": 0.1416,
+      "step": 5593
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8634920634920638e-06,
+      "loss": 0.1353,
+      "step": 5594
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8603174603174605e-06,
+      "loss": 0.135,
+      "step": 5595
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.1386,
+      "step": 5596
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.853968253968254e-06,
+      "loss": 0.1446,
+      "step": 5597
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.850793650793651e-06,
+      "loss": 0.1341,
+      "step": 5598
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.847619047619048e-06,
+      "loss": 0.1276,
+      "step": 5599
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8444444444444446e-06,
+      "loss": 0.1395,
+      "step": 5600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8412698412698414e-06,
+      "loss": 0.1356,
+      "step": 5601
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.838095238095238e-06,
+      "loss": 0.1293,
+      "step": 5602
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8349206349206353e-06,
+      "loss": 0.1435,
+      "step": 5603
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.831746031746032e-06,
+      "loss": 0.1355,
+      "step": 5604
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8285714285714288e-06,
+      "loss": 0.1314,
+      "step": 5605
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8253968253968255e-06,
+      "loss": 0.1467,
+      "step": 5606
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8222222222222223e-06,
+      "loss": 0.1344,
+      "step": 5607
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.819047619047619e-06,
+      "loss": 0.1268,
+      "step": 5608
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.815873015873016e-06,
+      "loss": 0.1299,
+      "step": 5609
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.812698412698413e-06,
+      "loss": 0.1409,
+      "step": 5610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8095238095238096e-06,
+      "loss": 0.1352,
+      "step": 5611
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8063492063492064e-06,
+      "loss": 0.1277,
+      "step": 5612
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.803174603174603e-06,
+      "loss": 0.1257,
+      "step": 5613
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.1215,
+      "step": 5614
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.796825396825397e-06,
+      "loss": 0.1458,
+      "step": 5615
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7936507936507938e-06,
+      "loss": 0.1301,
+      "step": 5616
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7904761904761905e-06,
+      "loss": 0.1455,
+      "step": 5617
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7873015873015873e-06,
+      "loss": 0.1351,
+      "step": 5618
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.784126984126984e-06,
+      "loss": 0.1282,
+      "step": 5619
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7809523809523816e-06,
+      "loss": 0.1327,
+      "step": 5620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 0.134,
+      "step": 5621
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7746031746031747e-06,
+      "loss": 0.1389,
+      "step": 5622
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7714285714285714e-06,
+      "loss": 0.1417,
+      "step": 5623
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.768253968253968e-06,
+      "loss": 0.132,
+      "step": 5624
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.765079365079365e-06,
+      "loss": 0.1426,
+      "step": 5625
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7619047619047625e-06,
+      "loss": 0.1431,
+      "step": 5626
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.758730158730159e-06,
+      "loss": 0.1238,
+      "step": 5627
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.755555555555556e-06,
+      "loss": 0.136,
+      "step": 5628
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7523809523809527e-06,
+      "loss": 0.1302,
+      "step": 5629
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.749206349206349e-06,
+      "loss": 0.1356,
+      "step": 5630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7460317460317466e-06,
+      "loss": 0.1384,
+      "step": 5631
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7428571428571433e-06,
+      "loss": 0.1458,
+      "step": 5632
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.73968253968254e-06,
+      "loss": 0.1286,
+      "step": 5633
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.736507936507937e-06,
+      "loss": 0.1307,
+      "step": 5634
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7333333333333336e-06,
+      "loss": 0.1332,
+      "step": 5635
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7301587301587303e-06,
+      "loss": 0.141,
+      "step": 5636
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7269841269841275e-06,
+      "loss": 0.1248,
+      "step": 5637
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.723809523809524e-06,
+      "loss": 0.1297,
+      "step": 5638
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.720634920634921e-06,
+      "loss": 0.1256,
+      "step": 5639
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7174603174603177e-06,
+      "loss": 0.1346,
+      "step": 5640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7142857142857144e-06,
+      "loss": 0.145,
+      "step": 5641
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7111111111111116e-06,
+      "loss": 0.1381,
+      "step": 5642
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7079365079365083e-06,
+      "loss": 0.1296,
+      "step": 5643
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.704761904761905e-06,
+      "loss": 0.1326,
+      "step": 5644
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.701587301587302e-06,
+      "loss": 0.1435,
+      "step": 5645
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6984126984126986e-06,
+      "loss": 0.1491,
+      "step": 5646
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6952380952380953e-06,
+      "loss": 0.1314,
+      "step": 5647
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6920634920634925e-06,
+      "loss": 0.142,
+      "step": 5648
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6888888888888892e-06,
+      "loss": 0.1402,
+      "step": 5649
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.685714285714286e-06,
+      "loss": 0.133,
+      "step": 5650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6825396825396827e-06,
+      "loss": 0.1426,
+      "step": 5651
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6793650793650794e-06,
+      "loss": 0.1343,
+      "step": 5652
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6761904761904766e-06,
+      "loss": 0.136,
+      "step": 5653
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6730158730158733e-06,
+      "loss": 0.1332,
+      "step": 5654
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.66984126984127e-06,
+      "loss": 0.1331,
+      "step": 5655
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.1394,
+      "step": 5656
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6634920634920636e-06,
+      "loss": 0.1328,
+      "step": 5657
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6603174603174603e-06,
+      "loss": 0.1313,
+      "step": 5658
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6571428571428575e-06,
+      "loss": 0.1286,
+      "step": 5659
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6539682539682542e-06,
+      "loss": 0.1315,
+      "step": 5660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.650793650793651e-06,
+      "loss": 0.1474,
+      "step": 5661
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6476190476190477e-06,
+      "loss": 0.1474,
+      "step": 5662
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6444444444444444e-06,
+      "loss": 0.1363,
+      "step": 5663
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6412698412698416e-06,
+      "loss": 0.1423,
+      "step": 5664
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6380952380952383e-06,
+      "loss": 0.1263,
+      "step": 5665
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.634920634920635e-06,
+      "loss": 0.142,
+      "step": 5666
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.631746031746032e-06,
+      "loss": 0.1379,
+      "step": 5667
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6285714285714286e-06,
+      "loss": 0.1344,
+      "step": 5668
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6253968253968253e-06,
+      "loss": 0.125,
+      "step": 5669
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6222222222222225e-06,
+      "loss": 0.1319,
+      "step": 5670
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6190476190476192e-06,
+      "loss": 0.141,
+      "step": 5671
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.615873015873016e-06,
+      "loss": 0.1387,
+      "step": 5672
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6126984126984127e-06,
+      "loss": 0.1427,
+      "step": 5673
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6095238095238094e-06,
+      "loss": 0.1266,
+      "step": 5674
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.606349206349206e-06,
+      "loss": 0.1281,
+      "step": 5675
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6031746031746038e-06,
+      "loss": 0.1404,
+      "step": 5676
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6e-06,
+      "loss": 0.1333,
+      "step": 5677
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.596825396825397e-06,
+      "loss": 0.1357,
+      "step": 5678
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5936507936507936e-06,
+      "loss": 0.1375,
+      "step": 5679
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5904761904761903e-06,
+      "loss": 0.1365,
+      "step": 5680
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.587301587301588e-06,
+      "loss": 0.1344,
+      "step": 5681
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5841269841269846e-06,
+      "loss": 0.1335,
+      "step": 5682
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5809523809523814e-06,
+      "loss": 0.1313,
+      "step": 5683
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.577777777777778e-06,
+      "loss": 0.1304,
+      "step": 5684
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5746031746031744e-06,
+      "loss": 0.1321,
+      "step": 5685
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": 0.1287,
+      "step": 5686
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5682539682539688e-06,
+      "loss": 0.1435,
+      "step": 5687
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5650793650793655e-06,
+      "loss": 0.132,
+      "step": 5688
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5619047619047623e-06,
+      "loss": 0.1207,
+      "step": 5689
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.558730158730159e-06,
+      "loss": 0.135,
+      "step": 5690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5555555555555557e-06,
+      "loss": 0.1434,
+      "step": 5691
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.552380952380953e-06,
+      "loss": 0.1353,
+      "step": 5692
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5492063492063497e-06,
+      "loss": 0.1449,
+      "step": 5693
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5460317460317464e-06,
+      "loss": 0.1363,
+      "step": 5694
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.542857142857143e-06,
+      "loss": 0.1261,
+      "step": 5695
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.53968253968254e-06,
+      "loss": 0.1315,
+      "step": 5696
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5365079365079366e-06,
+      "loss": 0.1367,
+      "step": 5697
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5333333333333338e-06,
+      "loss": 0.1407,
+      "step": 5698
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5301587301587305e-06,
+      "loss": 0.1307,
+      "step": 5699
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5269841269841273e-06,
+      "loss": 0.1385,
+      "step": 5700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.523809523809524e-06,
+      "loss": 0.1267,
+      "step": 5701
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5206349206349207e-06,
+      "loss": 0.1471,
+      "step": 5702
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.517460317460318e-06,
+      "loss": 0.1407,
+      "step": 5703
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5142857142857147e-06,
+      "loss": 0.1432,
+      "step": 5704
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5111111111111114e-06,
+      "loss": 0.1311,
+      "step": 5705
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.507936507936508e-06,
+      "loss": 0.1333,
+      "step": 5706
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.504761904761905e-06,
+      "loss": 0.1447,
+      "step": 5707
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.5015873015873016e-06,
+      "loss": 0.1438,
+      "step": 5708
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4984126984126984e-06,
+      "loss": 0.1365,
+      "step": 5709
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4952380952380955e-06,
+      "loss": 0.143,
+      "step": 5710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4920634920634923e-06,
+      "loss": 0.1352,
+      "step": 5711
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.488888888888889e-06,
+      "loss": 0.1376,
+      "step": 5712
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.485714285714286e-06,
+      "loss": 0.1304,
+      "step": 5713
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.482539682539683e-06,
+      "loss": 0.1404,
+      "step": 5714
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4793650793650797e-06,
+      "loss": 0.1324,
+      "step": 5715
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4761904761904764e-06,
+      "loss": 0.1341,
+      "step": 5716
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.473015873015873e-06,
+      "loss": 0.138,
+      "step": 5717
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.46984126984127e-06,
+      "loss": 0.1383,
+      "step": 5718
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.466666666666667e-06,
+      "loss": 0.1341,
+      "step": 5719
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.463492063492064e-06,
+      "loss": 0.1406,
+      "step": 5720
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4603174603174605e-06,
+      "loss": 0.1356,
+      "step": 5721
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4571428571428573e-06,
+      "loss": 0.1383,
+      "step": 5722
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.453968253968254e-06,
+      "loss": 0.1378,
+      "step": 5723
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.450793650793651e-06,
+      "loss": 0.1412,
+      "step": 5724
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.447619047619048e-06,
+      "loss": 0.1322,
+      "step": 5725
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4444444444444447e-06,
+      "loss": 0.1264,
+      "step": 5726
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4412698412698414e-06,
+      "loss": 0.1395,
+      "step": 5727
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.438095238095238e-06,
+      "loss": 0.1329,
+      "step": 5728
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.434920634920635e-06,
+      "loss": 0.1423,
+      "step": 5729
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.431746031746032e-06,
+      "loss": 0.1275,
+      "step": 5730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.428571428571429e-06,
+      "loss": 0.1266,
+      "step": 5731
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4253968253968255e-06,
+      "loss": 0.1355,
+      "step": 5732
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4222222222222223e-06,
+      "loss": 0.1298,
+      "step": 5733
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.419047619047619e-06,
+      "loss": 0.1359,
+      "step": 5734
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.415873015873016e-06,
+      "loss": 0.1375,
+      "step": 5735
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.412698412698413e-06,
+      "loss": 0.1373,
+      "step": 5736
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4095238095238097e-06,
+      "loss": 0.1289,
+      "step": 5737
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.406349206349207e-06,
+      "loss": 0.1337,
+      "step": 5738
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4031746031746036e-06,
+      "loss": 0.1361,
+      "step": 5739
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.1366,
+      "step": 5740
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.396825396825397e-06,
+      "loss": 0.1306,
+      "step": 5741
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.393650793650794e-06,
+      "loss": 0.1287,
+      "step": 5742
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3904761904761905e-06,
+      "loss": 0.1299,
+      "step": 5743
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3873015873015877e-06,
+      "loss": 0.1359,
+      "step": 5744
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3841269841269844e-06,
+      "loss": 0.1336,
+      "step": 5745
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.380952380952381e-06,
+      "loss": 0.1233,
+      "step": 5746
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.377777777777778e-06,
+      "loss": 0.1355,
+      "step": 5747
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3746031746031747e-06,
+      "loss": 0.1378,
+      "step": 5748
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.371428571428572e-06,
+      "loss": 0.1348,
+      "step": 5749
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3682539682539686e-06,
+      "loss": 0.1314,
+      "step": 5750
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3650793650793653e-06,
+      "loss": 0.1342,
+      "step": 5751
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.361904761904762e-06,
+      "loss": 0.1345,
+      "step": 5752
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.358730158730159e-06,
+      "loss": 0.137,
+      "step": 5753
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3555555555555555e-06,
+      "loss": 0.1372,
+      "step": 5754
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3523809523809527e-06,
+      "loss": 0.1306,
+      "step": 5755
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3492063492063494e-06,
+      "loss": 0.1342,
+      "step": 5756
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.346031746031746e-06,
+      "loss": 0.1413,
+      "step": 5757
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.342857142857143e-06,
+      "loss": 0.1423,
+      "step": 5758
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3396825396825397e-06,
+      "loss": 0.1371,
+      "step": 5759
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.336507936507937e-06,
+      "loss": 0.1336,
+      "step": 5760
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 0.1338,
+      "step": 5761
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3301587301587303e-06,
+      "loss": 0.1243,
+      "step": 5762
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3269841269841275e-06,
+      "loss": 0.1282,
+      "step": 5763
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.323809523809524e-06,
+      "loss": 0.1373,
+      "step": 5764
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3206349206349205e-06,
+      "loss": 0.1417,
+      "step": 5765
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3174603174603177e-06,
+      "loss": 0.1318,
+      "step": 5766
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3142857142857145e-06,
+      "loss": 0.1371,
+      "step": 5767
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.311111111111111e-06,
+      "loss": 0.1351,
+      "step": 5768
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3079365079365084e-06,
+      "loss": 0.1276,
+      "step": 5769
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.304761904761905e-06,
+      "loss": 0.1364,
+      "step": 5770
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.301587301587302e-06,
+      "loss": 0.1436,
+      "step": 5771
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2984126984126986e-06,
+      "loss": 0.1391,
+      "step": 5772
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2952380952380953e-06,
+      "loss": 0.1387,
+      "step": 5773
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2920634920634925e-06,
+      "loss": 0.1407,
+      "step": 5774
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2888888888888892e-06,
+      "loss": 0.1358,
+      "step": 5775
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 0.1359,
+      "step": 5776
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2825396825396827e-06,
+      "loss": 0.1292,
+      "step": 5777
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2793650793650795e-06,
+      "loss": 0.1473,
+      "step": 5778
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.276190476190476e-06,
+      "loss": 0.1376,
+      "step": 5779
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2730158730158734e-06,
+      "loss": 0.1313,
+      "step": 5780
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.26984126984127e-06,
+      "loss": 0.1322,
+      "step": 5781
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.266666666666667e-06,
+      "loss": 0.1323,
+      "step": 5782
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2634920634920636e-06,
+      "loss": 0.1284,
+      "step": 5783
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2603174603174603e-06,
+      "loss": 0.1402,
+      "step": 5784
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.257142857142857e-06,
+      "loss": 0.1296,
+      "step": 5785
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2539682539682542e-06,
+      "loss": 0.1288,
+      "step": 5786
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.250793650793651e-06,
+      "loss": 0.1355,
+      "step": 5787
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2476190476190477e-06,
+      "loss": 0.1416,
+      "step": 5788
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2444444444444445e-06,
+      "loss": 0.1343,
+      "step": 5789
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.241269841269841e-06,
+      "loss": 0.1294,
+      "step": 5790
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2380952380952384e-06,
+      "loss": 0.1296,
+      "step": 5791
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.234920634920635e-06,
+      "loss": 0.1495,
+      "step": 5792
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.231746031746032e-06,
+      "loss": 0.1316,
+      "step": 5793
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.228571428571429e-06,
+      "loss": 0.1362,
+      "step": 5794
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2253968253968258e-06,
+      "loss": 0.1306,
+      "step": 5795
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 0.1441,
+      "step": 5796
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2190476190476192e-06,
+      "loss": 0.1423,
+      "step": 5797
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.215873015873016e-06,
+      "loss": 0.1324,
+      "step": 5798
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.212698412698413e-06,
+      "loss": 0.1271,
+      "step": 5799
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.20952380952381e-06,
+      "loss": 0.1353,
+      "step": 5800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2063492063492066e-06,
+      "loss": 0.1467,
+      "step": 5801
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2031746031746034e-06,
+      "loss": 0.1326,
+      "step": 5802
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2e-06,
+      "loss": 0.1374,
+      "step": 5803
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.196825396825397e-06,
+      "loss": 0.1315,
+      "step": 5804
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.193650793650794e-06,
+      "loss": 0.1342,
+      "step": 5805
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1904761904761908e-06,
+      "loss": 0.1286,
+      "step": 5806
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1873015873015875e-06,
+      "loss": 0.1408,
+      "step": 5807
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1841269841269842e-06,
+      "loss": 0.1392,
+      "step": 5808
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.180952380952381e-06,
+      "loss": 0.1347,
+      "step": 5809
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1777777777777777e-06,
+      "loss": 0.1395,
+      "step": 5810
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.174603174603175e-06,
+      "loss": 0.1359,
+      "step": 5811
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1714285714285716e-06,
+      "loss": 0.137,
+      "step": 5812
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1682539682539684e-06,
+      "loss": 0.133,
+      "step": 5813
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.165079365079365e-06,
+      "loss": 0.1341,
+      "step": 5814
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.161904761904762e-06,
+      "loss": 0.1475,
+      "step": 5815
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.158730158730159e-06,
+      "loss": 0.1273,
+      "step": 5816
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1555555555555558e-06,
+      "loss": 0.1278,
+      "step": 5817
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1523809523809525e-06,
+      "loss": 0.1426,
+      "step": 5818
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1492063492063497e-06,
+      "loss": 0.1287,
+      "step": 5819
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.146031746031746e-06,
+      "loss": 0.1363,
+      "step": 5820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.131,
+      "step": 5821
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.13968253968254e-06,
+      "loss": 0.1237,
+      "step": 5822
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1365079365079366e-06,
+      "loss": 0.1419,
+      "step": 5823
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 0.1387,
+      "step": 5824
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1301587301587305e-06,
+      "loss": 0.133,
+      "step": 5825
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1269841269841273e-06,
+      "loss": 0.1262,
+      "step": 5826
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.123809523809524e-06,
+      "loss": 0.1334,
+      "step": 5827
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1206349206349208e-06,
+      "loss": 0.1299,
+      "step": 5828
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1174603174603175e-06,
+      "loss": 0.1255,
+      "step": 5829
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1142857142857147e-06,
+      "loss": 0.1367,
+      "step": 5830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1111111111111114e-06,
+      "loss": 0.136,
+      "step": 5831
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.107936507936508e-06,
+      "loss": 0.1309,
+      "step": 5832
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.104761904761905e-06,
+      "loss": 0.1369,
+      "step": 5833
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1015873015873016e-06,
+      "loss": 0.1267,
+      "step": 5834
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0984126984126984e-06,
+      "loss": 0.1357,
+      "step": 5835
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0952380952380955e-06,
+      "loss": 0.1354,
+      "step": 5836
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0920634920634923e-06,
+      "loss": 0.1399,
+      "step": 5837
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.088888888888889e-06,
+      "loss": 0.1439,
+      "step": 5838
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0857142857142858e-06,
+      "loss": 0.1288,
+      "step": 5839
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0825396825396825e-06,
+      "loss": 0.1385,
+      "step": 5840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0793650793650797e-06,
+      "loss": 0.1393,
+      "step": 5841
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0761904761904764e-06,
+      "loss": 0.1403,
+      "step": 5842
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.073015873015873e-06,
+      "loss": 0.1292,
+      "step": 5843
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.06984126984127e-06,
+      "loss": 0.1428,
+      "step": 5844
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0666666666666666e-06,
+      "loss": 0.132,
+      "step": 5845
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0634920634920634e-06,
+      "loss": 0.1364,
+      "step": 5846
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0603174603174606e-06,
+      "loss": 0.1319,
+      "step": 5847
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0571428571428573e-06,
+      "loss": 0.1464,
+      "step": 5848
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.053968253968254e-06,
+      "loss": 0.1371,
+      "step": 5849
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.050793650793651e-06,
+      "loss": 0.1452,
+      "step": 5850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.047619047619048e-06,
+      "loss": 0.1301,
+      "step": 5851
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0444444444444447e-06,
+      "loss": 0.1294,
+      "step": 5852
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0412698412698414e-06,
+      "loss": 0.1354,
+      "step": 5853
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.038095238095238e-06,
+      "loss": 0.1257,
+      "step": 5854
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0349206349206353e-06,
+      "loss": 0.1322,
+      "step": 5855
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.031746031746032e-06,
+      "loss": 0.1297,
+      "step": 5856
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.028571428571429e-06,
+      "loss": 0.1446,
+      "step": 5857
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0253968253968256e-06,
+      "loss": 0.1268,
+      "step": 5858
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0222222222222223e-06,
+      "loss": 0.1368,
+      "step": 5859
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.019047619047619e-06,
+      "loss": 0.1434,
+      "step": 5860
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.015873015873016e-06,
+      "loss": 0.1361,
+      "step": 5861
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.012698412698413e-06,
+      "loss": 0.1434,
+      "step": 5862
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0095238095238097e-06,
+      "loss": 0.1315,
+      "step": 5863
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0063492063492064e-06,
+      "loss": 0.1368,
+      "step": 5864
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.003174603174603e-06,
+      "loss": 0.1328,
+      "step": 5865
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.1327,
+      "step": 5866
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.996825396825397e-06,
+      "loss": 0.1329,
+      "step": 5867
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.993650793650794e-06,
+      "loss": 0.1376,
+      "step": 5868
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9904761904761906e-06,
+      "loss": 0.1304,
+      "step": 5869
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9873015873015873e-06,
+      "loss": 0.126,
+      "step": 5870
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.984126984126984e-06,
+      "loss": 0.1309,
+      "step": 5871
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.980952380952381e-06,
+      "loss": 0.1302,
+      "step": 5872
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.977777777777778e-06,
+      "loss": 0.1342,
+      "step": 5873
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9746031746031747e-06,
+      "loss": 0.1391,
+      "step": 5874
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9714285714285714e-06,
+      "loss": 0.1448,
+      "step": 5875
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.968253968253968e-06,
+      "loss": 0.1327,
+      "step": 5876
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9650793650793653e-06,
+      "loss": 0.1403,
+      "step": 5877
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.961904761904762e-06,
+      "loss": 0.1411,
+      "step": 5878
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.958730158730159e-06,
+      "loss": 0.1349,
+      "step": 5879
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.955555555555556e-06,
+      "loss": 0.1323,
+      "step": 5880
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9523809523809527e-06,
+      "loss": 0.1407,
+      "step": 5881
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9492063492063495e-06,
+      "loss": 0.1313,
+      "step": 5882
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.946031746031746e-06,
+      "loss": 0.1298,
+      "step": 5883
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.942857142857143e-06,
+      "loss": 0.1441,
+      "step": 5884
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9396825396825397e-06,
+      "loss": 0.1339,
+      "step": 5885
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.936507936507937e-06,
+      "loss": 0.1425,
+      "step": 5886
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9333333333333336e-06,
+      "loss": 0.1386,
+      "step": 5887
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9301587301587303e-06,
+      "loss": 0.1218,
+      "step": 5888
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.926984126984127e-06,
+      "loss": 0.1345,
+      "step": 5889
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.923809523809524e-06,
+      "loss": 0.1462,
+      "step": 5890
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.920634920634921e-06,
+      "loss": 0.1311,
+      "step": 5891
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9174603174603177e-06,
+      "loss": 0.1357,
+      "step": 5892
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9142857142857145e-06,
+      "loss": 0.1251,
+      "step": 5893
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9111111111111112e-06,
+      "loss": 0.1495,
+      "step": 5894
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.907936507936508e-06,
+      "loss": 0.1321,
+      "step": 5895
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 0.1382,
+      "step": 5896
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9015873015873019e-06,
+      "loss": 0.1415,
+      "step": 5897
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8984126984126986e-06,
+      "loss": 0.1335,
+      "step": 5898
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8952380952380953e-06,
+      "loss": 0.1359,
+      "step": 5899
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8920634920634923e-06,
+      "loss": 0.1286,
+      "step": 5900
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.888888888888889e-06,
+      "loss": 0.1244,
+      "step": 5901
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.885714285714286e-06,
+      "loss": 0.1384,
+      "step": 5902
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8825396825396827e-06,
+      "loss": 0.1382,
+      "step": 5903
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8793650793650795e-06,
+      "loss": 0.1416,
+      "step": 5904
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8761904761904764e-06,
+      "loss": 0.1367,
+      "step": 5905
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8730158730158732e-06,
+      "loss": 0.1363,
+      "step": 5906
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.86984126984127e-06,
+      "loss": 0.1313,
+      "step": 5907
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 0.1211,
+      "step": 5908
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8634920634920636e-06,
+      "loss": 0.1195,
+      "step": 5909
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8603174603174603e-06,
+      "loss": 0.1322,
+      "step": 5910
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8571428571428573e-06,
+      "loss": 0.1257,
+      "step": 5911
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.853968253968254e-06,
+      "loss": 0.1295,
+      "step": 5912
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8507936507936508e-06,
+      "loss": 0.1265,
+      "step": 5913
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8476190476190477e-06,
+      "loss": 0.1286,
+      "step": 5914
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8444444444444445e-06,
+      "loss": 0.1365,
+      "step": 5915
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8412698412698416e-06,
+      "loss": 0.1366,
+      "step": 5916
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8380952380952382e-06,
+      "loss": 0.1231,
+      "step": 5917
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.834920634920635e-06,
+      "loss": 0.136,
+      "step": 5918
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.831746031746032e-06,
+      "loss": 0.1315,
+      "step": 5919
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8285714285714288e-06,
+      "loss": 0.1362,
+      "step": 5920
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8253968253968254e-06,
+      "loss": 0.1333,
+      "step": 5921
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8222222222222225e-06,
+      "loss": 0.1262,
+      "step": 5922
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8190476190476193e-06,
+      "loss": 0.1361,
+      "step": 5923
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.815873015873016e-06,
+      "loss": 0.1348,
+      "step": 5924
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.812698412698413e-06,
+      "loss": 0.143,
+      "step": 5925
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8095238095238097e-06,
+      "loss": 0.126,
+      "step": 5926
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8063492063492066e-06,
+      "loss": 0.1346,
+      "step": 5927
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8031746031746034e-06,
+      "loss": 0.1398,
+      "step": 5928
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.1308,
+      "step": 5929
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.796825396825397e-06,
+      "loss": 0.1309,
+      "step": 5930
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7936507936507938e-06,
+      "loss": 0.1297,
+      "step": 5931
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7904761904761906e-06,
+      "loss": 0.142,
+      "step": 5932
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7873015873015875e-06,
+      "loss": 0.1334,
+      "step": 5933
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7841269841269843e-06,
+      "loss": 0.129,
+      "step": 5934
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.780952380952381e-06,
+      "loss": 0.1373,
+      "step": 5935
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 0.1429,
+      "step": 5936
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7746031746031747e-06,
+      "loss": 0.1394,
+      "step": 5937
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7714285714285714e-06,
+      "loss": 0.1355,
+      "step": 5938
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7682539682539684e-06,
+      "loss": 0.1462,
+      "step": 5939
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7650793650793651e-06,
+      "loss": 0.1287,
+      "step": 5940
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.761904761904762e-06,
+      "loss": 0.1444,
+      "step": 5941
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7587301587301588e-06,
+      "loss": 0.1387,
+      "step": 5942
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7555555555555556e-06,
+      "loss": 0.1335,
+      "step": 5943
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7523809523809525e-06,
+      "loss": 0.1411,
+      "step": 5944
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7492063492063493e-06,
+      "loss": 0.1416,
+      "step": 5945
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.746031746031746e-06,
+      "loss": 0.1299,
+      "step": 5946
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7428571428571432e-06,
+      "loss": 0.1298,
+      "step": 5947
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7396825396825397e-06,
+      "loss": 0.1359,
+      "step": 5948
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7365079365079364e-06,
+      "loss": 0.1346,
+      "step": 5949
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7333333333333336e-06,
+      "loss": 0.1353,
+      "step": 5950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7301587301587303e-06,
+      "loss": 0.1309,
+      "step": 5951
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7269841269841273e-06,
+      "loss": 0.1394,
+      "step": 5952
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.723809523809524e-06,
+      "loss": 0.1235,
+      "step": 5953
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7206349206349208e-06,
+      "loss": 0.1327,
+      "step": 5954
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7174603174603177e-06,
+      "loss": 0.1396,
+      "step": 5955
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 0.127,
+      "step": 5956
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7111111111111112e-06,
+      "loss": 0.1346,
+      "step": 5957
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7079365079365082e-06,
+      "loss": 0.1315,
+      "step": 5958
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.704761904761905e-06,
+      "loss": 0.1366,
+      "step": 5959
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7015873015873017e-06,
+      "loss": 0.1314,
+      "step": 5960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6984126984126986e-06,
+      "loss": 0.1316,
+      "step": 5961
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6952380952380954e-06,
+      "loss": 0.1382,
+      "step": 5962
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.692063492063492e-06,
+      "loss": 0.1298,
+      "step": 5963
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.688888888888889e-06,
+      "loss": 0.1322,
+      "step": 5964
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6857142857142858e-06,
+      "loss": 0.1252,
+      "step": 5965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6825396825396827e-06,
+      "loss": 0.1337,
+      "step": 5966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6793650793650795e-06,
+      "loss": 0.1385,
+      "step": 5967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6761904761904762e-06,
+      "loss": 0.1297,
+      "step": 5968
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6730158730158732e-06,
+      "loss": 0.137,
+      "step": 5969
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.66984126984127e-06,
+      "loss": 0.1326,
+      "step": 5970
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.1321,
+      "step": 5971
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6634920634920636e-06,
+      "loss": 0.1264,
+      "step": 5972
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6603174603174604e-06,
+      "loss": 0.1387,
+      "step": 5973
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.657142857142857e-06,
+      "loss": 0.1368,
+      "step": 5974
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6539682539682543e-06,
+      "loss": 0.1291,
+      "step": 5975
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6507936507936508e-06,
+      "loss": 0.1329,
+      "step": 5976
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6476190476190475e-06,
+      "loss": 0.1399,
+      "step": 5977
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6444444444444447e-06,
+      "loss": 0.1282,
+      "step": 5978
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6412698412698414e-06,
+      "loss": 0.137,
+      "step": 5979
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6380952380952384e-06,
+      "loss": 0.1317,
+      "step": 5980
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6349206349206351e-06,
+      "loss": 0.1278,
+      "step": 5981
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6317460317460319e-06,
+      "loss": 0.135,
+      "step": 5982
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6285714285714288e-06,
+      "loss": 0.1319,
+      "step": 5983
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6253968253968256e-06,
+      "loss": 0.1309,
+      "step": 5984
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6222222222222223e-06,
+      "loss": 0.1414,
+      "step": 5985
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6190476190476193e-06,
+      "loss": 0.1386,
+      "step": 5986
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.615873015873016e-06,
+      "loss": 0.1305,
+      "step": 5987
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6126984126984127e-06,
+      "loss": 0.1262,
+      "step": 5988
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6095238095238097e-06,
+      "loss": 0.1405,
+      "step": 5989
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6063492063492064e-06,
+      "loss": 0.1365,
+      "step": 5990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6031746031746034e-06,
+      "loss": 0.1347,
+      "step": 5991
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.1341,
+      "step": 5992
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5968253968253969e-06,
+      "loss": 0.1396,
+      "step": 5993
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5936507936507938e-06,
+      "loss": 0.1285,
+      "step": 5994
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5904761904761906e-06,
+      "loss": 0.1321,
+      "step": 5995
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5873015873015873e-06,
+      "loss": 0.1339,
+      "step": 5996
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5841269841269843e-06,
+      "loss": 0.1433,
+      "step": 5997
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.580952380952381e-06,
+      "loss": 0.1312,
+      "step": 5998
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5777777777777778e-06,
+      "loss": 0.1401,
+      "step": 5999
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5746031746031747e-06,
+      "loss": 0.1338,
+      "step": 6000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5714285714285714e-06,
+      "loss": 0.14,
+      "step": 6001
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5682539682539682e-06,
+      "loss": 0.1322,
+      "step": 6002
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5650793650793654e-06,
+      "loss": 0.1394,
+      "step": 6003
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5619047619047619e-06,
+      "loss": 0.1408,
+      "step": 6004
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.558730158730159e-06,
+      "loss": 0.1288,
+      "step": 6005
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5555555555555558e-06,
+      "loss": 0.1376,
+      "step": 6006
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5523809523809525e-06,
+      "loss": 0.1336,
+      "step": 6007
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5492063492063495e-06,
+      "loss": 0.1267,
+      "step": 6008
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5460317460317462e-06,
+      "loss": 0.1306,
+      "step": 6009
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.542857142857143e-06,
+      "loss": 0.1321,
+      "step": 6010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.53968253968254e-06,
+      "loss": 0.1381,
+      "step": 6011
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5365079365079367e-06,
+      "loss": 0.1251,
+      "step": 6012
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5333333333333334e-06,
+      "loss": 0.1347,
+      "step": 6013
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5301587301587304e-06,
+      "loss": 0.1326,
+      "step": 6014
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.526984126984127e-06,
+      "loss": 0.1361,
+      "step": 6015
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.523809523809524e-06,
+      "loss": 0.1412,
+      "step": 6016
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5206349206349208e-06,
+      "loss": 0.1335,
+      "step": 6017
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5174603174603175e-06,
+      "loss": 0.1353,
+      "step": 6018
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5142857142857145e-06,
+      "loss": 0.133,
+      "step": 6019
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5111111111111112e-06,
+      "loss": 0.1395,
+      "step": 6020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.507936507936508e-06,
+      "loss": 0.1319,
+      "step": 6021
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.504761904761905e-06,
+      "loss": 0.1372,
+      "step": 6022
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5015873015873017e-06,
+      "loss": 0.1322,
+      "step": 6023
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4984126984126984e-06,
+      "loss": 0.1318,
+      "step": 6024
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4952380952380954e-06,
+      "loss": 0.1181,
+      "step": 6025
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.492063492063492e-06,
+      "loss": 0.1342,
+      "step": 6026
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4888888888888888e-06,
+      "loss": 0.1316,
+      "step": 6027
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4857142857142858e-06,
+      "loss": 0.1263,
+      "step": 6028
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4825396825396825e-06,
+      "loss": 0.1393,
+      "step": 6029
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4793650793650797e-06,
+      "loss": 0.1405,
+      "step": 6030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4761904761904762e-06,
+      "loss": 0.1406,
+      "step": 6031
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.473015873015873e-06,
+      "loss": 0.1383,
+      "step": 6032
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4698412698412701e-06,
+      "loss": 0.1313,
+      "step": 6033
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4666666666666669e-06,
+      "loss": 0.1293,
+      "step": 6034
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4634920634920636e-06,
+      "loss": 0.1318,
+      "step": 6035
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4603174603174606e-06,
+      "loss": 0.1337,
+      "step": 6036
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4571428571428573e-06,
+      "loss": 0.1377,
+      "step": 6037
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.453968253968254e-06,
+      "loss": 0.1252,
+      "step": 6038
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.450793650793651e-06,
+      "loss": 0.1303,
+      "step": 6039
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4476190476190478e-06,
+      "loss": 0.135,
+      "step": 6040
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4444444444444445e-06,
+      "loss": 0.1383,
+      "step": 6041
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4412698412698414e-06,
+      "loss": 0.1504,
+      "step": 6042
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4380952380952382e-06,
+      "loss": 0.1271,
+      "step": 6043
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4349206349206351e-06,
+      "loss": 0.1381,
+      "step": 6044
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4317460317460319e-06,
+      "loss": 0.1343,
+      "step": 6045
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.1317,
+      "step": 6046
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4253968253968256e-06,
+      "loss": 0.1298,
+      "step": 6047
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4222222222222223e-06,
+      "loss": 0.129,
+      "step": 6048
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.419047619047619e-06,
+      "loss": 0.1145,
+      "step": 6049
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.415873015873016e-06,
+      "loss": 0.1272,
+      "step": 6050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4126984126984128e-06,
+      "loss": 0.1298,
+      "step": 6051
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4095238095238095e-06,
+      "loss": 0.1315,
+      "step": 6052
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4063492063492065e-06,
+      "loss": 0.1373,
+      "step": 6053
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4031746031746032e-06,
+      "loss": 0.1295,
+      "step": 6054
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.1323,
+      "step": 6055
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3968253968253969e-06,
+      "loss": 0.1274,
+      "step": 6056
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3936507936507936e-06,
+      "loss": 0.1402,
+      "step": 6057
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3904761904761908e-06,
+      "loss": 0.1313,
+      "step": 6058
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3873015873015873e-06,
+      "loss": 0.138,
+      "step": 6059
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.384126984126984e-06,
+      "loss": 0.127,
+      "step": 6060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3809523809523812e-06,
+      "loss": 0.1415,
+      "step": 6061
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.377777777777778e-06,
+      "loss": 0.1306,
+      "step": 6062
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3746031746031745e-06,
+      "loss": 0.145,
+      "step": 6063
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3714285714285717e-06,
+      "loss": 0.1332,
+      "step": 6064
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3682539682539684e-06,
+      "loss": 0.1274,
+      "step": 6065
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3650793650793652e-06,
+      "loss": 0.125,
+      "step": 6066
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.361904761904762e-06,
+      "loss": 0.1288,
+      "step": 6067
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3587301587301588e-06,
+      "loss": 0.1279,
+      "step": 6068
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3555555555555558e-06,
+      "loss": 0.1419,
+      "step": 6069
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3523809523809525e-06,
+      "loss": 0.1348,
+      "step": 6070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3492063492063493e-06,
+      "loss": 0.1453,
+      "step": 6071
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3460317460317462e-06,
+      "loss": 0.1388,
+      "step": 6072
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.342857142857143e-06,
+      "loss": 0.1444,
+      "step": 6073
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3396825396825397e-06,
+      "loss": 0.1405,
+      "step": 6074
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3365079365079367e-06,
+      "loss": 0.1346,
+      "step": 6075
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.1236,
+      "step": 6076
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3301587301587302e-06,
+      "loss": 0.1352,
+      "step": 6077
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3269841269841271e-06,
+      "loss": 0.1326,
+      "step": 6078
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3238095238095239e-06,
+      "loss": 0.139,
+      "step": 6079
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3206349206349208e-06,
+      "loss": 0.1275,
+      "step": 6080
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3174603174603175e-06,
+      "loss": 0.1344,
+      "step": 6081
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3142857142857143e-06,
+      "loss": 0.1342,
+      "step": 6082
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3111111111111112e-06,
+      "loss": 0.1359,
+      "step": 6083
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.307936507936508e-06,
+      "loss": 0.1313,
+      "step": 6084
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3047619047619047e-06,
+      "loss": 0.1276,
+      "step": 6085
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3015873015873019e-06,
+      "loss": 0.1488,
+      "step": 6086
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2984126984126984e-06,
+      "loss": 0.1371,
+      "step": 6087
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2952380952380952e-06,
+      "loss": 0.1354,
+      "step": 6088
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2920634920634923e-06,
+      "loss": 0.135,
+      "step": 6089
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.288888888888889e-06,
+      "loss": 0.1272,
+      "step": 6090
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2857142857142856e-06,
+      "loss": 0.1372,
+      "step": 6091
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2825396825396828e-06,
+      "loss": 0.1394,
+      "step": 6092
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2793650793650795e-06,
+      "loss": 0.1383,
+      "step": 6093
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2761904761904765e-06,
+      "loss": 0.1346,
+      "step": 6094
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2730158730158732e-06,
+      "loss": 0.1348,
+      "step": 6095
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.26984126984127e-06,
+      "loss": 0.1304,
+      "step": 6096
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2666666666666669e-06,
+      "loss": 0.1293,
+      "step": 6097
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2634920634920636e-06,
+      "loss": 0.1382,
+      "step": 6098
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2603174603174604e-06,
+      "loss": 0.1376,
+      "step": 6099
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2571428571428573e-06,
+      "loss": 0.1297,
+      "step": 6100
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.253968253968254e-06,
+      "loss": 0.1309,
+      "step": 6101
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2507936507936508e-06,
+      "loss": 0.1359,
+      "step": 6102
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2476190476190478e-06,
+      "loss": 0.146,
+      "step": 6103
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2444444444444445e-06,
+      "loss": 0.1388,
+      "step": 6104
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2412698412698415e-06,
+      "loss": 0.1408,
+      "step": 6105
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2380952380952382e-06,
+      "loss": 0.1316,
+      "step": 6106
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.234920634920635e-06,
+      "loss": 0.1395,
+      "step": 6107
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.231746031746032e-06,
+      "loss": 0.1307,
+      "step": 6108
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2285714285714286e-06,
+      "loss": 0.1403,
+      "step": 6109
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2253968253968256e-06,
+      "loss": 0.1276,
+      "step": 6110
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2222222222222223e-06,
+      "loss": 0.1313,
+      "step": 6111
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.219047619047619e-06,
+      "loss": 0.125,
+      "step": 6112
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.215873015873016e-06,
+      "loss": 0.1242,
+      "step": 6113
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2126984126984128e-06,
+      "loss": 0.1258,
+      "step": 6114
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2095238095238095e-06,
+      "loss": 0.1295,
+      "step": 6115
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2063492063492065e-06,
+      "loss": 0.1321,
+      "step": 6116
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2031746031746034e-06,
+      "loss": 0.1334,
+      "step": 6117
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.1281,
+      "step": 6118
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.196825396825397e-06,
+      "loss": 0.1422,
+      "step": 6119
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1936507936507939e-06,
+      "loss": 0.1415,
+      "step": 6120
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1904761904761906e-06,
+      "loss": 0.1357,
+      "step": 6121
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1873015873015873e-06,
+      "loss": 0.1367,
+      "step": 6122
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1841269841269843e-06,
+      "loss": 0.1314,
+      "step": 6123
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.180952380952381e-06,
+      "loss": 0.1343,
+      "step": 6124
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1777777777777778e-06,
+      "loss": 0.1414,
+      "step": 6125
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1746031746031747e-06,
+      "loss": 0.132,
+      "step": 6126
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1714285714285715e-06,
+      "loss": 0.1283,
+      "step": 6127
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1682539682539684e-06,
+      "loss": 0.1283,
+      "step": 6128
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1650793650793652e-06,
+      "loss": 0.1355,
+      "step": 6129
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.161904761904762e-06,
+      "loss": 0.1314,
+      "step": 6130
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1587301587301589e-06,
+      "loss": 0.1297,
+      "step": 6131
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1555555555555556e-06,
+      "loss": 0.1404,
+      "step": 6132
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1523809523809526e-06,
+      "loss": 0.1209,
+      "step": 6133
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1492063492063493e-06,
+      "loss": 0.1271,
+      "step": 6134
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1460317460317462e-06,
+      "loss": 0.1312,
+      "step": 6135
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 0.1345,
+      "step": 6136
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1396825396825397e-06,
+      "loss": 0.1333,
+      "step": 6137
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1365079365079367e-06,
+      "loss": 0.1335,
+      "step": 6138
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1333333333333334e-06,
+      "loss": 0.1336,
+      "step": 6139
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1301587301587302e-06,
+      "loss": 0.1401,
+      "step": 6140
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1269841269841271e-06,
+      "loss": 0.1371,
+      "step": 6141
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1238095238095239e-06,
+      "loss": 0.1354,
+      "step": 6142
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1206349206349206e-06,
+      "loss": 0.1283,
+      "step": 6143
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1174603174603176e-06,
+      "loss": 0.1328,
+      "step": 6144
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1142857142857145e-06,
+      "loss": 0.1402,
+      "step": 6145
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 0.1303,
+      "step": 6146
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.107936507936508e-06,
+      "loss": 0.1293,
+      "step": 6147
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.104761904761905e-06,
+      "loss": 0.1337,
+      "step": 6148
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1015873015873017e-06,
+      "loss": 0.143,
+      "step": 6149
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0984126984126984e-06,
+      "loss": 0.1301,
+      "step": 6150
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0952380952380954e-06,
+      "loss": 0.1299,
+      "step": 6151
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0920634920634921e-06,
+      "loss": 0.1344,
+      "step": 6152
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0888888888888889e-06,
+      "loss": 0.1436,
+      "step": 6153
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0857142857142858e-06,
+      "loss": 0.1205,
+      "step": 6154
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0825396825396826e-06,
+      "loss": 0.1289,
+      "step": 6155
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0793650793650795e-06,
+      "loss": 0.1418,
+      "step": 6156
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0761904761904763e-06,
+      "loss": 0.14,
+      "step": 6157
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.073015873015873e-06,
+      "loss": 0.1374,
+      "step": 6158
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.06984126984127e-06,
+      "loss": 0.1294,
+      "step": 6159
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 0.1304,
+      "step": 6160
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0634920634920636e-06,
+      "loss": 0.1335,
+      "step": 6161
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0603174603174604e-06,
+      "loss": 0.1254,
+      "step": 6162
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0571428571428573e-06,
+      "loss": 0.1328,
+      "step": 6163
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.053968253968254e-06,
+      "loss": 0.137,
+      "step": 6164
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0507936507936508e-06,
+      "loss": 0.1288,
+      "step": 6165
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0476190476190478e-06,
+      "loss": 0.1254,
+      "step": 6166
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0444444444444445e-06,
+      "loss": 0.143,
+      "step": 6167
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0412698412698413e-06,
+      "loss": 0.1374,
+      "step": 6168
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0380952380952382e-06,
+      "loss": 0.1342,
+      "step": 6169
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.034920634920635e-06,
+      "loss": 0.1412,
+      "step": 6170
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0317460317460317e-06,
+      "loss": 0.1293,
+      "step": 6171
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0285714285714286e-06,
+      "loss": 0.1373,
+      "step": 6172
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0253968253968256e-06,
+      "loss": 0.1428,
+      "step": 6173
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0222222222222223e-06,
+      "loss": 0.1265,
+      "step": 6174
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.019047619047619e-06,
+      "loss": 0.1343,
+      "step": 6175
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.015873015873016e-06,
+      "loss": 0.1345,
+      "step": 6176
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0126984126984128e-06,
+      "loss": 0.1298,
+      "step": 6177
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0095238095238095e-06,
+      "loss": 0.1357,
+      "step": 6178
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0063492063492065e-06,
+      "loss": 0.1443,
+      "step": 6179
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0031746031746032e-06,
+      "loss": 0.1357,
+      "step": 6180
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.1282,
+      "step": 6181
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.96825396825397e-07,
+      "loss": 0.1263,
+      "step": 6182
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.936507936507936e-07,
+      "loss": 0.1271,
+      "step": 6183
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.904761904761906e-07,
+      "loss": 0.1357,
+      "step": 6184
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.873015873015873e-07,
+      "loss": 0.1426,
+      "step": 6185
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.84126984126984e-07,
+      "loss": 0.1293,
+      "step": 6186
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.80952380952381e-07,
+      "loss": 0.1268,
+      "step": 6187
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.77777777777778e-07,
+      "loss": 0.1404,
+      "step": 6188
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.746031746031747e-07,
+      "loss": 0.1326,
+      "step": 6189
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.714285714285715e-07,
+      "loss": 0.1312,
+      "step": 6190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.682539682539684e-07,
+      "loss": 0.1294,
+      "step": 6191
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.650793650793652e-07,
+      "loss": 0.1306,
+      "step": 6192
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.61904761904762e-07,
+      "loss": 0.1492,
+      "step": 6193
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.587301587301589e-07,
+      "loss": 0.1469,
+      "step": 6194
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.555555555555556e-07,
+      "loss": 0.1187,
+      "step": 6195
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 0.1214,
+      "step": 6196
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.492063492063493e-07,
+      "loss": 0.133,
+      "step": 6197
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.460317460317461e-07,
+      "loss": 0.1392,
+      "step": 6198
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.42857142857143e-07,
+      "loss": 0.1287,
+      "step": 6199
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.396825396825397e-07,
+      "loss": 0.133,
+      "step": 6200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.365079365079366e-07,
+      "loss": 0.1324,
+      "step": 6201
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.333333333333334e-07,
+      "loss": 0.1304,
+      "step": 6202
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.301587301587302e-07,
+      "loss": 0.1351,
+      "step": 6203
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.26984126984127e-07,
+      "loss": 0.1314,
+      "step": 6204
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.238095238095239e-07,
+      "loss": 0.1239,
+      "step": 6205
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.206349206349208e-07,
+      "loss": 0.1429,
+      "step": 6206
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.174603174603175e-07,
+      "loss": 0.1307,
+      "step": 6207
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.142857142857144e-07,
+      "loss": 0.1286,
+      "step": 6208
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.111111111111113e-07,
+      "loss": 0.1302,
+      "step": 6209
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.07936507936508e-07,
+      "loss": 0.1361,
+      "step": 6210
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.047619047619048e-07,
+      "loss": 0.1284,
+      "step": 6211
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.015873015873017e-07,
+      "loss": 0.1301,
+      "step": 6212
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.984126984126985e-07,
+      "loss": 0.1342,
+      "step": 6213
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.952380952380953e-07,
+      "loss": 0.1277,
+      "step": 6214
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.920634920634921e-07,
+      "loss": 0.1277,
+      "step": 6215
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 0.1397,
+      "step": 6216
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.857142857142857e-07,
+      "loss": 0.1277,
+      "step": 6217
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.825396825396826e-07,
+      "loss": 0.1365,
+      "step": 6218
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.793650793650794e-07,
+      "loss": 0.1396,
+      "step": 6219
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.761904761904763e-07,
+      "loss": 0.1407,
+      "step": 6220
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.73015873015873e-07,
+      "loss": 0.1385,
+      "step": 6221
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.698412698412699e-07,
+      "loss": 0.121,
+      "step": 6222
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.666666666666668e-07,
+      "loss": 0.1368,
+      "step": 6223
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.634920634920637e-07,
+      "loss": 0.1349,
+      "step": 6224
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.603174603174604e-07,
+      "loss": 0.1364,
+      "step": 6225
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.571428571428572e-07,
+      "loss": 0.1304,
+      "step": 6226
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.539682539682541e-07,
+      "loss": 0.1345,
+      "step": 6227
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.507936507936508e-07,
+      "loss": 0.1332,
+      "step": 6228
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.476190476190477e-07,
+      "loss": 0.1324,
+      "step": 6229
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.444444444444445e-07,
+      "loss": 0.1423,
+      "step": 6230
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.412698412698414e-07,
+      "loss": 0.1388,
+      "step": 6231
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.380952380952381e-07,
+      "loss": 0.1219,
+      "step": 6232
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.34920634920635e-07,
+      "loss": 0.1357,
+      "step": 6233
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.317460317460318e-07,
+      "loss": 0.1432,
+      "step": 6234
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.285714285714285e-07,
+      "loss": 0.1309,
+      "step": 6235
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.253968253968254e-07,
+      "loss": 0.137,
+      "step": 6236
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.222222222222223e-07,
+      "loss": 0.1338,
+      "step": 6237
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.190476190476192e-07,
+      "loss": 0.124,
+      "step": 6238
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.158730158730159e-07,
+      "loss": 0.1357,
+      "step": 6239
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.126984126984128e-07,
+      "loss": 0.1275,
+      "step": 6240
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.095238095238096e-07,
+      "loss": 0.1229,
+      "step": 6241
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.063492063492064e-07,
+      "loss": 0.1326,
+      "step": 6242
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.031746031746032e-07,
+      "loss": 0.1341,
+      "step": 6243
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.132,
+      "step": 6244
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.968253968253969e-07,
+      "loss": 0.1341,
+      "step": 6245
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.936507936507937e-07,
+      "loss": 0.1357,
+      "step": 6246
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.904761904761905e-07,
+      "loss": 0.1374,
+      "step": 6247
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.873015873015874e-07,
+      "loss": 0.1208,
+      "step": 6248
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.841269841269841e-07,
+      "loss": 0.1238,
+      "step": 6249
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.809523809523809e-07,
+      "loss": 0.1354,
+      "step": 6250
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.777777777777779e-07,
+      "loss": 0.1301,
+      "step": 6251
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.746031746031747e-07,
+      "loss": 0.1297,
+      "step": 6252
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.714285714285715e-07,
+      "loss": 0.1307,
+      "step": 6253
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.682539682539683e-07,
+      "loss": 0.1349,
+      "step": 6254
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.650793650793652e-07,
+      "loss": 0.126,
+      "step": 6255
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.61904761904762e-07,
+      "loss": 0.1253,
+      "step": 6256
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.587301587301588e-07,
+      "loss": 0.1343,
+      "step": 6257
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.555555555555556e-07,
+      "loss": 0.1265,
+      "step": 6258
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.523809523809525e-07,
+      "loss": 0.1437,
+      "step": 6259
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.492063492063492e-07,
+      "loss": 0.1423,
+      "step": 6260
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.46031746031746e-07,
+      "loss": 0.1316,
+      "step": 6261
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.428571428571429e-07,
+      "loss": 0.1352,
+      "step": 6262
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.396825396825399e-07,
+      "loss": 0.1471,
+      "step": 6263
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.365079365079365e-07,
+      "loss": 0.1293,
+      "step": 6264
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.333333333333334e-07,
+      "loss": 0.1316,
+      "step": 6265
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.301587301587303e-07,
+      "loss": 0.1261,
+      "step": 6266
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.26984126984127e-07,
+      "loss": 0.1332,
+      "step": 6267
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.238095238095239e-07,
+      "loss": 0.1236,
+      "step": 6268
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.206349206349207e-07,
+      "loss": 0.128,
+      "step": 6269
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.174603174603176e-07,
+      "loss": 0.1339,
+      "step": 6270
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 0.1301,
+      "step": 6271
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.111111111111112e-07,
+      "loss": 0.1293,
+      "step": 6272
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.07936507936508e-07,
+      "loss": 0.1323,
+      "step": 6273
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.047619047619048e-07,
+      "loss": 0.1275,
+      "step": 6274
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.015873015873016e-07,
+      "loss": 0.1285,
+      "step": 6275
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.984126984126984e-07,
+      "loss": 0.1268,
+      "step": 6276
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.952380952380954e-07,
+      "loss": 0.127,
+      "step": 6277
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.92063492063492e-07,
+      "loss": 0.1269,
+      "step": 6278
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.88888888888889e-07,
+      "loss": 0.1453,
+      "step": 6279
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.857142857142858e-07,
+      "loss": 0.1246,
+      "step": 6280
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.825396825396826e-07,
+      "loss": 0.1331,
+      "step": 6281
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.793650793650794e-07,
+      "loss": 0.1306,
+      "step": 6282
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.761904761904763e-07,
+      "loss": 0.1302,
+      "step": 6283
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.730158730158731e-07,
+      "loss": 0.1203,
+      "step": 6284
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.698412698412699e-07,
+      "loss": 0.1294,
+      "step": 6285
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 0.1335,
+      "step": 6286
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.634920634920636e-07,
+      "loss": 0.1324,
+      "step": 6287
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.603174603174604e-07,
+      "loss": 0.1332,
+      "step": 6288
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.571428571428571e-07,
+      "loss": 0.1397,
+      "step": 6289
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.53968253968254e-07,
+      "loss": 0.1353,
+      "step": 6290
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.507936507936509e-07,
+      "loss": 0.1269,
+      "step": 6291
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.476190476190476e-07,
+      "loss": 0.1292,
+      "step": 6292
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.444444444444445e-07,
+      "loss": 0.1338,
+      "step": 6293
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.412698412698414e-07,
+      "loss": 0.1365,
+      "step": 6294
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.380952380952382e-07,
+      "loss": 0.1307,
+      "step": 6295
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.34920634920635e-07,
+      "loss": 0.1332,
+      "step": 6296
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.317460317460318e-07,
+      "loss": 0.1345,
+      "step": 6297
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.285714285714287e-07,
+      "loss": 0.1179,
+      "step": 6298
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.253968253968254e-07,
+      "loss": 0.132,
+      "step": 6299
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.222222222222223e-07,
+      "loss": 0.1351,
+      "step": 6300
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.190476190476191e-07,
+      "loss": 0.135,
+      "step": 6301
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.15873015873016e-07,
+      "loss": 0.1295,
+      "step": 6302
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.126984126984128e-07,
+      "loss": 0.124,
+      "step": 6303
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.095238095238095e-07,
+      "loss": 0.1271,
+      "step": 6304
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.063492063492064e-07,
+      "loss": 0.1158,
+      "step": 6305
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.031746031746032e-07,
+      "loss": 0.1275,
+      "step": 6306
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 0.1256,
+      "step": 6307
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.968253968253969e-07,
+      "loss": 0.1337,
+      "step": 6308
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.936507936507937e-07,
+      "loss": 0.1251,
+      "step": 6309
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.904761904761905e-07,
+      "loss": 0.131,
+      "step": 6310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.873015873015874e-07,
+      "loss": 0.1247,
+      "step": 6311
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.841269841269842e-07,
+      "loss": 0.1278,
+      "step": 6312
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.80952380952381e-07,
+      "loss": 0.1379,
+      "step": 6313
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.777777777777778e-07,
+      "loss": 0.1388,
+      "step": 6314
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.746031746031746e-07,
+      "loss": 0.1297,
+      "step": 6315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 0.1294,
+      "step": 6316
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.682539682539683e-07,
+      "loss": 0.138,
+      "step": 6317
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.650793650793651e-07,
+      "loss": 0.1357,
+      "step": 6318
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.619047619047619e-07,
+      "loss": 0.1374,
+      "step": 6319
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.587301587301588e-07,
+      "loss": 0.1338,
+      "step": 6320
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 0.1267,
+      "step": 6321
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.523809523809525e-07,
+      "loss": 0.1357,
+      "step": 6322
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.492063492063492e-07,
+      "loss": 0.1325,
+      "step": 6323
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.460317460317461e-07,
+      "loss": 0.1344,
+      "step": 6324
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.428571428571429e-07,
+      "loss": 0.1272,
+      "step": 6325
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.396825396825398e-07,
+      "loss": 0.1306,
+      "step": 6326
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.365079365079365e-07,
+      "loss": 0.1407,
+      "step": 6327
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 0.1247,
+      "step": 6328
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.301587301587302e-07,
+      "loss": 0.131,
+      "step": 6329
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.26984126984127e-07,
+      "loss": 0.1335,
+      "step": 6330
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.238095238095239e-07,
+      "loss": 0.1326,
+      "step": 6331
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.206349206349206e-07,
+      "loss": 0.1301,
+      "step": 6332
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.174603174603175e-07,
+      "loss": 0.1152,
+      "step": 6333
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.142857142857143e-07,
+      "loss": 0.1319,
+      "step": 6334
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.111111111111112e-07,
+      "loss": 0.1268,
+      "step": 6335
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.07936507936508e-07,
+      "loss": 0.1255,
+      "step": 6336
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.047619047619048e-07,
+      "loss": 0.139,
+      "step": 6337
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.015873015873016e-07,
+      "loss": 0.1383,
+      "step": 6338
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.984126984126985e-07,
+      "loss": 0.1306,
+      "step": 6339
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.952380952380953e-07,
+      "loss": 0.1296,
+      "step": 6340
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.92063492063492e-07,
+      "loss": 0.1326,
+      "step": 6341
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.88888888888889e-07,
+      "loss": 0.1301,
+      "step": 6342
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.857142857142857e-07,
+      "loss": 0.1246,
+      "step": 6343
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.825396825396826e-07,
+      "loss": 0.1189,
+      "step": 6344
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.793650793650794e-07,
+      "loss": 0.1411,
+      "step": 6345
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7619047619047623e-07,
+      "loss": 0.1267,
+      "step": 6346
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.730158730158731e-07,
+      "loss": 0.1216,
+      "step": 6347
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.6984126984126987e-07,
+      "loss": 0.1317,
+      "step": 6348
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.666666666666667e-07,
+      "loss": 0.1271,
+      "step": 6349
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.634920634920635e-07,
+      "loss": 0.1278,
+      "step": 6350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.603174603174604e-07,
+      "loss": 0.1301,
+      "step": 6351
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.571428571428572e-07,
+      "loss": 0.1306,
+      "step": 6352
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.53968253968254e-07,
+      "loss": 0.1371,
+      "step": 6353
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5079365079365085e-07,
+      "loss": 0.1224,
+      "step": 6354
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.4761904761904764e-07,
+      "loss": 0.1364,
+      "step": 6355
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 0.1298,
+      "step": 6356
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.412698412698413e-07,
+      "loss": 0.1247,
+      "step": 6357
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.3809523809523813e-07,
+      "loss": 0.1315,
+      "step": 6358
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.349206349206349e-07,
+      "loss": 0.1268,
+      "step": 6359
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.317460317460318e-07,
+      "loss": 0.1294,
+      "step": 6360
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.285714285714286e-07,
+      "loss": 0.1195,
+      "step": 6361
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.253968253968254e-07,
+      "loss": 0.1325,
+      "step": 6362
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2222222222222226e-07,
+      "loss": 0.1217,
+      "step": 6363
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1904761904761906e-07,
+      "loss": 0.1248,
+      "step": 6364
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.158730158730159e-07,
+      "loss": 0.128,
+      "step": 6365
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.126984126984127e-07,
+      "loss": 0.1372,
+      "step": 6366
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.095238095238096e-07,
+      "loss": 0.1344,
+      "step": 6367
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.063492063492064e-07,
+      "loss": 0.1298,
+      "step": 6368
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.031746031746032e-07,
+      "loss": 0.1244,
+      "step": 6369
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.1299,
+      "step": 6370
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9682539682539683e-07,
+      "loss": 0.1245,
+      "step": 6371
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.936507936507937e-07,
+      "loss": 0.1313,
+      "step": 6372
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9047619047619047e-07,
+      "loss": 0.1272,
+      "step": 6373
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8730158730158737e-07,
+      "loss": 0.1281,
+      "step": 6374
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8412698412698417e-07,
+      "loss": 0.1283,
+      "step": 6375
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.80952380952381e-07,
+      "loss": 0.1339,
+      "step": 6376
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.777777777777778e-07,
+      "loss": 0.1344,
+      "step": 6377
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.746031746031746e-07,
+      "loss": 0.1318,
+      "step": 6378
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7142857142857145e-07,
+      "loss": 0.133,
+      "step": 6379
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6825396825396824e-07,
+      "loss": 0.1292,
+      "step": 6380
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6507936507936514e-07,
+      "loss": 0.1324,
+      "step": 6381
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6190476190476194e-07,
+      "loss": 0.1352,
+      "step": 6382
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.587301587301588e-07,
+      "loss": 0.115,
+      "step": 6383
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.555555555555556e-07,
+      "loss": 0.1274,
+      "step": 6384
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.523809523809524e-07,
+      "loss": 0.1276,
+      "step": 6385
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.492063492063492e-07,
+      "loss": 0.13,
+      "step": 6386
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.46031746031746e-07,
+      "loss": 0.1372,
+      "step": 6387
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.428571428571429e-07,
+      "loss": 0.1253,
+      "step": 6388
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.396825396825397e-07,
+      "loss": 0.1205,
+      "step": 6389
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3650793650793656e-07,
+      "loss": 0.1252,
+      "step": 6390
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3333333333333335e-07,
+      "loss": 0.135,
+      "step": 6391
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.301587301587302e-07,
+      "loss": 0.1229,
+      "step": 6392
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.26984126984127e-07,
+      "loss": 0.1298,
+      "step": 6393
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.238095238095238e-07,
+      "loss": 0.1368,
+      "step": 6394
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.206349206349207e-07,
+      "loss": 0.1401,
+      "step": 6395
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.174603174603175e-07,
+      "loss": 0.1264,
+      "step": 6396
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1428571428571433e-07,
+      "loss": 0.1137,
+      "step": 6397
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.111111111111111e-07,
+      "loss": 0.1304,
+      "step": 6398
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.07936507936508e-07,
+      "loss": 0.1402,
+      "step": 6399
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0476190476190477e-07,
+      "loss": 0.1298,
+      "step": 6400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.015873015873016e-07,
+      "loss": 0.1357,
+      "step": 6401
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9841269841269846e-07,
+      "loss": 0.1513,
+      "step": 6402
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9523809523809526e-07,
+      "loss": 0.1325,
+      "step": 6403
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.920634920634921e-07,
+      "loss": 0.1393,
+      "step": 6404
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.888888888888889e-07,
+      "loss": 0.1332,
+      "step": 6405
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.8571428571428575e-07,
+      "loss": 0.1354,
+      "step": 6406
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.8253968253968254e-07,
+      "loss": 0.1374,
+      "step": 6407
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.793650793650794e-07,
+      "loss": 0.1479,
+      "step": 6408
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7619047619047624e-07,
+      "loss": 0.1264,
+      "step": 6409
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7301587301587303e-07,
+      "loss": 0.128,
+      "step": 6410
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.698412698412699e-07,
+      "loss": 0.1336,
+      "step": 6411
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 0.1324,
+      "step": 6412
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.634920634920635e-07,
+      "loss": 0.1381,
+      "step": 6413
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.603174603174603e-07,
+      "loss": 0.1334,
+      "step": 6414
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.5714285714285716e-07,
+      "loss": 0.1246,
+      "step": 6415
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.53968253968254e-07,
+      "loss": 0.127,
+      "step": 6416
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.507936507936508e-07,
+      "loss": 0.1251,
+      "step": 6417
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4761904761904765e-07,
+      "loss": 0.1361,
+      "step": 6418
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.444444444444445e-07,
+      "loss": 0.1414,
+      "step": 6419
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.412698412698413e-07,
+      "loss": 0.1294,
+      "step": 6420
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3809523809523811e-07,
+      "loss": 0.1353,
+      "step": 6421
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3492063492063493e-07,
+      "loss": 0.122,
+      "step": 6422
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3174603174603176e-07,
+      "loss": 0.1487,
+      "step": 6423
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.285714285714286e-07,
+      "loss": 0.1245,
+      "step": 6424
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2539682539682542e-07,
+      "loss": 0.1287,
+      "step": 6425
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2222222222222224e-07,
+      "loss": 0.126,
+      "step": 6426
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1904761904761907e-07,
+      "loss": 0.1376,
+      "step": 6427
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.158730158730159e-07,
+      "loss": 0.1245,
+      "step": 6428
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.126984126984127e-07,
+      "loss": 0.1325,
+      "step": 6429
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0952380952380953e-07,
+      "loss": 0.1376,
+      "step": 6430
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0634920634920635e-07,
+      "loss": 0.1339,
+      "step": 6431
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.031746031746032e-07,
+      "loss": 0.1339,
+      "step": 6432
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.1288,
+      "step": 6433
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9682539682539684e-07,
+      "loss": 0.116,
+      "step": 6434
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9365079365079369e-07,
+      "loss": 0.1348,
+      "step": 6435
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.904761904761905e-07,
+      "loss": 0.1432,
+      "step": 6436
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.873015873015873e-07,
+      "loss": 0.1253,
+      "step": 6437
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8412698412698412e-07,
+      "loss": 0.1281,
+      "step": 6438
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8095238095238097e-07,
+      "loss": 0.1428,
+      "step": 6439
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.777777777777778e-07,
+      "loss": 0.1373,
+      "step": 6440
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.746031746031746e-07,
+      "loss": 0.1235,
+      "step": 6441
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7142857142857146e-07,
+      "loss": 0.1352,
+      "step": 6442
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6825396825396828e-07,
+      "loss": 0.1276,
+      "step": 6443
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.650793650793651e-07,
+      "loss": 0.1355,
+      "step": 6444
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.619047619047619e-07,
+      "loss": 0.1301,
+      "step": 6445
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5873015873015874e-07,
+      "loss": 0.1242,
+      "step": 6446
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5555555555555556e-07,
+      "loss": 0.1256,
+      "step": 6447
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5238095238095238e-07,
+      "loss": 0.1374,
+      "step": 6448
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4920634920634923e-07,
+      "loss": 0.1358,
+      "step": 6449
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4603174603174605e-07,
+      "loss": 0.1405,
+      "step": 6450
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4285714285714287e-07,
+      "loss": 0.1296,
+      "step": 6451
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.396825396825397e-07,
+      "loss": 0.1255,
+      "step": 6452
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3650793650793652e-07,
+      "loss": 0.1464,
+      "step": 6453
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3333333333333336e-07,
+      "loss": 0.1297,
+      "step": 6454
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3015873015873016e-07,
+      "loss": 0.1388,
+      "step": 6455
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.26984126984127e-07,
+      "loss": 0.1359,
+      "step": 6456
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2380952380952383e-07,
+      "loss": 0.1366,
+      "step": 6457
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2063492063492065e-07,
+      "loss": 0.1271,
+      "step": 6458
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1746031746031747e-07,
+      "loss": 0.1297,
+      "step": 6459
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.142857142857143e-07,
+      "loss": 0.1329,
+      "step": 6460
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1111111111111112e-07,
+      "loss": 0.1341,
+      "step": 6461
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0793650793650796e-07,
+      "loss": 0.1412,
+      "step": 6462
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0476190476190476e-07,
+      "loss": 0.1295,
+      "step": 6463
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.015873015873016e-07,
+      "loss": 0.1169,
+      "step": 6464
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.841269841269842e-08,
+      "loss": 0.1362,
+      "step": 6465
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.523809523809525e-08,
+      "loss": 0.1328,
+      "step": 6466
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.206349206349206e-08,
+      "loss": 0.1354,
+      "step": 6467
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.88888888888889e-08,
+      "loss": 0.1265,
+      "step": 6468
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.571428571428573e-08,
+      "loss": 0.1297,
+      "step": 6469
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.253968253968255e-08,
+      "loss": 0.1353,
+      "step": 6470
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.936507936507937e-08,
+      "loss": 0.1288,
+      "step": 6471
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.619047619047619e-08,
+      "loss": 0.1299,
+      "step": 6472
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.301587301587303e-08,
+      "loss": 0.1388,
+      "step": 6473
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.984126984126985e-08,
+      "loss": 0.1294,
+      "step": 6474
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.666666666666668e-08,
+      "loss": 0.1262,
+      "step": 6475
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.34920634920635e-08,
+      "loss": 0.1371,
+      "step": 6476
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.031746031746032e-08,
+      "loss": 0.1418,
+      "step": 6477
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.714285714285715e-08,
+      "loss": 0.1277,
+      "step": 6478
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.396825396825398e-08,
+      "loss": 0.12,
+      "step": 6479
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.07936507936508e-08,
+      "loss": 0.1299,
+      "step": 6480
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.7619047619047627e-08,
+      "loss": 0.127,
+      "step": 6481
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.444444444444445e-08,
+      "loss": 0.1275,
+      "step": 6482
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1269841269841275e-08,
+      "loss": 0.1281,
+      "step": 6483
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.8095238095238096e-08,
+      "loss": 0.1346,
+      "step": 6484
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.4920634920634924e-08,
+      "loss": 0.1311,
+      "step": 6485
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.174603174603175e-08,
+      "loss": 0.1379,
+      "step": 6486
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8571428571428575e-08,
+      "loss": 0.1277,
+      "step": 6487
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.53968253968254e-08,
+      "loss": 0.1297,
+      "step": 6488
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2222222222222224e-08,
+      "loss": 0.1302,
+      "step": 6489
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9047619047619048e-08,
+      "loss": 0.1308,
+      "step": 6490
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5873015873015876e-08,
+      "loss": 0.1354,
+      "step": 6491
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.26984126984127e-08,
+      "loss": 0.1299,
+      "step": 6492
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.523809523809524e-09,
+      "loss": 0.1359,
+      "step": 6493
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.34920634920635e-09,
+      "loss": 0.127,
+      "step": 6494
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.174603174603175e-09,
+      "loss": 0.1321,
+      "step": 6495
+    },
+    {
+      "epoch": 3.0,
+      "step": 6495,
+      "total_flos": 2739424075186176.0,
+      "train_loss": 0.3304880689195801,
+      "train_runtime": 205756.3721,
+      "train_samples_per_second": 4.041,
+      "train_steps_per_second": 0.032
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6495,
+  "num_train_epochs": 3,
+  "save_steps": 100000,
+  "total_flos": 2739424075186176.0,
+  "trial_name": null,
+  "trial_params": null
+}