diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9517 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 1582,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0,
+      "loss": 1.5654,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0,
+      "loss": 1.5907,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0,
+      "loss": 1.5874,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 1.562,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 1.5541,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 1.567,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 1.5696,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 1.6244,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 1.5748,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 1.5526,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 1.552,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0,
+      "loss": 1.5895,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0,
+      "loss": 1.5527,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0,
+      "loss": 1.5988,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0,
+      "loss": 1.5558,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0,
+      "loss": 1.5579,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0,
+      "loss": 1.5671,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0,
+      "loss": 1.5648,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.581044635020827e-06,
+      "loss": 1.6027,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.675821459916692e-06,
+      "loss": 1.589,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.162089270041654e-06,
+      "loss": 1.5901,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.31492814710052e-06,
+      "loss": 1.5896,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.256866094937518e-06,
+      "loss": 1.5966,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0053263282233697e-05,
+      "loss": 1.5472,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0743133905062482e-05,
+      "loss": 1.557,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1351642919833384e-05,
+      "loss": 1.5989,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1895972782121347e-05,
+      "loss": 1.5863,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2388379038142511e-05,
+      "loss": 1.5452,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2837910729958346e-05,
+      "loss": 1.5958,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.325143979986714e-05,
+      "loss": 1.6016,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3634307917254523e-05,
+      "loss": 1.5522,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.399074960701721e-05,
+      "loss": 1.5394,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4324178540083309e-05,
+      "loss": 1.547,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4637386878506516e-05,
+      "loss": 1.5934,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4932687554854209e-05,
+      "loss": 1.6129,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5212018031974513e-05,
+      "loss": 1.554,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5477017417142176e-05,
+      "loss": 1.5764,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5729084742150388e-05,
+      "loss": 1.5516,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.596942367316334e-05,
+      "loss": 1.6008,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6199077273922286e-05,
+      "loss": 1.5685,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6418955364979175e-05,
+      "loss": 1.5862,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.662985629420104e-05,
+      "loss": 1.5953,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6832484434887967e-05,
+      "loss": 1.5745,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7027464379750072e-05,
+      "loss": 1.5314,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.721535255227535e-05,
+      "loss": 1.5197,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7396646779634733e-05,
+      "loss": 1.5728,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.757179424203804e-05,
+      "loss": 1.5845,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7741198118150898e-05,
+      "loss": 1.5758,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7905223175104135e-05,
+      "loss": 1.5164,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.80642004980592e-05,
+      "loss": 1.5868,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8218431513527344e-05,
+      "loss": 1.5559,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8368191429334215e-05,
+      "loss": 1.5845,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8513732189875035e-05,
+      "loss": 1.5432,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8655285026376745e-05,
+      "loss": 1.5356,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8793062666995343e-05,
+      "loss": 1.5105,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.892726125978383e-05,
+      "loss": 1.5472,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9058062052163002e-05,
+      "loss": 1.5493,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9185632862982668e-05,
+      "loss": 1.5358,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9310129377171218e-05,
+      "loss": 1.5605,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9431696288028554e-05,
+      "loss": 1.5145,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9550468308184164e-05,
+      "loss": 1.5266,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96665710669339e-05,
+      "loss": 1.4972,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9780121908943112e-05,
+      "loss": 1.5375,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98912306070467e-05,
+      "loss": 1.4777,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 1.5253,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 1.4702,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9986962190352023e-05,
+      "loss": 1.5171,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9973924380704045e-05,
+      "loss": 1.5054,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9960886571056063e-05,
+      "loss": 1.494,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9947848761408084e-05,
+      "loss": 1.5168,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9934810951760106e-05,
+      "loss": 1.4954,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9921773142112127e-05,
+      "loss": 1.521,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990873533246415e-05,
+      "loss": 1.4507,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9895697522816167e-05,
+      "loss": 1.4855,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.988265971316819e-05,
+      "loss": 1.4561,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.986962190352021e-05,
+      "loss": 1.4458,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9856584093872232e-05,
+      "loss": 1.4755,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.984354628422425e-05,
+      "loss": 1.4816,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9830508474576275e-05,
+      "loss": 1.4276,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9817470664928293e-05,
+      "loss": 1.432,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9804432855280314e-05,
+      "loss": 1.4384,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9791395045632336e-05,
+      "loss": 1.468,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778357235984354e-05,
+      "loss": 1.455,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.976531942633638e-05,
+      "loss": 1.4219,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9752281616688397e-05,
+      "loss": 1.4377,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.973924380704042e-05,
+      "loss": 1.4341,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.972620599739244e-05,
+      "loss": 1.4198,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9713168187744462e-05,
+      "loss": 1.4469,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970013037809648e-05,
+      "loss": 1.4092,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96870925684485e-05,
+      "loss": 1.4312,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9674054758800523e-05,
+      "loss": 1.4139,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9661016949152545e-05,
+      "loss": 1.4419,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9647979139504566e-05,
+      "loss": 1.4158,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9634941329856584e-05,
+      "loss": 1.4456,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621903520208606e-05,
+      "loss": 1.4178,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608865710560627e-05,
+      "loss": 1.4132,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959582790091265e-05,
+      "loss": 1.4174,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958279009126467e-05,
+      "loss": 1.4387,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.956975228161669e-05,
+      "loss": 1.3621,
+      "step": 99
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9556714471968713e-05,
+      "loss": 1.4186,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.954367666232073e-05,
+      "loss": 1.3762,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530638852672753e-05,
+      "loss": 1.4232,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517601043024775e-05,
+      "loss": 1.3823,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504563233376793e-05,
+      "loss": 1.3579,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491525423728814e-05,
+      "loss": 1.3751,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478487614080836e-05,
+      "loss": 1.3787,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9465449804432857e-05,
+      "loss": 1.3884,
+      "step": 107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9452411994784876e-05,
+      "loss": 1.4022,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94393741851369e-05,
+      "loss": 1.3467,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942633637548892e-05,
+      "loss": 1.3571,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941329856584094e-05,
+      "loss": 1.4038,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940026075619296e-05,
+      "loss": 1.3687,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938722294654498e-05,
+      "loss": 1.3716,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374185136897005e-05,
+      "loss": 1.3608,
+      "step": 114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9361147327249023e-05,
+      "loss": 1.3607,
+      "step": 115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9348109517601044e-05,
+      "loss": 1.342,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9335071707953066e-05,
+      "loss": 1.3776,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9322033898305087e-05,
+      "loss": 1.3895,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930899608865711e-05,
+      "loss": 1.3678,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295958279009127e-05,
+      "loss": 1.3893,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928292046936115e-05,
+      "loss": 1.373,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926988265971317e-05,
+      "loss": 1.3845,
+      "step": 122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9256844850065192e-05,
+      "loss": 1.3303,
+      "step": 123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.924380704041721e-05,
+      "loss": 1.3756,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.3575,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9217731421121253e-05,
+      "loss": 1.3395,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9204693611473274e-05,
+      "loss": 1.3496,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191655801825296e-05,
+      "loss": 1.3474,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178617992177314e-05,
+      "loss": 1.3518,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916558018252934e-05,
+      "loss": 1.3422,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9152542372881357e-05,
+      "loss": 1.3752,
+      "step": 131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.913950456323338e-05,
+      "loss": 1.3889,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.91264667535854e-05,
+      "loss": 1.354,
+      "step": 133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.911342894393742e-05,
+      "loss": 1.3274,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.910039113428944e-05,
+      "loss": 1.3337,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.908735332464146e-05,
+      "loss": 1.3867,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9074315514993483e-05,
+      "loss": 1.3417,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9061277705345505e-05,
+      "loss": 1.3513,
+      "step": 138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9048239895697526e-05,
+      "loss": 1.3544,
+      "step": 139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9035202086049544e-05,
+      "loss": 1.3694,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9022164276401566e-05,
+      "loss": 1.3424,
+      "step": 141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9009126466753587e-05,
+      "loss": 1.3429,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8996088657105605e-05,
+      "loss": 1.3328,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.898305084745763e-05,
+      "loss": 1.3157,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.897001303780965e-05,
+      "loss": 1.335,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.895697522816167e-05,
+      "loss": 1.3044,
+      "step": 146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.894393741851369e-05,
+      "loss": 1.3512,
+      "step": 147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8930899608865713e-05,
+      "loss": 1.3273,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8917861799217735e-05,
+      "loss": 1.3615,
+      "step": 149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8904823989569753e-05,
+      "loss": 1.3544,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8891786179921774e-05,
+      "loss": 1.3354,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8878748370273796e-05,
+      "loss": 1.3845,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8865710560625817e-05,
+      "loss": 1.3317,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8852672750977836e-05,
+      "loss": 1.3263,
+      "step": 154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8839634941329857e-05,
+      "loss": 1.3234,
+      "step": 155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.882659713168188e-05,
+      "loss": 1.3269,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.88135593220339e-05,
+      "loss": 1.2842,
+      "step": 157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.880052151238592e-05,
+      "loss": 1.3395,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.878748370273794e-05,
+      "loss": 1.312,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8774445893089965e-05,
+      "loss": 1.2876,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8761408083441983e-05,
+      "loss": 1.3389,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8748370273794004e-05,
+      "loss": 1.3048,
+      "step": 162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8735332464146026e-05,
+      "loss": 1.3628,
+      "step": 163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8722294654498044e-05,
+      "loss": 1.303,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.870925684485007e-05,
+      "loss": 1.3248,
+      "step": 165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8696219035202087e-05,
+      "loss": 1.3054,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.868318122555411e-05,
+      "loss": 1.3245,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.867014341590613e-05,
+      "loss": 1.3425,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8657105606258152e-05,
+      "loss": 1.3424,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.864406779661017e-05,
+      "loss": 1.3203,
+      "step": 170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.863102998696219e-05,
+      "loss": 1.3473,
+      "step": 171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8617992177314213e-05,
+      "loss": 1.366,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.860495436766623e-05,
+      "loss": 1.2986,
+      "step": 173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8591916558018256e-05,
+      "loss": 1.3368,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8578878748370274e-05,
+      "loss": 1.3269,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8565840938722296e-05,
+      "loss": 1.2887,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8552803129074317e-05,
+      "loss": 1.3384,
+      "step": 177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.853976531942634e-05,
+      "loss": 1.3027,
+      "step": 178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.852672750977836e-05,
+      "loss": 1.308,
+      "step": 179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.851368970013038e-05,
+      "loss": 1.3005,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.85006518904824e-05,
+      "loss": 1.3064,
+      "step": 181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.848761408083442e-05,
+      "loss": 1.3042,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8474576271186443e-05,
+      "loss": 1.2847,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.2906,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8448500651890483e-05,
+      "loss": 1.3507,
+      "step": 185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8435462842242504e-05,
+      "loss": 1.3068,
+      "step": 186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8422425032594526e-05,
+      "loss": 1.3324,
+      "step": 187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8409387222946547e-05,
+      "loss": 1.324,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8396349413298566e-05,
+      "loss": 1.3074,
+      "step": 189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.838331160365059e-05,
+      "loss": 1.3286,
+      "step": 190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.837027379400261e-05,
+      "loss": 1.2745,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.835723598435463e-05,
+      "loss": 1.25,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.834419817470665e-05,
+      "loss": 1.3082,
+      "step": 193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.833116036505867e-05,
+      "loss": 1.3041,
+      "step": 194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8318122555410695e-05,
+      "loss": 1.2632,
+      "step": 195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8305084745762713e-05,
+      "loss": 1.2894,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8292046936114734e-05,
+      "loss": 1.292,
+      "step": 197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8279009126466756e-05,
+      "loss": 1.3499,
+      "step": 198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8265971316818777e-05,
+      "loss": 1.3208,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8252933507170796e-05,
+      "loss": 1.3112,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8239895697522817e-05,
+      "loss": 1.3215,
+      "step": 201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.822685788787484e-05,
+      "loss": 1.3517,
+      "step": 202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.821382007822686e-05,
+      "loss": 1.2925,
+      "step": 203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8200782268578882e-05,
+      "loss": 1.3116,
+      "step": 204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.81877444589309e-05,
+      "loss": 1.3151,
+      "step": 205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.817470664928292e-05,
+      "loss": 1.2968,
+      "step": 206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8161668839634943e-05,
+      "loss": 1.3361,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8148631029986964e-05,
+      "loss": 1.2667,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8135593220338986e-05,
+      "loss": 1.3018,
+      "step": 209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8122555410691004e-05,
+      "loss": 1.3289,
+      "step": 210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8109517601043026e-05,
+      "loss": 1.2583,
+      "step": 211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8096479791395047e-05,
+      "loss": 1.3079,
+      "step": 212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.808344198174707e-05,
+      "loss": 1.303,
+      "step": 213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.807040417209909e-05,
+      "loss": 1.325,
+      "step": 214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.805736636245111e-05,
+      "loss": 1.264,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.804432855280313e-05,
+      "loss": 1.2833,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.803129074315515e-05,
+      "loss": 1.2604,
+      "step": 217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8018252933507173e-05,
+      "loss": 1.3169,
+      "step": 218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.800521512385919e-05,
+      "loss": 1.3234,
+      "step": 219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7992177314211213e-05,
+      "loss": 1.2804,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7979139504563234e-05,
+      "loss": 1.2938,
+      "step": 221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7966101694915256e-05,
+      "loss": 1.3267,
+      "step": 222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7953063885267277e-05,
+      "loss": 1.2988,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7940026075619295e-05,
+      "loss": 1.34,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.792698826597132e-05,
+      "loss": 1.2924,
+      "step": 225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.791395045632334e-05,
+      "loss": 1.276,
+      "step": 226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.790091264667536e-05,
+      "loss": 1.3103,
+      "step": 227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.788787483702738e-05,
+      "loss": 1.3031,
+      "step": 228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.78748370273794e-05,
+      "loss": 1.2872,
+      "step": 229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7861799217731425e-05,
+      "loss": 1.301,
+      "step": 230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7848761408083443e-05,
+      "loss": 1.239,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7835723598435464e-05,
+      "loss": 1.3019,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7822685788787486e-05,
+      "loss": 1.3147,
+      "step": 233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7809647979139507e-05,
+      "loss": 1.2756,
+      "step": 234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7796610169491526e-05,
+      "loss": 1.2791,
+      "step": 235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7783572359843547e-05,
+      "loss": 1.2959,
+      "step": 236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.777053455019557e-05,
+      "loss": 1.2891,
+      "step": 237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7757496740547587e-05,
+      "loss": 1.2815,
+      "step": 238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.774445893089961e-05,
+      "loss": 1.304,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.773142112125163e-05,
+      "loss": 1.2603,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.771838331160365e-05,
+      "loss": 1.3018,
+      "step": 241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7705345501955673e-05,
+      "loss": 1.3156,
+      "step": 242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.262,
+      "step": 243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7679269882659716e-05,
+      "loss": 1.2852,
+      "step": 244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7666232073011734e-05,
+      "loss": 1.2615,
+      "step": 245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7653194263363756e-05,
+      "loss": 1.3041,
+      "step": 246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7640156453715777e-05,
+      "loss": 1.2882,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.76271186440678e-05,
+      "loss": 1.2821,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.761408083441982e-05,
+      "loss": 1.2638,
+      "step": 249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.760104302477184e-05,
+      "loss": 1.3298,
+      "step": 250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.758800521512386e-05,
+      "loss": 1.2759,
+      "step": 251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.757496740547588e-05,
+      "loss": 1.2672,
+      "step": 252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7561929595827903e-05,
+      "loss": 1.248,
+      "step": 253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.754889178617992e-05,
+      "loss": 1.2805,
+      "step": 254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7535853976531946e-05,
+      "loss": 1.2679,
+      "step": 255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7522816166883964e-05,
+      "loss": 1.2874,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7509778357235986e-05,
+      "loss": 1.2874,
+      "step": 257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7496740547588007e-05,
+      "loss": 1.2979,
+      "step": 258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7483702737940025e-05,
+      "loss": 1.2655,
+      "step": 259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.747066492829205e-05,
+      "loss": 1.2812,
+      "step": 260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.745762711864407e-05,
+      "loss": 1.261,
+      "step": 261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.744458930899609e-05,
+      "loss": 1.3212,
+      "step": 262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.743155149934811e-05,
+      "loss": 1.2789,
+      "step": 263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7418513689700133e-05,
+      "loss": 1.3164,
+      "step": 264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.740547588005215e-05,
+      "loss": 1.2963,
+      "step": 265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7392438070404173e-05,
+      "loss": 1.2548,
+      "step": 266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7379400260756194e-05,
+      "loss": 1.2684,
+      "step": 267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7366362451108216e-05,
+      "loss": 1.286,
+      "step": 268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7353324641460237e-05,
+      "loss": 1.3213,
+      "step": 269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7340286831812255e-05,
+      "loss": 1.3036,
+      "step": 270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7327249022164277e-05,
+      "loss": 1.2478,
+      "step": 271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.73142112125163e-05,
+      "loss": 1.307,
+      "step": 272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.730117340286832e-05,
+      "loss": 1.2959,
+      "step": 273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.728813559322034e-05,
+      "loss": 1.238,
+      "step": 274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.727509778357236e-05,
+      "loss": 1.2871,
+      "step": 275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.726205997392438e-05,
+      "loss": 1.3219,
+      "step": 276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7249022164276403e-05,
+      "loss": 1.283,
+      "step": 277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7235984354628424e-05,
+      "loss": 1.2689,
+      "step": 278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7222946544980446e-05,
+      "loss": 1.2626,
+      "step": 279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7209908735332464e-05,
+      "loss": 1.2833,
+      "step": 280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7196870925684486e-05,
+      "loss": 1.2628,
+      "step": 281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7183833116036507e-05,
+      "loss": 1.2903,
+      "step": 282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.717079530638853e-05,
+      "loss": 1.2909,
+      "step": 283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7157757496740547e-05,
+      "loss": 1.2816,
+      "step": 284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.714471968709257e-05,
+      "loss": 1.3226,
+      "step": 285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.713168187744459e-05,
+      "loss": 1.252,
+      "step": 286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.711864406779661e-05,
+      "loss": 1.3066,
+      "step": 287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7105606258148633e-05,
+      "loss": 1.2614,
+      "step": 288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.709256844850065e-05,
+      "loss": 1.2714,
+      "step": 289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7079530638852676e-05,
+      "loss": 1.2695,
+      "step": 290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7066492829204694e-05,
+      "loss": 1.2495,
+      "step": 291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7053455019556716e-05,
+      "loss": 1.3086,
+      "step": 292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7040417209908737e-05,
+      "loss": 1.2717,
+      "step": 293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.702737940026076e-05,
+      "loss": 1.2803,
+      "step": 294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7014341590612777e-05,
+      "loss": 1.3202,
+      "step": 295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.70013037809648e-05,
+      "loss": 1.2826,
+      "step": 296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.698826597131682e-05,
+      "loss": 1.2496,
+      "step": 297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.697522816166884e-05,
+      "loss": 1.2594,
+      "step": 298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6962190352020863e-05,
+      "loss": 1.2823,
+      "step": 299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.694915254237288e-05,
+      "loss": 1.2816,
+      "step": 300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6936114732724903e-05,
+      "loss": 1.3027,
+      "step": 301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.2732,
+      "step": 302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6910039113428946e-05,
+      "loss": 1.2601,
+      "step": 303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6897001303780967e-05,
+      "loss": 1.2765,
+      "step": 304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6883963494132985e-05,
+      "loss": 1.2701,
+      "step": 305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.687092568448501e-05,
+      "loss": 1.3153,
+      "step": 306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.685788787483703e-05,
+      "loss": 1.2693,
+      "step": 307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.684485006518905e-05,
+      "loss": 1.2637,
+      "step": 308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.683181225554107e-05,
+      "loss": 1.2845,
+      "step": 309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.681877444589309e-05,
+      "loss": 1.2961,
+      "step": 310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.680573663624511e-05,
+      "loss": 1.2662,
+      "step": 311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6792698826597133e-05,
+      "loss": 1.2484,
+      "step": 312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6779661016949154e-05,
+      "loss": 1.2079,
+      "step": 313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6766623207301176e-05,
+      "loss": 1.2853,
+      "step": 314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6753585397653197e-05,
+      "loss": 1.2921,
+      "step": 315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6740547588005215e-05,
+      "loss": 1.2922,
+      "step": 316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6727509778357237e-05,
+      "loss": 1.2723,
+      "step": 317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.671447196870926e-05,
+      "loss": 1.2664,
+      "step": 318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6701434159061277e-05,
+      "loss": 1.2556,
+      "step": 319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.66883963494133e-05,
+      "loss": 1.2946,
+      "step": 320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.667535853976532e-05,
+      "loss": 1.2504,
+      "step": 321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.666232073011734e-05,
+      "loss": 1.2742,
+      "step": 322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6649282920469363e-05,
+      "loss": 1.2996,
+      "step": 323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6636245110821384e-05,
+      "loss": 1.2412,
+      "step": 324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6623207301173406e-05,
+      "loss": 1.2687,
+      "step": 325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6610169491525424e-05,
+      "loss": 1.2854,
+      "step": 326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6597131681877446e-05,
+      "loss": 1.2737,
+      "step": 327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6584093872229467e-05,
+      "loss": 1.2399,
+      "step": 328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.657105606258149e-05,
+      "loss": 1.2695,
+      "step": 329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6558018252933507e-05,
+      "loss": 1.264,
+      "step": 330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.654498044328553e-05,
+      "loss": 1.2629,
+      "step": 331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.653194263363755e-05,
+      "loss": 1.27,
+      "step": 332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.651890482398957e-05,
+      "loss": 1.243,
+      "step": 333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6505867014341593e-05,
+      "loss": 1.2568,
+      "step": 334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.649282920469361e-05,
+      "loss": 1.2618,
+      "step": 335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6479791395045636e-05,
+      "loss": 1.2542,
+      "step": 336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6466753585397654e-05,
+      "loss": 1.2521,
+      "step": 337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6453715775749676e-05,
+      "loss": 1.2804,
+      "step": 338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6440677966101697e-05,
+      "loss": 1.2739,
+      "step": 339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6427640156453715e-05,
+      "loss": 1.2427,
+      "step": 340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6414602346805737e-05,
+      "loss": 1.287,
+      "step": 341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.640156453715776e-05,
+      "loss": 1.2985,
+      "step": 342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.638852672750978e-05,
+      "loss": 1.2667,
+      "step": 343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.63754889178618e-05,
+      "loss": 1.257,
+      "step": 344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6362451108213823e-05,
+      "loss": 1.2571,
+      "step": 345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.634941329856584e-05,
+      "loss": 1.2734,
+      "step": 346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6336375488917863e-05,
+      "loss": 1.2654,
+      "step": 347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6323337679269884e-05,
+      "loss": 1.2743,
+      "step": 348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6310299869621902e-05,
+      "loss": 1.2706,
+      "step": 349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6297262059973927e-05,
+      "loss": 1.2262,
+      "step": 350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6284224250325945e-05,
+      "loss": 1.2726,
+      "step": 351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6271186440677967e-05,
+      "loss": 1.2593,
+      "step": 352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.625814863102999e-05,
+      "loss": 1.2729,
+      "step": 353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.624511082138201e-05,
+      "loss": 1.29,
+      "step": 354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.623207301173403e-05,
+      "loss": 1.2191,
+      "step": 355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.621903520208605e-05,
+      "loss": 1.2603,
+      "step": 356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.620599739243807e-05,
+      "loss": 1.2557,
+      "step": 357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6192959582790093e-05,
+      "loss": 1.2937,
+      "step": 358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6179921773142114e-05,
+      "loss": 1.3005,
+      "step": 359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6166883963494132e-05,
+      "loss": 1.2596,
+      "step": 360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 1.3081,
+      "step": 361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6140808344198176e-05,
+      "loss": 1.2283,
+      "step": 362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6127770534550197e-05,
+      "loss": 1.24,
+      "step": 363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.611473272490222e-05,
+      "loss": 1.2529,
+      "step": 364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6101694915254237e-05,
+      "loss": 1.2649,
+      "step": 365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.608865710560626e-05,
+      "loss": 1.2634,
+      "step": 366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.607561929595828e-05,
+      "loss": 1.2498,
+      "step": 367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.60625814863103e-05,
+      "loss": 1.2787,
+      "step": 368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6049543676662323e-05,
+      "loss": 1.319,
+      "step": 369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.603650586701434e-05,
+      "loss": 1.2584,
+      "step": 370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6023468057366366e-05,
+      "loss": 1.2662,
+      "step": 371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6010430247718384e-05,
+      "loss": 1.2046,
+      "step": 372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5997392438070406e-05,
+      "loss": 1.246,
+      "step": 373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5984354628422427e-05,
+      "loss": 1.2365,
+      "step": 374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.597131681877445e-05,
+      "loss": 1.2535,
+      "step": 375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5958279009126467e-05,
+      "loss": 1.2272,
+      "step": 376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.594524119947849e-05,
+      "loss": 1.1935,
+      "step": 377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.593220338983051e-05,
+      "loss": 1.2494,
+      "step": 378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5919165580182528e-05,
+      "loss": 1.2434,
+      "step": 379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5906127770534553e-05,
+      "loss": 1.2443,
+      "step": 380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.589308996088657e-05,
+      "loss": 1.2432,
+      "step": 381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5880052151238593e-05,
+      "loss": 1.2986,
+      "step": 382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5867014341590614e-05,
+      "loss": 1.2634,
+      "step": 383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5853976531942636e-05,
+      "loss": 1.2103,
+      "step": 384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5840938722294657e-05,
+      "loss": 1.2687,
+      "step": 385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5827900912646675e-05,
+      "loss": 1.2755,
+      "step": 386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5814863102998697e-05,
+      "loss": 1.2452,
+      "step": 387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.580182529335072e-05,
+      "loss": 1.2461,
+      "step": 388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.578878748370274e-05,
+      "loss": 1.2838,
+      "step": 389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.577574967405476e-05,
+      "loss": 1.2122,
+      "step": 390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.576271186440678e-05,
+      "loss": 1.2466,
+      "step": 391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.57496740547588e-05,
+      "loss": 1.248,
+      "step": 392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5736636245110823e-05,
+      "loss": 1.281,
+      "step": 393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5723598435462844e-05,
+      "loss": 1.2549,
+      "step": 394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5710560625814862e-05,
+      "loss": 1.2725,
+      "step": 395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5697522816166887e-05,
+      "loss": 1.264,
+      "step": 396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5684485006518905e-05,
+      "loss": 1.2319,
+      "step": 397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5671447196870927e-05,
+      "loss": 1.2664,
+      "step": 398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.565840938722295e-05,
+      "loss": 1.2634,
+      "step": 399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5645371577574967e-05,
+      "loss": 1.2524,
+      "step": 400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.563233376792699e-05,
+      "loss": 1.2482,
+      "step": 401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.561929595827901e-05,
+      "loss": 1.2443,
+      "step": 402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.560625814863103e-05,
+      "loss": 1.271,
+      "step": 403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5593220338983053e-05,
+      "loss": 1.2687,
+      "step": 404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5580182529335074e-05,
+      "loss": 1.2257,
+      "step": 405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5567144719687092e-05,
+      "loss": 1.2828,
+      "step": 406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5554106910039114e-05,
+      "loss": 1.2346,
+      "step": 407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5541069100391136e-05,
+      "loss": 1.2562,
+      "step": 408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5528031290743157e-05,
+      "loss": 1.2599,
+      "step": 409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.551499348109518e-05,
+      "loss": 1.2144,
+      "step": 410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5501955671447197e-05,
+      "loss": 1.3008,
+      "step": 411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5488917861799218e-05,
+      "loss": 1.2646,
+      "step": 412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.547588005215124e-05,
+      "loss": 1.2612,
+      "step": 413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.546284224250326e-05,
+      "loss": 1.2371,
+      "step": 414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5449804432855283e-05,
+      "loss": 1.2661,
+      "step": 415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.54367666232073e-05,
+      "loss": 1.2742,
+      "step": 416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5423728813559326e-05,
+      "loss": 1.2406,
+      "step": 417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5410691003911344e-05,
+      "loss": 1.2693,
+      "step": 418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5397653194263366e-05,
+      "loss": 1.2902,
+      "step": 419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.3,
+      "step": 420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5371577574967405e-05,
+      "loss": 1.2482,
+      "step": 421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5358539765319427e-05,
+      "loss": 1.2594,
+      "step": 422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.534550195567145e-05,
+      "loss": 1.2493,
+      "step": 423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.533246414602347e-05,
+      "loss": 1.2396,
+      "step": 424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5319426336375488e-05,
+      "loss": 1.2704,
+      "step": 425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5306388526727513e-05,
+      "loss": 1.2302,
+      "step": 426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.529335071707953e-05,
+      "loss": 1.2756,
+      "step": 427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5280312907431553e-05,
+      "loss": 1.2382,
+      "step": 428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5267275097783574e-05,
+      "loss": 1.2432,
+      "step": 429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5254237288135594e-05,
+      "loss": 1.2303,
+      "step": 430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5241199478487616e-05,
+      "loss": 1.2335,
+      "step": 431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5228161668839635e-05,
+      "loss": 1.2186,
+      "step": 432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5215123859191657e-05,
+      "loss": 1.2338,
+      "step": 433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5202086049543678e-05,
+      "loss": 1.2839,
+      "step": 434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5189048239895698e-05,
+      "loss": 1.2556,
+      "step": 435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.517601043024772e-05,
+      "loss": 1.2371,
+      "step": 436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5162972620599741e-05,
+      "loss": 1.1931,
+      "step": 437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5149934810951761e-05,
+      "loss": 1.2543,
+      "step": 438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5136897001303783e-05,
+      "loss": 1.2205,
+      "step": 439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5123859191655803e-05,
+      "loss": 1.2842,
+      "step": 440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5110821382007822e-05,
+      "loss": 1.2721,
+      "step": 441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5097783572359846e-05,
+      "loss": 1.2308,
+      "step": 442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5084745762711865e-05,
+      "loss": 1.2686,
+      "step": 443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5071707953063885e-05,
+      "loss": 1.2467,
+      "step": 444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5058670143415907e-05,
+      "loss": 1.2302,
+      "step": 445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5045632333767928e-05,
+      "loss": 1.2609,
+      "step": 446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.503259452411995e-05,
+      "loss": 1.2659,
+      "step": 447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.501955671447197e-05,
+      "loss": 1.228,
+      "step": 448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.500651890482399e-05,
+      "loss": 1.2555,
+      "step": 449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4993481095176013e-05,
+      "loss": 1.2301,
+      "step": 450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4980443285528033e-05,
+      "loss": 1.2632,
+      "step": 451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4967405475880053e-05,
+      "loss": 1.2665,
+      "step": 452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4954367666232074e-05,
+      "loss": 1.2283,
+      "step": 453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4941329856584096e-05,
+      "loss": 1.2106,
+      "step": 454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4928292046936117e-05,
+      "loss": 1.2329,
+      "step": 455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4915254237288137e-05,
+      "loss": 1.2777,
+      "step": 456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4902216427640157e-05,
+      "loss": 1.2612,
+      "step": 457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.488917861799218e-05,
+      "loss": 1.2745,
+      "step": 458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.48761408083442e-05,
+      "loss": 1.2434,
+      "step": 459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.486310299869622e-05,
+      "loss": 1.2522,
+      "step": 460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4850065189048241e-05,
+      "loss": 1.2383,
+      "step": 461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4837027379400261e-05,
+      "loss": 1.272,
+      "step": 462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4823989569752283e-05,
+      "loss": 1.2487,
+      "step": 463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4810951760104304e-05,
+      "loss": 1.2676,
+      "step": 464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4797913950456324e-05,
+      "loss": 1.2345,
+      "step": 465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4784876140808346e-05,
+      "loss": 1.224,
+      "step": 466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4771838331160367e-05,
+      "loss": 1.2468,
+      "step": 467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4758800521512387e-05,
+      "loss": 1.2506,
+      "step": 468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4745762711864408e-05,
+      "loss": 1.2106,
+      "step": 469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4732724902216428e-05,
+      "loss": 1.2383,
+      "step": 470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4719687092568448e-05,
+      "loss": 1.2518,
+      "step": 471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4706649282920471e-05,
+      "loss": 1.2449,
+      "step": 472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4693611473272491e-05,
+      "loss": 1.2302,
+      "step": 473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4680573663624513e-05,
+      "loss": 1.2455,
+      "step": 474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4667535853976533e-05,
+      "loss": 1.2515,
+      "step": 475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4654498044328554e-05,
+      "loss": 1.2377,
+      "step": 476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4641460234680576e-05,
+      "loss": 1.2371,
+      "step": 477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4628422425032595e-05,
+      "loss": 1.1954,
+      "step": 478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 1.2476,
+      "step": 479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4602346805736639e-05,
+      "loss": 1.2271,
+      "step": 480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4589308996088658e-05,
+      "loss": 1.2389,
+      "step": 481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4576271186440678e-05,
+      "loss": 1.2675,
+      "step": 482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.45632333767927e-05,
+      "loss": 1.2733,
+      "step": 483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.455019556714472e-05,
+      "loss": 1.2128,
+      "step": 484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4537157757496743e-05,
+      "loss": 1.2449,
+      "step": 485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4524119947848763e-05,
+      "loss": 1.2703,
+      "step": 486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4511082138200782e-05,
+      "loss": 1.2638,
+      "step": 487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4498044328552806e-05,
+      "loss": 1.2646,
+      "step": 488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4485006518904826e-05,
+      "loss": 1.2557,
+      "step": 489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4471968709256845e-05,
+      "loss": 1.2294,
+      "step": 490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4458930899608867e-05,
+      "loss": 1.2352,
+      "step": 491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4445893089960887e-05,
+      "loss": 1.2557,
+      "step": 492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.443285528031291e-05,
+      "loss": 1.2339,
+      "step": 493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.441981747066493e-05,
+      "loss": 1.2534,
+      "step": 494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.440677966101695e-05,
+      "loss": 1.2535,
+      "step": 495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4393741851368971e-05,
+      "loss": 1.2371,
+      "step": 496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4380704041720993e-05,
+      "loss": 1.2476,
+      "step": 497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4367666232073013e-05,
+      "loss": 1.2487,
+      "step": 498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4354628422425034e-05,
+      "loss": 1.23,
+      "step": 499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4341590612777054e-05,
+      "loss": 1.2228,
+      "step": 500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4328552803129077e-05,
+      "loss": 1.2144,
+      "step": 501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4315514993481097e-05,
+      "loss": 1.1974,
+      "step": 502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4302477183833117e-05,
+      "loss": 1.2505,
+      "step": 503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4289439374185138e-05,
+      "loss": 1.2189,
+      "step": 504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4276401564537158e-05,
+      "loss": 1.3004,
+      "step": 505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.426336375488918e-05,
+      "loss": 1.2303,
+      "step": 506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4250325945241201e-05,
+      "loss": 1.2507,
+      "step": 507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4237288135593221e-05,
+      "loss": 1.2526,
+      "step": 508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4224250325945241e-05,
+      "loss": 1.1986,
+      "step": 509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4211212516297264e-05,
+      "loss": 1.256,
+      "step": 510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4198174706649284e-05,
+      "loss": 1.2341,
+      "step": 511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4185136897001306e-05,
+      "loss": 1.2296,
+      "step": 512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4172099087353325e-05,
+      "loss": 1.222,
+      "step": 513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4159061277705345e-05,
+      "loss": 1.2379,
+      "step": 514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4146023468057368e-05,
+      "loss": 1.266,
+      "step": 515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4132985658409388e-05,
+      "loss": 1.2318,
+      "step": 516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4119947848761408e-05,
+      "loss": 1.224,
+      "step": 517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4106910039113431e-05,
+      "loss": 1.2378,
+      "step": 518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4093872229465451e-05,
+      "loss": 1.2502,
+      "step": 519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4080834419817473e-05,
+      "loss": 1.2629,
+      "step": 520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4067796610169493e-05,
+      "loss": 1.2478,
+      "step": 521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4054758800521512e-05,
+      "loss": 1.2451,
+      "step": 522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4041720990873536e-05,
+      "loss": 1.2341,
+      "step": 523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4028683181225555e-05,
+      "loss": 1.2424,
+      "step": 524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4015645371577575e-05,
+      "loss": 1.2479,
+      "step": 525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4002607561929597e-05,
+      "loss": 1.2714,
+      "step": 526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3989569752281618e-05,
+      "loss": 1.2075,
+      "step": 527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3976531942633638e-05,
+      "loss": 1.2098,
+      "step": 528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.396349413298566e-05,
+      "loss": 1.2498,
+      "step": 529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.395045632333768e-05,
+      "loss": 1.2395,
+      "step": 530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3937418513689703e-05,
+      "loss": 1.2181,
+      "step": 531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3924380704041723e-05,
+      "loss": 1.2526,
+      "step": 532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3911342894393742e-05,
+      "loss": 1.2103,
+      "step": 533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3898305084745764e-05,
+      "loss": 1.2539,
+      "step": 534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3885267275097784e-05,
+      "loss": 1.2194,
+      "step": 535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3872229465449805e-05,
+      "loss": 1.2319,
+      "step": 536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3859191655801827e-05,
+      "loss": 1.2878,
+      "step": 537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.2428,
+      "step": 538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.383311603650587e-05,
+      "loss": 1.2076,
+      "step": 539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.382007822685789e-05,
+      "loss": 1.2354,
+      "step": 540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.380704041720991e-05,
+      "loss": 1.215,
+      "step": 541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3794002607561931e-05,
+      "loss": 1.2285,
+      "step": 542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3780964797913951e-05,
+      "loss": 1.2247,
+      "step": 543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3767926988265971e-05,
+      "loss": 1.2616,
+      "step": 544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3754889178617994e-05,
+      "loss": 1.2357,
+      "step": 545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3741851368970014e-05,
+      "loss": 1.2456,
+      "step": 546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3728813559322034e-05,
+      "loss": 1.2196,
+      "step": 547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3715775749674057e-05,
+      "loss": 1.2515,
+      "step": 548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3702737940026077e-05,
+      "loss": 1.2621,
+      "step": 549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3689700130378098e-05,
+      "loss": 1.2527,
+      "step": 550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3676662320730118e-05,
+      "loss": 1.2391,
+      "step": 551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3663624511082138e-05,
+      "loss": 1.2236,
+      "step": 552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3650586701434161e-05,
+      "loss": 1.2314,
+      "step": 553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3637548891786181e-05,
+      "loss": 1.2709,
+      "step": 554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3624511082138201e-05,
+      "loss": 1.2664,
+      "step": 555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3611473272490223e-05,
+      "loss": 1.2439,
+      "step": 556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3598435462842244e-05,
+      "loss": 1.2516,
+      "step": 557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3585397653194266e-05,
+      "loss": 1.1953,
+      "step": 558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3572359843546285e-05,
+      "loss": 1.2264,
+      "step": 559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3559322033898305e-05,
+      "loss": 1.2375,
+      "step": 560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3546284224250328e-05,
+      "loss": 1.2517,
+      "step": 561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3533246414602348e-05,
+      "loss": 1.2524,
+      "step": 562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3520208604954368e-05,
+      "loss": 1.2576,
+      "step": 563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.350717079530639e-05,
+      "loss": 1.2208,
+      "step": 564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.349413298565841e-05,
+      "loss": 1.2578,
+      "step": 565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3481095176010431e-05,
+      "loss": 1.2454,
+      "step": 566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3468057366362453e-05,
+      "loss": 1.2465,
+      "step": 567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3455019556714472e-05,
+      "loss": 1.1848,
+      "step": 568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3441981747066496e-05,
+      "loss": 1.2263,
+      "step": 569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3428943937418515e-05,
+      "loss": 1.2611,
+      "step": 570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3415906127770535e-05,
+      "loss": 1.267,
+      "step": 571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3402868318122557e-05,
+      "loss": 1.2039,
+      "step": 572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3389830508474577e-05,
+      "loss": 1.2028,
+      "step": 573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3376792698826597e-05,
+      "loss": 1.2123,
+      "step": 574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.336375488917862e-05,
+      "loss": 1.2323,
+      "step": 575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.335071707953064e-05,
+      "loss": 1.2263,
+      "step": 576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3337679269882661e-05,
+      "loss": 1.2428,
+      "step": 577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3324641460234683e-05,
+      "loss": 1.2151,
+      "step": 578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3311603650586703e-05,
+      "loss": 1.2126,
+      "step": 579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3298565840938724e-05,
+      "loss": 1.2164,
+      "step": 580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3285528031290744e-05,
+      "loss": 1.2487,
+      "step": 581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3272490221642764e-05,
+      "loss": 1.2478,
+      "step": 582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3259452411994787e-05,
+      "loss": 1.2043,
+      "step": 583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3246414602346807e-05,
+      "loss": 1.24,
+      "step": 584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3233376792698827e-05,
+      "loss": 1.2255,
+      "step": 585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3220338983050848e-05,
+      "loss": 1.2319,
+      "step": 586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.320730117340287e-05,
+      "loss": 1.2208,
+      "step": 587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3194263363754891e-05,
+      "loss": 1.2074,
+      "step": 588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3181225554106911e-05,
+      "loss": 1.2559,
+      "step": 589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3168187744458931e-05,
+      "loss": 1.2375,
+      "step": 590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3155149934810954e-05,
+      "loss": 1.2394,
+      "step": 591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3142112125162974e-05,
+      "loss": 1.1995,
+      "step": 592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3129074315514994e-05,
+      "loss": 1.2228,
+      "step": 593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3116036505867015e-05,
+      "loss": 1.2177,
+      "step": 594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3102998696219035e-05,
+      "loss": 1.212,
+      "step": 595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3089960886571058e-05,
+      "loss": 1.2421,
+      "step": 596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.1732,
+      "step": 597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3063885267275098e-05,
+      "loss": 1.2311,
+      "step": 598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.305084745762712e-05,
+      "loss": 1.1901,
+      "step": 599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3037809647979141e-05,
+      "loss": 1.2166,
+      "step": 600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3024771838331161e-05,
+      "loss": 1.2255,
+      "step": 601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3011734028683183e-05,
+      "loss": 1.2339,
+      "step": 602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2998696219035202e-05,
+      "loss": 1.1895,
+      "step": 603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2985658409387226e-05,
+      "loss": 1.249,
+      "step": 604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2972620599739245e-05,
+      "loss": 1.2374,
+      "step": 605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2959582790091265e-05,
+      "loss": 1.233,
+      "step": 606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2946544980443287e-05,
+      "loss": 1.2328,
+      "step": 607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2933507170795307e-05,
+      "loss": 1.2249,
+      "step": 608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2920469361147328e-05,
+      "loss": 1.2308,
+      "step": 609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.290743155149935e-05,
+      "loss": 1.2108,
+      "step": 610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.289439374185137e-05,
+      "loss": 1.2081,
+      "step": 611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.288135593220339e-05,
+      "loss": 1.1864,
+      "step": 612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2868318122555413e-05,
+      "loss": 1.2281,
+      "step": 613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2855280312907432e-05,
+      "loss": 1.2393,
+      "step": 614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2842242503259454e-05,
+      "loss": 1.22,
+      "step": 615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2829204693611474e-05,
+      "loss": 1.2476,
+      "step": 616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2816166883963494e-05,
+      "loss": 1.2466,
+      "step": 617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2803129074315517e-05,
+      "loss": 1.2136,
+      "step": 618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2790091264667537e-05,
+      "loss": 1.2001,
+      "step": 619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2777053455019557e-05,
+      "loss": 1.2329,
+      "step": 620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.276401564537158e-05,
+      "loss": 1.2358,
+      "step": 621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.27509778357236e-05,
+      "loss": 1.1984,
+      "step": 622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2737940026075621e-05,
+      "loss": 1.2155,
+      "step": 623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2724902216427641e-05,
+      "loss": 1.1827,
+      "step": 624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2711864406779661e-05,
+      "loss": 1.2338,
+      "step": 625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2698826597131684e-05,
+      "loss": 1.1949,
+      "step": 626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2685788787483704e-05,
+      "loss": 1.257,
+      "step": 627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2672750977835724e-05,
+      "loss": 1.2255,
+      "step": 628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2659713168187745e-05,
+      "loss": 1.2471,
+      "step": 629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2646675358539767e-05,
+      "loss": 1.2235,
+      "step": 630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2633637548891787e-05,
+      "loss": 1.2341,
+      "step": 631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2620599739243808e-05,
+      "loss": 1.27,
+      "step": 632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2607561929595828e-05,
+      "loss": 1.2129,
+      "step": 633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2594524119947851e-05,
+      "loss": 1.2241,
+      "step": 634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2581486310299871e-05,
+      "loss": 1.2374,
+      "step": 635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2568448500651891e-05,
+      "loss": 1.2211,
+      "step": 636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2555410691003912e-05,
+      "loss": 1.2537,
+      "step": 637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2542372881355932e-05,
+      "loss": 1.2599,
+      "step": 638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2529335071707954e-05,
+      "loss": 1.209,
+      "step": 639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2516297262059975e-05,
+      "loss": 1.2432,
+      "step": 640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2503259452411995e-05,
+      "loss": 1.2017,
+      "step": 641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2490221642764018e-05,
+      "loss": 1.2379,
+      "step": 642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2477183833116038e-05,
+      "loss": 1.2214,
+      "step": 643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2464146023468058e-05,
+      "loss": 1.2206,
+      "step": 644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.245110821382008e-05,
+      "loss": 1.2076,
+      "step": 645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.24380704041721e-05,
+      "loss": 1.2047,
+      "step": 646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.242503259452412e-05,
+      "loss": 1.2461,
+      "step": 647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2411994784876143e-05,
+      "loss": 1.2225,
+      "step": 648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2398956975228162e-05,
+      "loss": 1.2384,
+      "step": 649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2385919165580182e-05,
+      "loss": 1.2668,
+      "step": 650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2372881355932205e-05,
+      "loss": 1.2402,
+      "step": 651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2359843546284225e-05,
+      "loss": 1.2328,
+      "step": 652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2346805736636247e-05,
+      "loss": 1.2308,
+      "step": 653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2333767926988267e-05,
+      "loss": 1.2576,
+      "step": 654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2320730117340287e-05,
+      "loss": 1.2299,
+      "step": 655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.2217,
+      "step": 656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.229465449804433e-05,
+      "loss": 1.2507,
+      "step": 657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.228161668839635e-05,
+      "loss": 1.2576,
+      "step": 658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2268578878748371e-05,
+      "loss": 1.2118,
+      "step": 659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2255541069100392e-05,
+      "loss": 1.251,
+      "step": 660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2242503259452414e-05,
+      "loss": 1.2106,
+      "step": 661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2229465449804434e-05,
+      "loss": 1.2234,
+      "step": 662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2216427640156454e-05,
+      "loss": 1.1669,
+      "step": 663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2203389830508477e-05,
+      "loss": 1.1932,
+      "step": 664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2190352020860497e-05,
+      "loss": 1.2327,
+      "step": 665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2177314211212517e-05,
+      "loss": 1.2292,
+      "step": 666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2164276401564538e-05,
+      "loss": 1.2108,
+      "step": 667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2151238591916558e-05,
+      "loss": 1.2222,
+      "step": 668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213820078226858e-05,
+      "loss": 1.2054,
+      "step": 669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2125162972620601e-05,
+      "loss": 1.2456,
+      "step": 670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2112125162972621e-05,
+      "loss": 1.2206,
+      "step": 671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2099087353324644e-05,
+      "loss": 1.2288,
+      "step": 672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2086049543676664e-05,
+      "loss": 1.2538,
+      "step": 673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2073011734028684e-05,
+      "loss": 1.2056,
+      "step": 674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2059973924380705e-05,
+      "loss": 1.1759,
+      "step": 675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2046936114732725e-05,
+      "loss": 1.2075,
+      "step": 676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2033898305084745e-05,
+      "loss": 1.2072,
+      "step": 677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2020860495436768e-05,
+      "loss": 1.2133,
+      "step": 678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2007822685788788e-05,
+      "loss": 1.2198,
+      "step": 679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.199478487614081e-05,
+      "loss": 1.2267,
+      "step": 680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1981747066492831e-05,
+      "loss": 1.2028,
+      "step": 681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1968709256844851e-05,
+      "loss": 1.2196,
+      "step": 682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1955671447196873e-05,
+      "loss": 1.1947,
+      "step": 683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1942633637548892e-05,
+      "loss": 1.1884,
+      "step": 684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1929595827900912e-05,
+      "loss": 1.2349,
+      "step": 685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1916558018252935e-05,
+      "loss": 1.2426,
+      "step": 686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1903520208604955e-05,
+      "loss": 1.217,
+      "step": 687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1890482398956975e-05,
+      "loss": 1.207,
+      "step": 688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1877444589308997e-05,
+      "loss": 1.2053,
+      "step": 689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1864406779661018e-05,
+      "loss": 1.2238,
+      "step": 690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.185136897001304e-05,
+      "loss": 1.2151,
+      "step": 691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.183833116036506e-05,
+      "loss": 1.2343,
+      "step": 692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.182529335071708e-05,
+      "loss": 1.2358,
+      "step": 693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1812255541069103e-05,
+      "loss": 1.2286,
+      "step": 694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1799217731421122e-05,
+      "loss": 1.1934,
+      "step": 695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1786179921773142e-05,
+      "loss": 1.1842,
+      "step": 696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1773142112125164e-05,
+      "loss": 1.2043,
+      "step": 697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1760104302477184e-05,
+      "loss": 1.234,
+      "step": 698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1747066492829207e-05,
+      "loss": 1.2546,
+      "step": 699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1734028683181227e-05,
+      "loss": 1.1737,
+      "step": 700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1720990873533247e-05,
+      "loss": 1.2408,
+      "step": 701
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.170795306388527e-05,
+      "loss": 1.2415,
+      "step": 702
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.169491525423729e-05,
+      "loss": 1.2068,
+      "step": 703
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.168187744458931e-05,
+      "loss": 1.2448,
+      "step": 704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1668839634941331e-05,
+      "loss": 1.2314,
+      "step": 705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.165580182529335e-05,
+      "loss": 1.2467,
+      "step": 706
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1642764015645374e-05,
+      "loss": 1.208,
+      "step": 707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1629726205997394e-05,
+      "loss": 1.2262,
+      "step": 708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1616688396349414e-05,
+      "loss": 1.1771,
+      "step": 709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1603650586701435e-05,
+      "loss": 1.1926,
+      "step": 710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1590612777053457e-05,
+      "loss": 1.2075,
+      "step": 711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1577574967405477e-05,
+      "loss": 1.2258,
+      "step": 712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1564537157757498e-05,
+      "loss": 1.2113,
+      "step": 713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1551499348109518e-05,
+      "loss": 1.2128,
+      "step": 714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 1.2002,
+      "step": 715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1525423728813561e-05,
+      "loss": 1.2115,
+      "step": 716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1512385919165581e-05,
+      "loss": 1.1877,
+      "step": 717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1499348109517602e-05,
+      "loss": 1.2522,
+      "step": 718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1486310299869622e-05,
+      "loss": 1.2061,
+      "step": 719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1473272490221644e-05,
+      "loss": 1.1938,
+      "step": 720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1460234680573665e-05,
+      "loss": 1.1967,
+      "step": 721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1447196870925685e-05,
+      "loss": 1.2426,
+      "step": 722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1434159061277705e-05,
+      "loss": 1.1947,
+      "step": 723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1421121251629728e-05,
+      "loss": 1.2333,
+      "step": 724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1408083441981748e-05,
+      "loss": 1.1958,
+      "step": 725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.139504563233377e-05,
+      "loss": 1.1732,
+      "step": 726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.138200782268579e-05,
+      "loss": 1.2079,
+      "step": 727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.136897001303781e-05,
+      "loss": 1.2202,
+      "step": 728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1355932203389833e-05,
+      "loss": 1.1875,
+      "step": 729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1342894393741852e-05,
+      "loss": 1.2174,
+      "step": 730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1329856584093872e-05,
+      "loss": 1.1995,
+      "step": 731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1316818774445895e-05,
+      "loss": 1.2087,
+      "step": 732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1303780964797915e-05,
+      "loss": 1.2072,
+      "step": 733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1290743155149935e-05,
+      "loss": 1.2104,
+      "step": 734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1277705345501957e-05,
+      "loss": 1.192,
+      "step": 735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1264667535853976e-05,
+      "loss": 1.2162,
+      "step": 736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1251629726206e-05,
+      "loss": 1.1765,
+      "step": 737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.123859191655802e-05,
+      "loss": 1.1801,
+      "step": 738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.122555410691004e-05,
+      "loss": 1.2258,
+      "step": 739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1212516297262061e-05,
+      "loss": 1.1937,
+      "step": 740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1199478487614082e-05,
+      "loss": 1.2339,
+      "step": 741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1186440677966102e-05,
+      "loss": 1.1935,
+      "step": 742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1173402868318124e-05,
+      "loss": 1.2578,
+      "step": 743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1160365058670144e-05,
+      "loss": 1.1941,
+      "step": 744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1147327249022167e-05,
+      "loss": 1.2177,
+      "step": 745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1134289439374187e-05,
+      "loss": 1.2064,
+      "step": 746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1121251629726207e-05,
+      "loss": 1.2084,
+      "step": 747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1108213820078228e-05,
+      "loss": 1.2031,
+      "step": 748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1095176010430248e-05,
+      "loss": 1.2231,
+      "step": 749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.108213820078227e-05,
+      "loss": 1.2166,
+      "step": 750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1069100391134291e-05,
+      "loss": 1.2198,
+      "step": 751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1056062581486311e-05,
+      "loss": 1.2103,
+      "step": 752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.104302477183833e-05,
+      "loss": 1.2113,
+      "step": 753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1029986962190354e-05,
+      "loss": 1.2184,
+      "step": 754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1016949152542374e-05,
+      "loss": 1.2081,
+      "step": 755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1003911342894395e-05,
+      "loss": 1.2292,
+      "step": 756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0990873533246415e-05,
+      "loss": 1.1931,
+      "step": 757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0977835723598435e-05,
+      "loss": 1.179,
+      "step": 758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0964797913950458e-05,
+      "loss": 1.2032,
+      "step": 759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0951760104302478e-05,
+      "loss": 1.2179,
+      "step": 760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0938722294654498e-05,
+      "loss": 1.2339,
+      "step": 761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.092568448500652e-05,
+      "loss": 1.2264,
+      "step": 762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0912646675358541e-05,
+      "loss": 1.2042,
+      "step": 763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0899608865710562e-05,
+      "loss": 1.1958,
+      "step": 764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0886571056062582e-05,
+      "loss": 1.2015,
+      "step": 765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0873533246414602e-05,
+      "loss": 1.2152,
+      "step": 766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0860495436766625e-05,
+      "loss": 1.2483,
+      "step": 767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0847457627118645e-05,
+      "loss": 1.2361,
+      "step": 768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0834419817470665e-05,
+      "loss": 1.2026,
+      "step": 769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0821382007822687e-05,
+      "loss": 1.2201,
+      "step": 770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0808344198174706e-05,
+      "loss": 1.2163,
+      "step": 771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0795306388526728e-05,
+      "loss": 1.2191,
+      "step": 772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.078226857887875e-05,
+      "loss": 1.2307,
+      "step": 773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.2061,
+      "step": 774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0756192959582793e-05,
+      "loss": 1.2155,
+      "step": 775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0743155149934812e-05,
+      "loss": 1.2186,
+      "step": 776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0730117340286832e-05,
+      "loss": 1.2218,
+      "step": 777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0717079530638854e-05,
+      "loss": 1.2332,
+      "step": 778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0704041720990874e-05,
+      "loss": 1.1931,
+      "step": 779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0691003911342893e-05,
+      "loss": 1.1971,
+      "step": 780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0677966101694917e-05,
+      "loss": 1.2264,
+      "step": 781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0664928292046937e-05,
+      "loss": 1.2499,
+      "step": 782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0651890482398958e-05,
+      "loss": 1.2117,
+      "step": 783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.063885267275098e-05,
+      "loss": 1.2285,
+      "step": 784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0625814863103e-05,
+      "loss": 1.2216,
+      "step": 785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0612777053455021e-05,
+      "loss": 1.2141,
+      "step": 786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.059973924380704e-05,
+      "loss": 1.1837,
+      "step": 787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.058670143415906e-05,
+      "loss": 1.1733,
+      "step": 788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0573663624511084e-05,
+      "loss": 1.2407,
+      "step": 789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0560625814863104e-05,
+      "loss": 1.198,
+      "step": 790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0547588005215125e-05,
+      "loss": 1.3012,
+      "step": 791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0534550195567145e-05,
+      "loss": 1.2131,
+      "step": 792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0521512385919167e-05,
+      "loss": 1.2198,
+      "step": 793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0508474576271188e-05,
+      "loss": 1.208,
+      "step": 794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0495436766623208e-05,
+      "loss": 1.2153,
+      "step": 795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0482398956975228e-05,
+      "loss": 1.2216,
+      "step": 796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0469361147327251e-05,
+      "loss": 1.2345,
+      "step": 797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0456323337679271e-05,
+      "loss": 1.224,
+      "step": 798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.044328552803129e-05,
+      "loss": 1.1936,
+      "step": 799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0430247718383312e-05,
+      "loss": 1.2423,
+      "step": 800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0417209908735332e-05,
+      "loss": 1.1975,
+      "step": 801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0404172099087355e-05,
+      "loss": 1.2197,
+      "step": 802
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0391134289439375e-05,
+      "loss": 1.2596,
+      "step": 803
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0378096479791395e-05,
+      "loss": 1.1724,
+      "step": 804
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0365058670143418e-05,
+      "loss": 1.2084,
+      "step": 805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0352020860495438e-05,
+      "loss": 1.1946,
+      "step": 806
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0338983050847458e-05,
+      "loss": 1.2518,
+      "step": 807
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.032594524119948e-05,
+      "loss": 1.1707,
+      "step": 808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.03129074315515e-05,
+      "loss": 1.2238,
+      "step": 809
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0299869621903522e-05,
+      "loss": 1.1854,
+      "step": 810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0286831812255542e-05,
+      "loss": 1.1964,
+      "step": 811
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0273794002607562e-05,
+      "loss": 1.1992,
+      "step": 812
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0260756192959584e-05,
+      "loss": 1.2385,
+      "step": 813
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0247718383311605e-05,
+      "loss": 1.2129,
+      "step": 814
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0234680573663625e-05,
+      "loss": 1.1818,
+      "step": 815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0221642764015647e-05,
+      "loss": 1.2023,
+      "step": 816
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0208604954367666e-05,
+      "loss": 1.2354,
+      "step": 817
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0195567144719686e-05,
+      "loss": 1.1669,
+      "step": 818
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.018252933507171e-05,
+      "loss": 1.2156,
+      "step": 819
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.016949152542373e-05,
+      "loss": 1.2052,
+      "step": 820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0156453715775751e-05,
+      "loss": 1.1943,
+      "step": 821
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.014341590612777e-05,
+      "loss": 1.208,
+      "step": 822
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0130378096479792e-05,
+      "loss": 1.2069,
+      "step": 823
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0117340286831814e-05,
+      "loss": 1.21,
+      "step": 824
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0104302477183834e-05,
+      "loss": 1.2218,
+      "step": 825
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0091264667535853e-05,
+      "loss": 1.1984,
+      "step": 826
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0078226857887877e-05,
+      "loss": 1.1804,
+      "step": 827
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0065189048239897e-05,
+      "loss": 1.213,
+      "step": 828
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0052151238591918e-05,
+      "loss": 1.2312,
+      "step": 829
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0039113428943938e-05,
+      "loss": 1.2502,
+      "step": 830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0026075619295958e-05,
+      "loss": 1.212,
+      "step": 831
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0013037809647981e-05,
+      "loss": 1.2222,
+      "step": 832
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 1.2419,
+      "step": 833
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.986962190352022e-06,
+      "loss": 1.1929,
+      "step": 834
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.973924380704042e-06,
+      "loss": 1.1882,
+      "step": 835
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.960886571056064e-06,
+      "loss": 1.2307,
+      "step": 836
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.947848761408084e-06,
+      "loss": 1.2312,
+      "step": 837
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.934810951760105e-06,
+      "loss": 1.1963,
+      "step": 838
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.921773142112125e-06,
+      "loss": 1.222,
+      "step": 839
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.908735332464146e-06,
+      "loss": 1.2078,
+      "step": 840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.895697522816168e-06,
+      "loss": 1.2306,
+      "step": 841
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.88265971316819e-06,
+      "loss": 1.1906,
+      "step": 842
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.86962190352021e-06,
+      "loss": 1.2075,
+      "step": 843
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.856584093872231e-06,
+      "loss": 1.222,
+      "step": 844
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.84354628422425e-06,
+      "loss": 1.2206,
+      "step": 845
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.830508474576272e-06,
+      "loss": 1.1727,
+      "step": 846
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.817470664928292e-06,
+      "loss": 1.2115,
+      "step": 847
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.804432855280314e-06,
+      "loss": 1.1754,
+      "step": 848
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.791395045632335e-06,
+      "loss": 1.2238,
+      "step": 849
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.778357235984357e-06,
+      "loss": 1.1927,
+      "step": 850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.765319426336377e-06,
+      "loss": 1.1995,
+      "step": 851
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.752281616688396e-06,
+      "loss": 1.1984,
+      "step": 852
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.739243807040418e-06,
+      "loss": 1.2366,
+      "step": 853
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.726205997392438e-06,
+      "loss": 1.1886,
+      "step": 854
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.71316818774446e-06,
+      "loss": 1.1871,
+      "step": 855
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.70013037809648e-06,
+      "loss": 1.1919,
+      "step": 856
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.687092568448502e-06,
+      "loss": 1.2045,
+      "step": 857
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.674054758800522e-06,
+      "loss": 1.2053,
+      "step": 858
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.661016949152544e-06,
+      "loss": 1.2543,
+      "step": 859
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.647979139504564e-06,
+      "loss": 1.1973,
+      "step": 860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.634941329856585e-06,
+      "loss": 1.2036,
+      "step": 861
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.621903520208605e-06,
+      "loss": 1.1744,
+      "step": 862
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.608865710560626e-06,
+      "loss": 1.2272,
+      "step": 863
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.595827900912648e-06,
+      "loss": 1.2266,
+      "step": 864
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.58279009126467e-06,
+      "loss": 1.1694,
+      "step": 865
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.56975228161669e-06,
+      "loss": 1.1765,
+      "step": 866
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.55671447196871e-06,
+      "loss": 1.1829,
+      "step": 867
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.54367666232073e-06,
+      "loss": 1.2461,
+      "step": 868
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.530638852672752e-06,
+      "loss": 1.2266,
+      "step": 869
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.517601043024772e-06,
+      "loss": 1.2247,
+      "step": 870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.504563233376794e-06,
+      "loss": 1.2061,
+      "step": 871
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.491525423728815e-06,
+      "loss": 1.1727,
+      "step": 872
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.478487614080835e-06,
+      "loss": 1.1575,
+      "step": 873
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.465449804432857e-06,
+      "loss": 1.1871,
+      "step": 874
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.452411994784876e-06,
+      "loss": 1.2117,
+      "step": 875
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.439374185136898e-06,
+      "loss": 1.2083,
+      "step": 876
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.426336375488918e-06,
+      "loss": 1.2404,
+      "step": 877
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.41329856584094e-06,
+      "loss": 1.1959,
+      "step": 878
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.40026075619296e-06,
+      "loss": 1.2286,
+      "step": 879
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.387222946544982e-06,
+      "loss": 1.1969,
+      "step": 880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.374185136897002e-06,
+      "loss": 1.1733,
+      "step": 881
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.361147327249022e-06,
+      "loss": 1.1996,
+      "step": 882
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.348109517601044e-06,
+      "loss": 1.2255,
+      "step": 883
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.335071707953065e-06,
+      "loss": 1.1917,
+      "step": 884
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.322033898305085e-06,
+      "loss": 1.1599,
+      "step": 885
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.308996088657106e-06,
+      "loss": 1.1926,
+      "step": 886
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.295958279009128e-06,
+      "loss": 1.2533,
+      "step": 887
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.282920469361148e-06,
+      "loss": 1.2264,
+      "step": 888
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.26988265971317e-06,
+      "loss": 1.1788,
+      "step": 889
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.25684485006519e-06,
+      "loss": 1.2297,
+      "step": 890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.24380704041721e-06,
+      "loss": 1.1923,
+      "step": 891
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.1813,
+      "step": 892
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.217731421121252e-06,
+      "loss": 1.2067,
+      "step": 893
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.204693611473274e-06,
+      "loss": 1.1957,
+      "step": 894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.191655801825295e-06,
+      "loss": 1.2053,
+      "step": 895
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.178617992177315e-06,
+      "loss": 1.2203,
+      "step": 896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.165580182529335e-06,
+      "loss": 1.1864,
+      "step": 897
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.152542372881356e-06,
+      "loss": 1.1949,
+      "step": 898
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.139504563233378e-06,
+      "loss": 1.2115,
+      "step": 899
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.126466753585398e-06,
+      "loss": 1.2247,
+      "step": 900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.11342894393742e-06,
+      "loss": 1.1991,
+      "step": 901
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.100391134289441e-06,
+      "loss": 1.184,
+      "step": 902
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.08735332464146e-06,
+      "loss": 1.2025,
+      "step": 903
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.074315514993482e-06,
+      "loss": 1.2031,
+      "step": 904
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.061277705345502e-06,
+      "loss": 1.2051,
+      "step": 905
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.048239895697524e-06,
+      "loss": 1.197,
+      "step": 906
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.035202086049545e-06,
+      "loss": 1.1852,
+      "step": 907
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.022164276401565e-06,
+      "loss": 1.2126,
+      "step": 908
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.009126466753587e-06,
+      "loss": 1.2097,
+      "step": 909
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.996088657105606e-06,
+      "loss": 1.1953,
+      "step": 910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.983050847457628e-06,
+      "loss": 1.2231,
+      "step": 911
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.970013037809648e-06,
+      "loss": 1.2372,
+      "step": 912
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.95697522816167e-06,
+      "loss": 1.1711,
+      "step": 913
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.94393741851369e-06,
+      "loss": 1.2177,
+      "step": 914
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.930899608865712e-06,
+      "loss": 1.1853,
+      "step": 915
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.917861799217732e-06,
+      "loss": 1.2188,
+      "step": 916
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.904823989569754e-06,
+      "loss": 1.1552,
+      "step": 917
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.891786179921774e-06,
+      "loss": 1.251,
+      "step": 918
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.878748370273793e-06,
+      "loss": 1.2551,
+      "step": 919
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.865710560625815e-06,
+      "loss": 1.1763,
+      "step": 920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.852672750977836e-06,
+      "loss": 1.2158,
+      "step": 921
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.839634941329858e-06,
+      "loss": 1.1896,
+      "step": 922
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.826597131681878e-06,
+      "loss": 1.2051,
+      "step": 923
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.8135593220339e-06,
+      "loss": 1.1897,
+      "step": 924
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.80052151238592e-06,
+      "loss": 1.2053,
+      "step": 925
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.78748370273794e-06,
+      "loss": 1.2063,
+      "step": 926
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.77444589308996e-06,
+      "loss": 1.2063,
+      "step": 927
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.761408083441982e-06,
+      "loss": 1.2058,
+      "step": 928
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.748370273794004e-06,
+      "loss": 1.1875,
+      "step": 929
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.735332464146025e-06,
+      "loss": 1.1816,
+      "step": 930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.722294654498045e-06,
+      "loss": 1.1742,
+      "step": 931
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.709256844850067e-06,
+      "loss": 1.2152,
+      "step": 932
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.696219035202086e-06,
+      "loss": 1.2369,
+      "step": 933
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.683181225554108e-06,
+      "loss": 1.212,
+      "step": 934
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.670143415906128e-06,
+      "loss": 1.214,
+      "step": 935
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.65710560625815e-06,
+      "loss": 1.2098,
+      "step": 936
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.64406779661017e-06,
+      "loss": 1.2069,
+      "step": 937
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.63102998696219e-06,
+      "loss": 1.187,
+      "step": 938
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.617992177314212e-06,
+      "loss": 1.1989,
+      "step": 939
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.604954367666232e-06,
+      "loss": 1.2207,
+      "step": 940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.591916558018254e-06,
+      "loss": 1.1844,
+      "step": 941
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.578878748370273e-06,
+      "loss": 1.2087,
+      "step": 942
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.565840938722295e-06,
+      "loss": 1.1847,
+      "step": 943
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.552803129074316e-06,
+      "loss": 1.21,
+      "step": 944
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.539765319426338e-06,
+      "loss": 1.1982,
+      "step": 945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.526727509778358e-06,
+      "loss": 1.1696,
+      "step": 946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.51368970013038e-06,
+      "loss": 1.1874,
+      "step": 947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.5006518904824e-06,
+      "loss": 1.1561,
+      "step": 948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.48761408083442e-06,
+      "loss": 1.1803,
+      "step": 949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.47457627118644e-06,
+      "loss": 1.1912,
+      "step": 950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.2262,
+      "step": 951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.448500651890484e-06,
+      "loss": 1.2173,
+      "step": 952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.435462842242505e-06,
+      "loss": 1.1965,
+      "step": 953
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.422425032594525e-06,
+      "loss": 1.1707,
+      "step": 954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.409387222946545e-06,
+      "loss": 1.1779,
+      "step": 955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.396349413298566e-06,
+      "loss": 1.1865,
+      "step": 956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.383311603650588e-06,
+      "loss": 1.1857,
+      "step": 957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.370273794002608e-06,
+      "loss": 1.2117,
+      "step": 958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.35723598435463e-06,
+      "loss": 1.1899,
+      "step": 959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.34419817470665e-06,
+      "loss": 1.2365,
+      "step": 960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.33116036505867e-06,
+      "loss": 1.187,
+      "step": 961
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.318122555410692e-06,
+      "loss": 1.1976,
+      "step": 962
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.305084745762712e-06,
+      "loss": 1.2101,
+      "step": 963
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.292046936114734e-06,
+      "loss": 1.1792,
+      "step": 964
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.279009126466753e-06,
+      "loss": 1.2073,
+      "step": 965
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.265971316818775e-06,
+      "loss": 1.1805,
+      "step": 966
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.252933507170796e-06,
+      "loss": 1.1981,
+      "step": 967
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.239895697522818e-06,
+      "loss": 1.2075,
+      "step": 968
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.226857887874838e-06,
+      "loss": 1.1788,
+      "step": 969
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.213820078226858e-06,
+      "loss": 1.2166,
+      "step": 970
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.20078226857888e-06,
+      "loss": 1.2134,
+      "step": 971
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.1877444589309e-06,
+      "loss": 1.2015,
+      "step": 972
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.17470664928292e-06,
+      "loss": 1.1886,
+      "step": 973
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.161668839634942e-06,
+      "loss": 1.2329,
+      "step": 974
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.148631029986964e-06,
+      "loss": 1.1957,
+      "step": 975
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.135593220338983e-06,
+      "loss": 1.1719,
+      "step": 976
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.122555410691005e-06,
+      "loss": 1.1837,
+      "step": 977
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.109517601043025e-06,
+      "loss": 1.2113,
+      "step": 978
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.096479791395046e-06,
+      "loss": 1.1625,
+      "step": 979
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.083441981747066e-06,
+      "loss": 1.2124,
+      "step": 980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.070404172099088e-06,
+      "loss": 1.2363,
+      "step": 981
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.05736636245111e-06,
+      "loss": 1.2057,
+      "step": 982
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.04432855280313e-06,
+      "loss": 1.1815,
+      "step": 983
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.03129074315515e-06,
+      "loss": 1.1846,
+      "step": 984
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.01825293350717e-06,
+      "loss": 1.1914,
+      "step": 985
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.005215123859192e-06,
+      "loss": 1.2354,
+      "step": 986
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.992177314211214e-06,
+      "loss": 1.2401,
+      "step": 987
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.979139504563233e-06,
+      "loss": 1.2151,
+      "step": 988
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.966101694915255e-06,
+      "loss": 1.1763,
+      "step": 989
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.953063885267276e-06,
+      "loss": 1.1848,
+      "step": 990
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.940026075619296e-06,
+      "loss": 1.1791,
+      "step": 991
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.926988265971318e-06,
+      "loss": 1.1991,
+      "step": 992
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.913950456323338e-06,
+      "loss": 1.2151,
+      "step": 993
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.90091264667536e-06,
+      "loss": 1.1592,
+      "step": 994
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.88787483702738e-06,
+      "loss": 1.1559,
+      "step": 995
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.8748370273794e-06,
+      "loss": 1.2266,
+      "step": 996
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.861799217731422e-06,
+      "loss": 1.1986,
+      "step": 997
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.848761408083444e-06,
+      "loss": 1.1744,
+      "step": 998
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.835723598435464e-06,
+      "loss": 1.2219,
+      "step": 999
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.822685788787483e-06,
+      "loss": 1.1947,
+      "step": 1000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.809647979139505e-06,
+      "loss": 1.192,
+      "step": 1001
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.796610169491526e-06,
+      "loss": 1.2159,
+      "step": 1002
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.783572359843546e-06,
+      "loss": 1.1974,
+      "step": 1003
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.770534550195568e-06,
+      "loss": 1.2177,
+      "step": 1004
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.75749674054759e-06,
+      "loss": 1.168,
+      "step": 1005
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.744458930899609e-06,
+      "loss": 1.1908,
+      "step": 1006
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.73142112125163e-06,
+      "loss": 1.2009,
+      "step": 1007
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.71838331160365e-06,
+      "loss": 1.2164,
+      "step": 1008
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.705345501955672e-06,
+      "loss": 1.1918,
+      "step": 1009
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.1906,
+      "step": 1010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.679269882659713e-06,
+      "loss": 1.2042,
+      "step": 1011
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.666232073011735e-06,
+      "loss": 1.1899,
+      "step": 1012
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.653194263363756e-06,
+      "loss": 1.1836,
+      "step": 1013
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.640156453715776e-06,
+      "loss": 1.1627,
+      "step": 1014
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.627118644067797e-06,
+      "loss": 1.1722,
+      "step": 1015
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.614080834419818e-06,
+      "loss": 1.2384,
+      "step": 1016
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.601043024771839e-06,
+      "loss": 1.1528,
+      "step": 1017
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.601043024771839e-06,
+      "loss": 1.2067,
+      "step": 1018
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.58800521512386e-06,
+      "loss": 1.196,
+      "step": 1019
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.574967405475881e-06,
+      "loss": 1.1886,
+      "step": 1020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.561929595827901e-06,
+      "loss": 1.225,
+      "step": 1021
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.548891786179923e-06,
+      "loss": 1.2195,
+      "step": 1022
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.535853976531943e-06,
+      "loss": 1.1768,
+      "step": 1023
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.522816166883964e-06,
+      "loss": 1.1818,
+      "step": 1024
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.509778357235985e-06,
+      "loss": 1.1764,
+      "step": 1025
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.496740547588006e-06,
+      "loss": 1.1895,
+      "step": 1026
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.483702737940026e-06,
+      "loss": 1.2224,
+      "step": 1027
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.470664928292048e-06,
+      "loss": 1.1943,
+      "step": 1028
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.4576271186440685e-06,
+      "loss": 1.2083,
+      "step": 1029
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.44458930899609e-06,
+      "loss": 1.2417,
+      "step": 1030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.43155149934811e-06,
+      "loss": 1.2249,
+      "step": 1031
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.4185136897001305e-06,
+      "loss": 1.2104,
+      "step": 1032
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.405475880052152e-06,
+      "loss": 1.2188,
+      "step": 1033
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.392438070404173e-06,
+      "loss": 1.2273,
+      "step": 1034
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.3794002607561934e-06,
+      "loss": 1.158,
+      "step": 1035
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.366362451108214e-06,
+      "loss": 1.1852,
+      "step": 1036
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.353324641460236e-06,
+      "loss": 1.2142,
+      "step": 1037
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.340286831812256e-06,
+      "loss": 1.1689,
+      "step": 1038
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.327249022164277e-06,
+      "loss": 1.2256,
+      "step": 1039
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.314211212516298e-06,
+      "loss": 1.2034,
+      "step": 1040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.301173402868319e-06,
+      "loss": 1.2177,
+      "step": 1041
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.288135593220339e-06,
+      "loss": 1.1689,
+      "step": 1042
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.27509778357236e-06,
+      "loss": 1.1826,
+      "step": 1043
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.262059973924381e-06,
+      "loss": 1.1881,
+      "step": 1044
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.249022164276403e-06,
+      "loss": 1.1848,
+      "step": 1045
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.235984354628423e-06,
+      "loss": 1.1866,
+      "step": 1046
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.222946544980443e-06,
+      "loss": 1.2186,
+      "step": 1047
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.209908735332465e-06,
+      "loss": 1.2205,
+      "step": 1048
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.196870925684486e-06,
+      "loss": 1.2001,
+      "step": 1049
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.183833116036506e-06,
+      "loss": 1.2109,
+      "step": 1050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.170795306388527e-06,
+      "loss": 1.1827,
+      "step": 1051
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.1577574967405485e-06,
+      "loss": 1.1857,
+      "step": 1052
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.144719687092569e-06,
+      "loss": 1.2012,
+      "step": 1053
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.13168187744459e-06,
+      "loss": 1.21,
+      "step": 1054
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.1186440677966106e-06,
+      "loss": 1.2158,
+      "step": 1055
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.105606258148632e-06,
+      "loss": 1.2012,
+      "step": 1056
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.092568448500653e-06,
+      "loss": 1.1998,
+      "step": 1057
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.079530638852673e-06,
+      "loss": 1.1769,
+      "step": 1058
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.066492829204694e-06,
+      "loss": 1.1912,
+      "step": 1059
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.053455019556716e-06,
+      "loss": 1.2153,
+      "step": 1060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.040417209908736e-06,
+      "loss": 1.1779,
+      "step": 1061
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.027379400260756e-06,
+      "loss": 1.1943,
+      "step": 1062
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.014341590612778e-06,
+      "loss": 1.1835,
+      "step": 1063
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.001303780964798e-06,
+      "loss": 1.21,
+      "step": 1064
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.988265971316819e-06,
+      "loss": 1.1901,
+      "step": 1065
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.97522816166884e-06,
+      "loss": 1.2056,
+      "step": 1066
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.962190352020861e-06,
+      "loss": 1.1823,
+      "step": 1067
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.949152542372882e-06,
+      "loss": 1.1978,
+      "step": 1068
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.936114732724903e-06,
+      "loss": 1.1902,
+      "step": 1069
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.204,
+      "step": 1070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.910039113428945e-06,
+      "loss": 1.2189,
+      "step": 1071
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.897001303780966e-06,
+      "loss": 1.2075,
+      "step": 1072
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.8839634941329854e-06,
+      "loss": 1.2048,
+      "step": 1073
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.870925684485007e-06,
+      "loss": 1.1964,
+      "step": 1074
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.8578878748370285e-06,
+      "loss": 1.2133,
+      "step": 1075
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.844850065189049e-06,
+      "loss": 1.1874,
+      "step": 1076
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.831812255541069e-06,
+      "loss": 1.1852,
+      "step": 1077
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.8187744458930906e-06,
+      "loss": 1.1666,
+      "step": 1078
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.805736636245111e-06,
+      "loss": 1.2155,
+      "step": 1079
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.792698826597133e-06,
+      "loss": 1.2058,
+      "step": 1080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.779661016949153e-06,
+      "loss": 1.1555,
+      "step": 1081
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.766623207301174e-06,
+      "loss": 1.1705,
+      "step": 1082
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.753585397653195e-06,
+      "loss": 1.1735,
+      "step": 1083
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.7405475880052155e-06,
+      "loss": 1.2366,
+      "step": 1084
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.727509778357236e-06,
+      "loss": 1.2048,
+      "step": 1085
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.714471968709258e-06,
+      "loss": 1.2014,
+      "step": 1086
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.7014341590612784e-06,
+      "loss": 1.1934,
+      "step": 1087
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.688396349413298e-06,
+      "loss": 1.1771,
+      "step": 1088
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.67535853976532e-06,
+      "loss": 1.1985,
+      "step": 1089
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.662320730117341e-06,
+      "loss": 1.1393,
+      "step": 1090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.649282920469362e-06,
+      "loss": 1.1864,
+      "step": 1091
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.636245110821382e-06,
+      "loss": 1.1731,
+      "step": 1092
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.623207301173403e-06,
+      "loss": 1.1901,
+      "step": 1093
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.610169491525424e-06,
+      "loss": 1.1917,
+      "step": 1094
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.597131681877446e-06,
+      "loss": 1.2357,
+      "step": 1095
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.5840938722294655e-06,
+      "loss": 1.1979,
+      "step": 1096
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.571056062581487e-06,
+      "loss": 1.1975,
+      "step": 1097
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.558018252933508e-06,
+      "loss": 1.1771,
+      "step": 1098
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.544980443285529e-06,
+      "loss": 1.2054,
+      "step": 1099
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.531942633637549e-06,
+      "loss": 1.177,
+      "step": 1100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.518904823989571e-06,
+      "loss": 1.1816,
+      "step": 1101
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.505867014341591e-06,
+      "loss": 1.2006,
+      "step": 1102
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.492829204693613e-06,
+      "loss": 1.2091,
+      "step": 1103
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.479791395045633e-06,
+      "loss": 1.1818,
+      "step": 1104
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.466753585397653e-06,
+      "loss": 1.2112,
+      "step": 1105
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.453715775749675e-06,
+      "loss": 1.1914,
+      "step": 1106
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.440677966101695e-06,
+      "loss": 1.2104,
+      "step": 1107
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.427640156453716e-06,
+      "loss": 1.1818,
+      "step": 1108
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.414602346805737e-06,
+      "loss": 1.1943,
+      "step": 1109
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4015645371577584e-06,
+      "loss": 1.2052,
+      "step": 1110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.388526727509778e-06,
+      "loss": 1.184,
+      "step": 1111
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3754889178618e-06,
+      "loss": 1.2203,
+      "step": 1112
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3624511082138205e-06,
+      "loss": 1.1899,
+      "step": 1113
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.349413298565842e-06,
+      "loss": 1.1703,
+      "step": 1114
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.336375488917862e-06,
+      "loss": 1.2048,
+      "step": 1115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.323337679269883e-06,
+      "loss": 1.2175,
+      "step": 1116
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.310299869621904e-06,
+      "loss": 1.173,
+      "step": 1117
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.297262059973926e-06,
+      "loss": 1.1984,
+      "step": 1118
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.2842242503259455e-06,
+      "loss": 1.1766,
+      "step": 1119
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.271186440677966e-06,
+      "loss": 1.1694,
+      "step": 1120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.258148631029988e-06,
+      "loss": 1.1685,
+      "step": 1121
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.245110821382009e-06,
+      "loss": 1.2151,
+      "step": 1122
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.232073011734029e-06,
+      "loss": 1.1942,
+      "step": 1123
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.21903520208605e-06,
+      "loss": 1.2227,
+      "step": 1124
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.205997392438071e-06,
+      "loss": 1.1964,
+      "step": 1125
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.192959582790091e-06,
+      "loss": 1.1869,
+      "step": 1126
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.179921773142113e-06,
+      "loss": 1.1987,
+      "step": 1127
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.166883963494133e-06,
+      "loss": 1.1984,
+      "step": 1128
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.1937,
+      "step": 1129
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.140808344198175e-06,
+      "loss": 1.1788,
+      "step": 1130
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.127770534550196e-06,
+      "loss": 1.1732,
+      "step": 1131
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.114732724902217e-06,
+      "loss": 1.1809,
+      "step": 1132
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.1016949152542385e-06,
+      "loss": 1.2456,
+      "step": 1133
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.088657105606258e-06,
+      "loss": 1.2018,
+      "step": 1134
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.075619295958279e-06,
+      "loss": 1.2062,
+      "step": 1135
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.0625814863103005e-06,
+      "loss": 1.1908,
+      "step": 1136
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.049543676662322e-06,
+      "loss": 1.1909,
+      "step": 1137
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.036505867014342e-06,
+      "loss": 1.1956,
+      "step": 1138
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.023468057366363e-06,
+      "loss": 1.1804,
+      "step": 1139
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.010430247718384e-06,
+      "loss": 1.1968,
+      "step": 1140
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.997392438070405e-06,
+      "loss": 1.179,
+      "step": 1141
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.9843546284224255e-06,
+      "loss": 1.1948,
+      "step": 1142
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.971316818774446e-06,
+      "loss": 1.2008,
+      "step": 1143
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.958279009126468e-06,
+      "loss": 1.1534,
+      "step": 1144
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.9452411994784875e-06,
+      "loss": 1.1687,
+      "step": 1145
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.932203389830509e-06,
+      "loss": 1.2438,
+      "step": 1146
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.91916558018253e-06,
+      "loss": 1.1913,
+      "step": 1147
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.906127770534551e-06,
+      "loss": 1.1941,
+      "step": 1148
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.893089960886571e-06,
+      "loss": 1.2301,
+      "step": 1149
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.880052151238592e-06,
+      "loss": 1.1956,
+      "step": 1150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.867014341590613e-06,
+      "loss": 1.2184,
+      "step": 1151
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.853976531942635e-06,
+      "loss": 1.2043,
+      "step": 1152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.840938722294655e-06,
+      "loss": 1.2234,
+      "step": 1153
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.827900912646675e-06,
+      "loss": 1.1857,
+      "step": 1154
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.814863102998697e-06,
+      "loss": 1.194,
+      "step": 1155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.801825293350718e-06,
+      "loss": 1.203,
+      "step": 1156
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.788787483702738e-06,
+      "loss": 1.1868,
+      "step": 1157
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.775749674054759e-06,
+      "loss": 1.2185,
+      "step": 1158
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.7627118644067805e-06,
+      "loss": 1.194,
+      "step": 1159
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.749674054758801e-06,
+      "loss": 1.1831,
+      "step": 1160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.736636245110822e-06,
+      "loss": 1.1635,
+      "step": 1161
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.723598435462843e-06,
+      "loss": 1.2037,
+      "step": 1162
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.710560625814864e-06,
+      "loss": 1.2056,
+      "step": 1163
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.697522816166885e-06,
+      "loss": 1.1947,
+      "step": 1164
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.684485006518905e-06,
+      "loss": 1.1895,
+      "step": 1165
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.671447196870926e-06,
+      "loss": 1.1859,
+      "step": 1166
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.658409387222948e-06,
+      "loss": 1.2039,
+      "step": 1167
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.6453715775749676e-06,
+      "loss": 1.1975,
+      "step": 1168
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.632333767926988e-06,
+      "loss": 1.1757,
+      "step": 1169
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.61929595827901e-06,
+      "loss": 1.1947,
+      "step": 1170
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.6062581486310305e-06,
+      "loss": 1.2039,
+      "step": 1171
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.593220338983051e-06,
+      "loss": 1.2054,
+      "step": 1172
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.580182529335072e-06,
+      "loss": 1.1926,
+      "step": 1173
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.567144719687093e-06,
+      "loss": 1.2003,
+      "step": 1174
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.554106910039114e-06,
+      "loss": 1.2111,
+      "step": 1175
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.541069100391135e-06,
+      "loss": 1.1934,
+      "step": 1176
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.5280312907431554e-06,
+      "loss": 1.2058,
+      "step": 1177
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.514993481095177e-06,
+      "loss": 1.1711,
+      "step": 1178
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.501955671447198e-06,
+      "loss": 1.1561,
+      "step": 1179
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.4889178617992175e-06,
+      "loss": 1.207,
+      "step": 1180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.475880052151239e-06,
+      "loss": 1.2012,
+      "step": 1181
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.46284224250326e-06,
+      "loss": 1.2057,
+      "step": 1182
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.449804432855281e-06,
+      "loss": 1.2152,
+      "step": 1183
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.436766623207301e-06,
+      "loss": 1.1687,
+      "step": 1184
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.423728813559323e-06,
+      "loss": 1.1841,
+      "step": 1185
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.410691003911343e-06,
+      "loss": 1.2261,
+      "step": 1186
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.397653194263364e-06,
+      "loss": 1.2334,
+      "step": 1187
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.2207,
+      "step": 1188
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.371577574967406e-06,
+      "loss": 1.2203,
+      "step": 1189
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.358539765319427e-06,
+      "loss": 1.1813,
+      "step": 1190
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.345501955671447e-06,
+      "loss": 1.1995,
+      "step": 1191
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.332464146023468e-06,
+      "loss": 1.2142,
+      "step": 1192
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.31942633637549e-06,
+      "loss": 1.1738,
+      "step": 1193
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.3063885267275105e-06,
+      "loss": 1.1791,
+      "step": 1194
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.29335071707953e-06,
+      "loss": 1.1622,
+      "step": 1195
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.280312907431552e-06,
+      "loss": 1.1842,
+      "step": 1196
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.2672750977835725e-06,
+      "loss": 1.1971,
+      "step": 1197
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.254237288135594e-06,
+      "loss": 1.1942,
+      "step": 1198
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.241199478487614e-06,
+      "loss": 1.2006,
+      "step": 1199
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.2281616688396354e-06,
+      "loss": 1.1666,
+      "step": 1200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.215123859191656e-06,
+      "loss": 1.2139,
+      "step": 1201
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.202086049543678e-06,
+      "loss": 1.2119,
+      "step": 1202
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.1890482398956975e-06,
+      "loss": 1.1692,
+      "step": 1203
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.176010430247719e-06,
+      "loss": 1.1866,
+      "step": 1204
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.16297262059974e-06,
+      "loss": 1.1819,
+      "step": 1205
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.149934810951761e-06,
+      "loss": 1.1796,
+      "step": 1206
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.136897001303781e-06,
+      "loss": 1.2026,
+      "step": 1207
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.123859191655803e-06,
+      "loss": 1.1776,
+      "step": 1208
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.110821382007823e-06,
+      "loss": 1.1797,
+      "step": 1209
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.097783572359843e-06,
+      "loss": 1.1918,
+      "step": 1210
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.084745762711865e-06,
+      "loss": 1.2203,
+      "step": 1211
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.071707953063885e-06,
+      "loss": 1.2372,
+      "step": 1212
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.058670143415907e-06,
+      "loss": 1.205,
+      "step": 1213
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.045632333767927e-06,
+      "loss": 1.2125,
+      "step": 1214
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.032594524119948e-06,
+      "loss": 1.1992,
+      "step": 1215
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.019556714471969e-06,
+      "loss": 1.2153,
+      "step": 1216
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.0065189048239905e-06,
+      "loss": 1.2025,
+      "step": 1217
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.993481095176011e-06,
+      "loss": 1.1792,
+      "step": 1218
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.980443285528032e-06,
+      "loss": 1.2062,
+      "step": 1219
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.9674054758800525e-06,
+      "loss": 1.172,
+      "step": 1220
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.954367666232073e-06,
+      "loss": 1.1832,
+      "step": 1221
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.941329856584095e-06,
+      "loss": 1.17,
+      "step": 1222
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.9282920469361155e-06,
+      "loss": 1.2041,
+      "step": 1223
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.915254237288136e-06,
+      "loss": 1.2073,
+      "step": 1224
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.902216427640157e-06,
+      "loss": 1.1891,
+      "step": 1225
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.889178617992178e-06,
+      "loss": 1.187,
+      "step": 1226
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.876140808344198e-06,
+      "loss": 1.1746,
+      "step": 1227
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.863102998696219e-06,
+      "loss": 1.1901,
+      "step": 1228
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.85006518904824e-06,
+      "loss": 1.2081,
+      "step": 1229
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.837027379400261e-06,
+      "loss": 1.1875,
+      "step": 1230
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.823989569752282e-06,
+      "loss": 1.1858,
+      "step": 1231
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.8109517601043025e-06,
+      "loss": 1.1667,
+      "step": 1232
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.797913950456324e-06,
+      "loss": 1.1757,
+      "step": 1233
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.784876140808345e-06,
+      "loss": 1.1704,
+      "step": 1234
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.771838331160365e-06,
+      "loss": 1.1675,
+      "step": 1235
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.758800521512386e-06,
+      "loss": 1.2047,
+      "step": 1236
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.745762711864408e-06,
+      "loss": 1.1677,
+      "step": 1237
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.732724902216428e-06,
+      "loss": 1.201,
+      "step": 1238
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.719687092568449e-06,
+      "loss": 1.187,
+      "step": 1239
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.70664928292047e-06,
+      "loss": 1.2085,
+      "step": 1240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.693611473272491e-06,
+      "loss": 1.1641,
+      "step": 1241
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.680573663624511e-06,
+      "loss": 1.1724,
+      "step": 1242
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.6675358539765326e-06,
+      "loss": 1.2327,
+      "step": 1243
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.654498044328553e-06,
+      "loss": 1.1848,
+      "step": 1244
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.641460234680574e-06,
+      "loss": 1.2128,
+      "step": 1245
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.628422425032595e-06,
+      "loss": 1.1923,
+      "step": 1246
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.1541,
+      "step": 1247
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.602346805736637e-06,
+      "loss": 1.1898,
+      "step": 1248
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.5893089960886575e-06,
+      "loss": 1.1731,
+      "step": 1249
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.576271186440678e-06,
+      "loss": 1.1793,
+      "step": 1250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.563233376792699e-06,
+      "loss": 1.1642,
+      "step": 1251
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.5501955671447204e-06,
+      "loss": 1.2009,
+      "step": 1252
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.537157757496741e-06,
+      "loss": 1.2172,
+      "step": 1253
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.524119947848762e-06,
+      "loss": 1.2015,
+      "step": 1254
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.5110821382007825e-06,
+      "loss": 1.2007,
+      "step": 1255
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.498044328552803e-06,
+      "loss": 1.1848,
+      "step": 1256
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.485006518904824e-06,
+      "loss": 1.1627,
+      "step": 1257
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.471968709256845e-06,
+      "loss": 1.168,
+      "step": 1258
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.458930899608866e-06,
+      "loss": 1.1614,
+      "step": 1259
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.445893089960887e-06,
+      "loss": 1.1678,
+      "step": 1260
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4328552803129075e-06,
+      "loss": 1.1947,
+      "step": 1261
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.419817470664929e-06,
+      "loss": 1.2125,
+      "step": 1262
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.40677966101695e-06,
+      "loss": 1.165,
+      "step": 1263
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.39374185136897e-06,
+      "loss": 1.1917,
+      "step": 1264
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.380704041720991e-06,
+      "loss": 1.2078,
+      "step": 1265
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.3676662320730126e-06,
+      "loss": 1.196,
+      "step": 1266
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.354628422425033e-06,
+      "loss": 1.1866,
+      "step": 1267
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.341590612777054e-06,
+      "loss": 1.1757,
+      "step": 1268
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.328552803129075e-06,
+      "loss": 1.1445,
+      "step": 1269
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.315514993481095e-06,
+      "loss": 1.2147,
+      "step": 1270
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.302477183833116e-06,
+      "loss": 1.1858,
+      "step": 1271
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.289439374185137e-06,
+      "loss": 1.2328,
+      "step": 1272
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.276401564537158e-06,
+      "loss": 1.1981,
+      "step": 1273
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.263363754889179e-06,
+      "loss": 1.1903,
+      "step": 1274
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2503259452412e-06,
+      "loss": 1.1758,
+      "step": 1275
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.23728813559322e-06,
+      "loss": 1.1921,
+      "step": 1276
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.224250325945242e-06,
+      "loss": 1.1982,
+      "step": 1277
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.2112125162972625e-06,
+      "loss": 1.1899,
+      "step": 1278
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.198174706649283e-06,
+      "loss": 1.1523,
+      "step": 1279
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.185136897001304e-06,
+      "loss": 1.1947,
+      "step": 1280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.172099087353325e-06,
+      "loss": 1.2357,
+      "step": 1281
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.159061277705346e-06,
+      "loss": 1.2098,
+      "step": 1282
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.146023468057367e-06,
+      "loss": 1.2078,
+      "step": 1283
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1329856584093875e-06,
+      "loss": 1.2032,
+      "step": 1284
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.119947848761409e-06,
+      "loss": 1.188,
+      "step": 1285
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.106910039113429e-06,
+      "loss": 1.1934,
+      "step": 1286
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.09387222946545e-06,
+      "loss": 1.1556,
+      "step": 1287
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.080834419817471e-06,
+      "loss": 1.2279,
+      "step": 1288
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.067796610169492e-06,
+      "loss": 1.1532,
+      "step": 1289
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.0547588005215124e-06,
+      "loss": 1.1934,
+      "step": 1290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.041720990873533e-06,
+      "loss": 1.1927,
+      "step": 1291
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.028683181225555e-06,
+      "loss": 1.1898,
+      "step": 1292
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.015645371577575e-06,
+      "loss": 1.1904,
+      "step": 1293
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.002607561929596e-06,
+      "loss": 1.1912,
+      "step": 1294
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.989569752281617e-06,
+      "loss": 1.2081,
+      "step": 1295
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.976531942633638e-06,
+      "loss": 1.1908,
+      "step": 1296
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.963494132985659e-06,
+      "loss": 1.2009,
+      "step": 1297
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.95045632333768e-06,
+      "loss": 1.1975,
+      "step": 1298
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9374185136897e-06,
+      "loss": 1.1805,
+      "step": 1299
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.924380704041722e-06,
+      "loss": 1.187,
+      "step": 1300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.911342894393742e-06,
+      "loss": 1.191,
+      "step": 1301
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.898305084745763e-06,
+      "loss": 1.1613,
+      "step": 1302
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.885267275097784e-06,
+      "loss": 1.1918,
+      "step": 1303
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8722294654498046e-06,
+      "loss": 1.1548,
+      "step": 1304
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.859191655801825e-06,
+      "loss": 1.1647,
+      "step": 1305
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.174,
+      "step": 1306
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8331160365058675e-06,
+      "loss": 1.1719,
+      "step": 1307
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.820078226857888e-06,
+      "loss": 1.1868,
+      "step": 1308
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.807040417209909e-06,
+      "loss": 1.1511,
+      "step": 1309
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.79400260756193e-06,
+      "loss": 1.1808,
+      "step": 1310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.7809647979139506e-06,
+      "loss": 1.1946,
+      "step": 1311
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.7679269882659713e-06,
+      "loss": 1.2063,
+      "step": 1312
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.7548891786179924e-06,
+      "loss": 1.2103,
+      "step": 1313
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.741851368970013e-06,
+      "loss": 1.1876,
+      "step": 1314
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.7288135593220342e-06,
+      "loss": 1.1633,
+      "step": 1315
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.715775749674055e-06,
+      "loss": 1.2017,
+      "step": 1316
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.702737940026076e-06,
+      "loss": 1.1765,
+      "step": 1317
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.6897001303780967e-06,
+      "loss": 1.1987,
+      "step": 1318
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.676662320730118e-06,
+      "loss": 1.188,
+      "step": 1319
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.6636245110821385e-06,
+      "loss": 1.2144,
+      "step": 1320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.6505867014341596e-06,
+      "loss": 1.1898,
+      "step": 1321
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.63754889178618e-06,
+      "loss": 1.1832,
+      "step": 1322
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.6245110821382014e-06,
+      "loss": 1.2081,
+      "step": 1323
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.6114732724902217e-06,
+      "loss": 1.1624,
+      "step": 1324
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.598435462842243e-06,
+      "loss": 1.1864,
+      "step": 1325
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.5853976531942635e-06,
+      "loss": 1.2061,
+      "step": 1326
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.5723598435462846e-06,
+      "loss": 1.2046,
+      "step": 1327
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.5593220338983053e-06,
+      "loss": 1.2097,
+      "step": 1328
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.5462842242503264e-06,
+      "loss": 1.1588,
+      "step": 1329
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.533246414602347e-06,
+      "loss": 1.174,
+      "step": 1330
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.520208604954368e-06,
+      "loss": 1.1781,
+      "step": 1331
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.507170795306389e-06,
+      "loss": 1.1865,
+      "step": 1332
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4941329856584096e-06,
+      "loss": 1.1814,
+      "step": 1333
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4810951760104307e-06,
+      "loss": 1.1775,
+      "step": 1334
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4680573663624513e-06,
+      "loss": 1.1725,
+      "step": 1335
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4550195567144725e-06,
+      "loss": 1.1921,
+      "step": 1336
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4419817470664927e-06,
+      "loss": 1.1787,
+      "step": 1337
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4289439374185143e-06,
+      "loss": 1.1797,
+      "step": 1338
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4159061277705345e-06,
+      "loss": 1.2207,
+      "step": 1339
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.4028683181225556e-06,
+      "loss": 1.1998,
+      "step": 1340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3898305084745763e-06,
+      "loss": 1.2008,
+      "step": 1341
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3767926988265974e-06,
+      "loss": 1.1925,
+      "step": 1342
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.363754889178618e-06,
+      "loss": 1.1863,
+      "step": 1343
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3507170795306392e-06,
+      "loss": 1.1804,
+      "step": 1344
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.33767926988266e-06,
+      "loss": 1.1815,
+      "step": 1345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.324641460234681e-06,
+      "loss": 1.1699,
+      "step": 1346
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3116036505867017e-06,
+      "loss": 1.2013,
+      "step": 1347
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.298565840938723e-06,
+      "loss": 1.2656,
+      "step": 1348
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2855280312907435e-06,
+      "loss": 1.2196,
+      "step": 1349
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2724902216427646e-06,
+      "loss": 1.2217,
+      "step": 1350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2594524119947853e-06,
+      "loss": 1.1862,
+      "step": 1351
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2464146023468064e-06,
+      "loss": 1.1954,
+      "step": 1352
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2333767926988267e-06,
+      "loss": 1.2104,
+      "step": 1353
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2203389830508473e-06,
+      "loss": 1.1951,
+      "step": 1354
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2073011734028685e-06,
+      "loss": 1.1914,
+      "step": 1355
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.194263363754889e-06,
+      "loss": 1.2131,
+      "step": 1356
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1812255541069103e-06,
+      "loss": 1.1925,
+      "step": 1357
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.168187744458931e-06,
+      "loss": 1.1793,
+      "step": 1358
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.155149934810952e-06,
+      "loss": 1.1929,
+      "step": 1359
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1421121251629727e-06,
+      "loss": 1.1423,
+      "step": 1360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.129074315514994e-06,
+      "loss": 1.2283,
+      "step": 1361
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1160365058670145e-06,
+      "loss": 1.1735,
+      "step": 1362
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1029986962190356e-06,
+      "loss": 1.1698,
+      "step": 1363
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.0899608865710563e-06,
+      "loss": 1.1942,
+      "step": 1364
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.1683,
+      "step": 1365
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.063885267275098e-06,
+      "loss": 1.1975,
+      "step": 1366
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.0508474576271192e-06,
+      "loss": 1.1697,
+      "step": 1367
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.0378096479791395e-06,
+      "loss": 1.1753,
+      "step": 1368
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.024771838331161e-06,
+      "loss": 1.2202,
+      "step": 1369
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.0117340286831813e-06,
+      "loss": 1.1853,
+      "step": 1370
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.9986962190352024e-06,
+      "loss": 1.174,
+      "step": 1371
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.985658409387223e-06,
+      "loss": 1.2139,
+      "step": 1372
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.9726205997392438e-06,
+      "loss": 1.1758,
+      "step": 1373
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.959582790091265e-06,
+      "loss": 1.1877,
+      "step": 1374
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.9465449804432856e-06,
+      "loss": 1.1782,
+      "step": 1375
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.9335071707953067e-06,
+      "loss": 1.2385,
+      "step": 1376
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.9204693611473274e-06,
+      "loss": 1.1855,
+      "step": 1377
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.9074315514993485e-06,
+      "loss": 1.1426,
+      "step": 1378
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.894393741851369e-06,
+      "loss": 1.1982,
+      "step": 1379
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8813559322033903e-06,
+      "loss": 1.2119,
+      "step": 1380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.868318122555411e-06,
+      "loss": 1.172,
+      "step": 1381
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.855280312907432e-06,
+      "loss": 1.1364,
+      "step": 1382
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8422425032594523e-06,
+      "loss": 1.1688,
+      "step": 1383
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.829204693611474e-06,
+      "loss": 1.1736,
+      "step": 1384
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.816166883963494e-06,
+      "loss": 1.1987,
+      "step": 1385
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8031290743155152e-06,
+      "loss": 1.2017,
+      "step": 1386
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.790091264667536e-06,
+      "loss": 1.1969,
+      "step": 1387
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.777053455019557e-06,
+      "loss": 1.2157,
+      "step": 1388
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7640156453715777e-06,
+      "loss": 1.1622,
+      "step": 1389
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.750977835723599e-06,
+      "loss": 1.2345,
+      "step": 1390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7379400260756195e-06,
+      "loss": 1.176,
+      "step": 1391
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7249022164276406e-06,
+      "loss": 1.2141,
+      "step": 1392
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7118644067796613e-06,
+      "loss": 1.2273,
+      "step": 1393
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.698826597131682e-06,
+      "loss": 1.1875,
+      "step": 1394
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.685788787483703e-06,
+      "loss": 1.2201,
+      "step": 1395
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.6727509778357234e-06,
+      "loss": 1.1882,
+      "step": 1396
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.659713168187745e-06,
+      "loss": 1.1979,
+      "step": 1397
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.646675358539765e-06,
+      "loss": 1.1603,
+      "step": 1398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.6336375488917863e-06,
+      "loss": 1.1632,
+      "step": 1399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.620599739243807e-06,
+      "loss": 1.1814,
+      "step": 1400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.607561929595828e-06,
+      "loss": 1.1837,
+      "step": 1401
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.5945241199478487e-06,
+      "loss": 1.1851,
+      "step": 1402
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.58148631029987e-06,
+      "loss": 1.1954,
+      "step": 1403
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.5684485006518905e-06,
+      "loss": 1.1973,
+      "step": 1404
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.5554106910039117e-06,
+      "loss": 1.2025,
+      "step": 1405
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.5423728813559323e-06,
+      "loss": 1.1947,
+      "step": 1406
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.5293350717079534e-06,
+      "loss": 1.1947,
+      "step": 1407
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.516297262059974e-06,
+      "loss": 1.1765,
+      "step": 1408
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.5032594524119952e-06,
+      "loss": 1.1755,
+      "step": 1409
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.490221642764016e-06,
+      "loss": 1.201,
+      "step": 1410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4771838331160366e-06,
+      "loss": 1.1942,
+      "step": 1411
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4641460234680577e-06,
+      "loss": 1.1595,
+      "step": 1412
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4511082138200784e-06,
+      "loss": 1.1615,
+      "step": 1413
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.438070404172099e-06,
+      "loss": 1.1903,
+      "step": 1414
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.42503259452412e-06,
+      "loss": 1.176,
+      "step": 1415
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.411994784876141e-06,
+      "loss": 1.1791,
+      "step": 1416
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.398956975228162e-06,
+      "loss": 1.2268,
+      "step": 1417
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.3859191655801827e-06,
+      "loss": 1.1869,
+      "step": 1418
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.372881355932204e-06,
+      "loss": 1.1893,
+      "step": 1419
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.3598435462842245e-06,
+      "loss": 1.1912,
+      "step": 1420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.3468057366362456e-06,
+      "loss": 1.1648,
+      "step": 1421
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.3337679269882663e-06,
+      "loss": 1.1484,
+      "step": 1422
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.320730117340287e-06,
+      "loss": 1.1672,
+      "step": 1423
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.1753,
+      "step": 1424
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2946544980443288e-06,
+      "loss": 1.181,
+      "step": 1425
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2816166883963494e-06,
+      "loss": 1.1902,
+      "step": 1426
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2685788787483706e-06,
+      "loss": 1.1926,
+      "step": 1427
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2555410691003912e-06,
+      "loss": 1.2024,
+      "step": 1428
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.242503259452412e-06,
+      "loss": 1.2125,
+      "step": 1429
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.229465449804433e-06,
+      "loss": 1.2313,
+      "step": 1430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2164276401564537e-06,
+      "loss": 1.2059,
+      "step": 1431
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.203389830508475e-06,
+      "loss": 1.161,
+      "step": 1432
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1903520208604955e-06,
+      "loss": 1.1731,
+      "step": 1433
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1773142112125166e-06,
+      "loss": 1.1658,
+      "step": 1434
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1642764015645373e-06,
+      "loss": 1.1925,
+      "step": 1435
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.151238591916558e-06,
+      "loss": 1.1989,
+      "step": 1436
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.138200782268579e-06,
+      "loss": 1.1588,
+      "step": 1437
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1251629726206e-06,
+      "loss": 1.1781,
+      "step": 1438
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.112125162972621e-06,
+      "loss": 1.1814,
+      "step": 1439
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0990873533246416e-06,
+      "loss": 1.1493,
+      "step": 1440
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0860495436766627e-06,
+      "loss": 1.2012,
+      "step": 1441
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0730117340286834e-06,
+      "loss": 1.1758,
+      "step": 1442
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0599739243807045e-06,
+      "loss": 1.1429,
+      "step": 1443
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.046936114732725e-06,
+      "loss": 1.1812,
+      "step": 1444
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.033898305084746e-06,
+      "loss": 1.1719,
+      "step": 1445
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0208604954367666e-06,
+      "loss": 1.202,
+      "step": 1446
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0078226857887877e-06,
+      "loss": 1.235,
+      "step": 1447
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9947848761408084e-06,
+      "loss": 1.1686,
+      "step": 1448
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9817470664928295e-06,
+      "loss": 1.1733,
+      "step": 1449
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.96870925684485e-06,
+      "loss": 1.204,
+      "step": 1450
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.955671447196871e-06,
+      "loss": 1.2109,
+      "step": 1451
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.942633637548892e-06,
+      "loss": 1.1962,
+      "step": 1452
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9295958279009126e-06,
+      "loss": 1.2164,
+      "step": 1453
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9165580182529337e-06,
+      "loss": 1.1941,
+      "step": 1454
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9035202086049544e-06,
+      "loss": 1.1825,
+      "step": 1455
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8904823989569753e-06,
+      "loss": 1.1798,
+      "step": 1456
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8774445893089962e-06,
+      "loss": 1.2008,
+      "step": 1457
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8644067796610171e-06,
+      "loss": 1.197,
+      "step": 1458
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.851368970013038e-06,
+      "loss": 1.17,
+      "step": 1459
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.838331160365059e-06,
+      "loss": 1.2112,
+      "step": 1460
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8252933507170798e-06,
+      "loss": 1.21,
+      "step": 1461
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8122555410691007e-06,
+      "loss": 1.2124,
+      "step": 1462
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.7992177314211214e-06,
+      "loss": 1.1791,
+      "step": 1463
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.7861799217731423e-06,
+      "loss": 1.1984,
+      "step": 1464
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.7731421121251632e-06,
+      "loss": 1.186,
+      "step": 1465
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.760104302477184e-06,
+      "loss": 1.1791,
+      "step": 1466
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.7470664928292048e-06,
+      "loss": 1.2003,
+      "step": 1467
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.7340286831812257e-06,
+      "loss": 1.2053,
+      "step": 1468
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.7209908735332464e-06,
+      "loss": 1.1984,
+      "step": 1469
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.7079530638852673e-06,
+      "loss": 1.1827,
+      "step": 1470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6949152542372882e-06,
+      "loss": 1.2075,
+      "step": 1471
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.681877444589309e-06,
+      "loss": 1.1782,
+      "step": 1472
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.66883963494133e-06,
+      "loss": 1.1774,
+      "step": 1473
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6558018252933508e-06,
+      "loss": 1.2073,
+      "step": 1474
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6427640156453717e-06,
+      "loss": 1.2136,
+      "step": 1475
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6297262059973926e-06,
+      "loss": 1.1663,
+      "step": 1476
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6166883963494133e-06,
+      "loss": 1.1614,
+      "step": 1477
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6036505867014342e-06,
+      "loss": 1.1744,
+      "step": 1478
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.5906127770534551e-06,
+      "loss": 1.1858,
+      "step": 1479
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.577574967405476e-06,
+      "loss": 1.1733,
+      "step": 1480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.564537157757497e-06,
+      "loss": 1.2163,
+      "step": 1481
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.5514993481095178e-06,
+      "loss": 1.1982,
+      "step": 1482
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.2268,
+      "step": 1483
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.5254237288135596e-06,
+      "loss": 1.229,
+      "step": 1484
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.5123859191655805e-06,
+      "loss": 1.2092,
+      "step": 1485
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4993481095176012e-06,
+      "loss": 1.213,
+      "step": 1486
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4863102998696219e-06,
+      "loss": 1.21,
+      "step": 1487
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4732724902216428e-06,
+      "loss": 1.1871,
+      "step": 1488
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4602346805736637e-06,
+      "loss": 1.1776,
+      "step": 1489
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4471968709256846e-06,
+      "loss": 1.2206,
+      "step": 1490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4341590612777055e-06,
+      "loss": 1.182,
+      "step": 1491
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4211212516297262e-06,
+      "loss": 1.2,
+      "step": 1492
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.408083441981747e-06,
+      "loss": 1.2351,
+      "step": 1493
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.395045632333768e-06,
+      "loss": 1.1742,
+      "step": 1494
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.3820078226857889e-06,
+      "loss": 1.181,
+      "step": 1495
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.3689700130378098e-06,
+      "loss": 1.1456,
+      "step": 1496
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.3559322033898307e-06,
+      "loss": 1.1967,
+      "step": 1497
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.3428943937418515e-06,
+      "loss": 1.1914,
+      "step": 1498
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3298565840938724e-06,
+      "loss": 1.2083,
+      "step": 1499
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3168187744458931e-06,
+      "loss": 1.1923,
+      "step": 1500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.303780964797914e-06,
+      "loss": 1.134,
+      "step": 1501
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.290743155149935e-06,
+      "loss": 1.2035,
+      "step": 1502
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2777053455019558e-06,
+      "loss": 1.2052,
+      "step": 1503
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2646675358539767e-06,
+      "loss": 1.1926,
+      "step": 1504
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2516297262059976e-06,
+      "loss": 1.1852,
+      "step": 1505
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2385919165580183e-06,
+      "loss": 1.1984,
+      "step": 1506
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2255541069100392e-06,
+      "loss": 1.1847,
+      "step": 1507
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.21251629726206e-06,
+      "loss": 1.1841,
+      "step": 1508
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.199478487614081e-06,
+      "loss": 1.2029,
+      "step": 1509
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.186440677966102e-06,
+      "loss": 1.2097,
+      "step": 1510
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1734028683181228e-06,
+      "loss": 1.173,
+      "step": 1511
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1603650586701435e-06,
+      "loss": 1.1948,
+      "step": 1512
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1473272490221644e-06,
+      "loss": 1.2197,
+      "step": 1513
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1342894393741853e-06,
+      "loss": 1.1564,
+      "step": 1514
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.121251629726206e-06,
+      "loss": 1.1813,
+      "step": 1515
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1082138200782269e-06,
+      "loss": 1.1864,
+      "step": 1516
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0951760104302478e-06,
+      "loss": 1.1919,
+      "step": 1517
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0821382007822687e-06,
+      "loss": 1.1797,
+      "step": 1518
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0691003911342896e-06,
+      "loss": 1.1964,
+      "step": 1519
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0560625814863105e-06,
+      "loss": 1.181,
+      "step": 1520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0430247718383314e-06,
+      "loss": 1.1708,
+      "step": 1521
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0299869621903522e-06,
+      "loss": 1.2242,
+      "step": 1522
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.016949152542373e-06,
+      "loss": 1.1693,
+      "step": 1523
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0039113428943938e-06,
+      "loss": 1.1864,
+      "step": 1524
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.908735332464147e-07,
+      "loss": 1.2212,
+      "step": 1525
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.778357235984354e-07,
+      "loss": 1.1969,
+      "step": 1526
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.647979139504563e-07,
+      "loss": 1.2076,
+      "step": 1527
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.517601043024772e-07,
+      "loss": 1.2249,
+      "step": 1528
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.387222946544981e-07,
+      "loss": 1.2036,
+      "step": 1529
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.25684485006519e-07,
+      "loss": 1.1963,
+      "step": 1530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.126466753585399e-07,
+      "loss": 1.1641,
+      "step": 1531
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.996088657105607e-07,
+      "loss": 1.2148,
+      "step": 1532
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.865710560625816e-07,
+      "loss": 1.2277,
+      "step": 1533
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.735332464146024e-07,
+      "loss": 1.1488,
+      "step": 1534
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.604954367666232e-07,
+      "loss": 1.158,
+      "step": 1535
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.474576271186441e-07,
+      "loss": 1.1635,
+      "step": 1536
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.34419817470665e-07,
+      "loss": 1.1802,
+      "step": 1537
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.213820078226859e-07,
+      "loss": 1.2195,
+      "step": 1538
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.083441981747067e-07,
+      "loss": 1.193,
+      "step": 1539
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.953063885267276e-07,
+      "loss": 1.1923,
+      "step": 1540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.822685788787485e-07,
+      "loss": 1.1888,
+      "step": 1541
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.1852,
+      "step": 1542
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.561929595827903e-07,
+      "loss": 1.1503,
+      "step": 1543
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.431551499348109e-07,
+      "loss": 1.1898,
+      "step": 1544
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.301173402868318e-07,
+      "loss": 1.199,
+      "step": 1545
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.170795306388527e-07,
+      "loss": 1.1647,
+      "step": 1546
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.040417209908735e-07,
+      "loss": 1.1854,
+      "step": 1547
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.910039113428944e-07,
+      "loss": 1.2312,
+      "step": 1548
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.779661016949153e-07,
+      "loss": 1.2179,
+      "step": 1549
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.649282920469362e-07,
+      "loss": 1.1716,
+      "step": 1550
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.51890482398957e-07,
+      "loss": 1.1946,
+      "step": 1551
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.388526727509779e-07,
+      "loss": 1.1576,
+      "step": 1552
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.258148631029988e-07,
+      "loss": 1.1829,
+      "step": 1553
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.127770534550196e-07,
+      "loss": 1.161,
+      "step": 1554
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.997392438070405e-07,
+      "loss": 1.207,
+      "step": 1555
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.867014341590614e-07,
+      "loss": 1.17,
+      "step": 1556
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.736636245110822e-07,
+      "loss": 1.2059,
+      "step": 1557
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.60625814863103e-07,
+      "loss": 1.1674,
+      "step": 1558
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.475880052151239e-07,
+      "loss": 1.202,
+      "step": 1559
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.345501955671448e-07,
+      "loss": 1.2046,
+      "step": 1560
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.215123859191657e-07,
+      "loss": 1.1973,
+      "step": 1561
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.084745762711865e-07,
+      "loss": 1.1875,
+      "step": 1562
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.954367666232074e-07,
+      "loss": 1.1938,
+      "step": 1563
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.823989569752282e-07,
+      "loss": 1.1562,
+      "step": 1564
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6936114732724906e-07,
+      "loss": 1.1699,
+      "step": 1565
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.5632333767926995e-07,
+      "loss": 1.1873,
+      "step": 1566
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.432855280312908e-07,
+      "loss": 1.1727,
+      "step": 1567
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.302477183833116e-07,
+      "loss": 1.1716,
+      "step": 1568
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.172099087353325e-07,
+      "loss": 1.1973,
+      "step": 1569
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0417209908735333e-07,
+      "loss": 1.2015,
+      "step": 1570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.9113428943937423e-07,
+      "loss": 1.2175,
+      "step": 1571
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7809647979139513e-07,
+      "loss": 1.2017,
+      "step": 1572
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.650586701434159e-07,
+      "loss": 1.1893,
+      "step": 1573
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.5202086049543677e-07,
+      "loss": 1.1995,
+      "step": 1574
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3898305084745766e-07,
+      "loss": 1.1456,
+      "step": 1575
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.259452411994785e-07,
+      "loss": 1.1541,
+      "step": 1576
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.129074315514994e-07,
+      "loss": 1.1317,
+      "step": 1577
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.9986962190352025e-07,
+      "loss": 1.2208,
+      "step": 1578
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.868318122555411e-07,
+      "loss": 1.1703,
+      "step": 1579
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7379400260756194e-07,
+      "loss": 1.1829,
+      "step": 1580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6075619295958284e-07,
+      "loss": 1.1837,
+      "step": 1581
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.477183833116037e-07,
+      "loss": 1.1523,
+      "step": 1582
+    },
+    {
+      "epoch": 2.0,
+      "step": 1582,
+      "total_flos": 1.712123314962432e+16,
+      "train_loss": 1.243581701922507,
+      "train_runtime": 26517.1703,
+      "train_samples_per_second": 7.627,
+      "train_steps_per_second": 0.06
+    }
+  ],
+  "max_steps": 1582,
+  "num_train_epochs": 2,
+  "total_flos": 1.712123314962432e+16,
+  "trial_name": null,
+  "trial_params": null
+}