diff --git "a/checkpoint-12185/trainer_state.json" "b/checkpoint-12185/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-12185/trainer_state.json"
@@ -0,0 +1,73290 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999928195554279,
+  "eval_steps": 609,
+  "global_step": 12185,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.101722723543889e-09,
+      "loss": 2.2972,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.203445447087778e-09,
+      "loss": 2.2753,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2305168170631666e-08,
+      "loss": 2.0002,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6406890894175556e-08,
+      "loss": 2.2223,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0508613617719444e-08,
+      "loss": 2.1817,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4610336341263332e-08,
+      "loss": 2.5169,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8712059064807224e-08,
+      "loss": 2.2092,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.281378178835111e-08,
+      "loss": 2.1944,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6915504511894997e-08,
+      "loss": 2.4209,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.101722723543889e-08,
+      "loss": 2.4294,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.511894995898278e-08,
+      "loss": 2.3176,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9220672682526664e-08,
+      "loss": 2.1067,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.332239540607055e-08,
+      "loss": 2.5917,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.742411812961445e-08,
+      "loss": 2.3246,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.152584085315833e-08,
+      "loss": 2.3478,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.562756357670222e-08,
+      "loss": 2.4847,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.972928630024611e-08,
+      "loss": 2.1185,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.383100902378999e-08,
+      "loss": 2.1788,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.793273174733389e-08,
+      "loss": 2.4704,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.203445447087778e-08,
+      "loss": 2.2347,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.613617719442167e-08,
+      "loss": 2.1863,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.023789991796556e-08,
+      "loss": 2.0941,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.433962264150944e-08,
+      "loss": 2.2202,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.844134536505333e-08,
+      "loss": 2.282,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0254306808859723e-07,
+      "loss": 2.1386,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.066447908121411e-07,
+      "loss": 2.3794,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.10746513535685e-07,
+      "loss": 2.0625,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.148482362592289e-07,
+      "loss": 2.8435,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1894995898277277e-07,
+      "loss": 2.2446,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2305168170631665e-07,
+      "loss": 2.1195,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2715340442986055e-07,
+      "loss": 2.3243,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3125512715340445e-07,
+      "loss": 2.1666,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3535684987694832e-07,
+      "loss": 2.1677,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3945857260049222e-07,
+      "loss": 1.9796,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4356029532403611e-07,
+      "loss": 2.0966,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4766201804757999e-07,
+      "loss": 2.0716,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5176374077112388e-07,
+      "loss": 1.9907,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5586546349466778e-07,
+      "loss": 2.1361,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5996718621821165e-07,
+      "loss": 2.2616,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6406890894175555e-07,
+      "loss": 2.0836,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6817063166529945e-07,
+      "loss": 2.3845,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7227235438884335e-07,
+      "loss": 2.0324,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7637407711238722e-07,
+      "loss": 2.0313,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8047579983593112e-07,
+      "loss": 2.4506,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.84577522559475e-07,
+      "loss": 1.9794,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.886792452830189e-07,
+      "loss": 2.0609,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9278096800656276e-07,
+      "loss": 2.0054,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9688269073010666e-07,
+      "loss": 2.4932,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0098441345365056e-07,
+      "loss": 2.1251,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0508613617719445e-07,
+      "loss": 2.2167,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.091878589007383e-07,
+      "loss": 2.0295,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.132895816242822e-07,
+      "loss": 2.1033,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.173913043478261e-07,
+      "loss": 2.225,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2149302707137e-07,
+      "loss": 1.9331,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.255947497949139e-07,
+      "loss": 2.0596,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.296964725184578e-07,
+      "loss": 1.9417,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3379819524200169e-07,
+      "loss": 1.9686,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3789991796554553e-07,
+      "loss": 1.8251,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.420016406890894e-07,
+      "loss": 2.4721,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.461033634126333e-07,
+      "loss": 2.104,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.502050861361772e-07,
+      "loss": 2.2616,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.543068088597211e-07,
+      "loss": 1.9587,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.58408531583265e-07,
+      "loss": 2.0699,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.625102543068089e-07,
+      "loss": 2.2282,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6661197703035274e-07,
+      "loss": 1.9006,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7071369975389664e-07,
+      "loss": 1.9224,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7481542247744053e-07,
+      "loss": 1.9694,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7891714520098443e-07,
+      "loss": 2.194,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8301886792452833e-07,
+      "loss": 1.7362,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8712059064807223e-07,
+      "loss": 1.8057,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9122231337161613e-07,
+      "loss": 2.1604,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9532403609515997e-07,
+      "loss": 2.1537,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9942575881870387e-07,
+      "loss": 2.0859,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0352748154224777e-07,
+      "loss": 1.8449,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0762920426579167e-07,
+      "loss": 1.9374,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1173092698933556e-07,
+      "loss": 2.24,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.158326497128794e-07,
+      "loss": 1.8642,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.199343724364233e-07,
+      "loss": 1.9807,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.240360951599672e-07,
+      "loss": 1.7175,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.281378178835111e-07,
+      "loss": 1.7293,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.32239540607055e-07,
+      "loss": 1.4999,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.363412633305989e-07,
+      "loss": 1.7512,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.404429860541428e-07,
+      "loss": 2.427,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.445447087776867e-07,
+      "loss": 1.8384,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4864643150123054e-07,
+      "loss": 1.5592,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5274815422477444e-07,
+      "loss": 1.8201,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5684987694831834e-07,
+      "loss": 1.602,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6095159967186224e-07,
+      "loss": 1.7784,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.650533223954061e-07,
+      "loss": 1.6625,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6915504511895e-07,
+      "loss": 1.8081,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.732567678424939e-07,
+      "loss": 1.5321,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.773584905660378e-07,
+      "loss": 1.4613,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.814602132895816e-07,
+      "loss": 1.8105,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.855619360131255e-07,
+      "loss": 1.6812,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.896636587366694e-07,
+      "loss": 1.3996,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.937653814602133e-07,
+      "loss": 1.5579,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.978671041837572e-07,
+      "loss": 1.836,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.019688269073011e-07,
+      "loss": 1.311,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.06070549630845e-07,
+      "loss": 1.2478,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.101722723543889e-07,
+      "loss": 1.6295,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.142739950779328e-07,
+      "loss": 1.395,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.183757178014766e-07,
+      "loss": 1.5346,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.224774405250205e-07,
+      "loss": 1.5468,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.265791632485644e-07,
+      "loss": 1.3017,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.306808859721083e-07,
+      "loss": 1.3618,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347826086956522e-07,
+      "loss": 1.3106,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.388843314191961e-07,
+      "loss": 1.3655,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4298605414274e-07,
+      "loss": 1.3215,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.470877768662839e-07,
+      "loss": 1.6217,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.511894995898278e-07,
+      "loss": 1.3996,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.552912223133717e-07,
+      "loss": 1.2089,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.593929450369156e-07,
+      "loss": 1.2472,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.634946677604595e-07,
+      "loss": 1.1707,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6759639048400337e-07,
+      "loss": 1.6083,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.716981132075472e-07,
+      "loss": 1.1726,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7579983593109106e-07,
+      "loss": 1.234,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.799015586546349e-07,
+      "loss": 1.3359,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.840032813781788e-07,
+      "loss": 1.3315,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.881050041017227e-07,
+      "loss": 1.1861,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.922067268252666e-07,
+      "loss": 1.2429,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963084495488105e-07,
+      "loss": 1.4431,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.004101722723544e-07,
+      "loss": 1.1228,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.045118949958983e-07,
+      "loss": 1.0371,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.086136177194422e-07,
+      "loss": 1.0382,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.127153404429861e-07,
+      "loss": 1.791,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1681706316653e-07,
+      "loss": 1.4074,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.209187858900739e-07,
+      "loss": 1.17,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.250205086136178e-07,
+      "loss": 1.0649,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.291222313371616e-07,
+      "loss": 1.0529,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.332239540607055e-07,
+      "loss": 0.9915,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.373256767842494e-07,
+      "loss": 0.9733,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.414273995077933e-07,
+      "loss": 1.0579,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.455291222313372e-07,
+      "loss": 1.1137,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.496308449548811e-07,
+      "loss": 1.0268,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.53732567678425e-07,
+      "loss": 1.0894,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.578342904019689e-07,
+      "loss": 1.1653,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.619360131255128e-07,
+      "loss": 0.9713,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.660377358490567e-07,
+      "loss": 1.1663,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.701394585726006e-07,
+      "loss": 1.0411,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.742411812961445e-07,
+      "loss": 1.0512,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.783429040196884e-07,
+      "loss": 1.2857,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.824446267432323e-07,
+      "loss": 1.0289,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.86546349466776e-07,
+      "loss": 1.0875,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.906480721903199e-07,
+      "loss": 0.9654,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.947497949138638e-07,
+      "loss": 1.0536,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.988515176374077e-07,
+      "loss": 0.9565,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.029532403609516e-07,
+      "loss": 1.1424,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.070549630844955e-07,
+      "loss": 1.1319,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.111566858080394e-07,
+      "loss": 1.0987,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.152584085315833e-07,
+      "loss": 0.9829,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.193601312551272e-07,
+      "loss": 1.063,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.234618539786711e-07,
+      "loss": 1.0743,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.275635767022149e-07,
+      "loss": 1.081,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.316652994257588e-07,
+      "loss": 0.9119,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.357670221493027e-07,
+      "loss": 1.1091,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.398687448728466e-07,
+      "loss": 0.9289,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.439704675963905e-07,
+      "loss": 1.1682,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.480721903199344e-07,
+      "loss": 1.047,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.521739130434783e-07,
+      "loss": 0.9096,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.562756357670222e-07,
+      "loss": 1.1586,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.603773584905661e-07,
+      "loss": 1.0682,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6447908121411e-07,
+      "loss": 1.19,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.685808039376539e-07,
+      "loss": 0.9906,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.726825266611978e-07,
+      "loss": 1.1366,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.767842493847417e-07,
+      "loss": 1.0745,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.808859721082856e-07,
+      "loss": 0.9525,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.849876948318295e-07,
+      "loss": 0.9142,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.890894175553734e-07,
+      "loss": 1.3039,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.931911402789172e-07,
+      "loss": 1.0889,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.972928630024611e-07,
+      "loss": 1.0314,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.01394585726005e-07,
+      "loss": 0.9759,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.054963084495489e-07,
+      "loss": 1.1925,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.095980311730928e-07,
+      "loss": 1.4058,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.136997538966367e-07,
+      "loss": 1.2465,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.178014766201806e-07,
+      "loss": 0.9483,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.219031993437245e-07,
+      "loss": 0.8943,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.260049220672683e-07,
+      "loss": 0.8675,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.301066447908122e-07,
+      "loss": 0.8807,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.342083675143561e-07,
+      "loss": 1.1085,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.383100902379e-07,
+      "loss": 0.8065,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.424118129614439e-07,
+      "loss": 1.1893,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.465135356849878e-07,
+      "loss": 1.0527,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.506152584085317e-07,
+      "loss": 1.0841,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.547169811320755e-07,
+      "loss": 0.9838,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.588187038556194e-07,
+      "loss": 1.1074,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.629204265791632e-07,
+      "loss": 1.1098,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.670221493027071e-07,
+      "loss": 0.8911,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.71123872026251e-07,
+      "loss": 1.0711,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.752255947497949e-07,
+      "loss": 1.0258,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.793273174733388e-07,
+      "loss": 1.2864,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.834290401968827e-07,
+      "loss": 0.9504,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.875307629204266e-07,
+      "loss": 0.9855,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.916324856439705e-07,
+      "loss": 1.2466,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.957342083675144e-07,
+      "loss": 0.8856,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.998359310910583e-07,
+      "loss": 0.9411,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.039376538146022e-07,
+      "loss": 0.9069,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.080393765381461e-07,
+      "loss": 1.0178,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.1214109926169e-07,
+      "loss": 1.0567,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.162428219852339e-07,
+      "loss": 0.8418,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.203445447087778e-07,
+      "loss": 0.8749,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.244462674323217e-07,
+      "loss": 0.8817,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.285479901558656e-07,
+      "loss": 0.8775,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.326497128794095e-07,
+      "loss": 1.0441,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.367514356029532e-07,
+      "loss": 1.047,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.408531583264971e-07,
+      "loss": 0.8892,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.44954881050041e-07,
+      "loss": 0.8804,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.490566037735849e-07,
+      "loss": 0.9061,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.531583264971288e-07,
+      "loss": 0.9381,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.572600492206727e-07,
+      "loss": 0.8673,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.613617719442166e-07,
+      "loss": 0.7823,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.654634946677605e-07,
+      "loss": 1.1427,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 0.9059,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.736669401148483e-07,
+      "loss": 1.1584,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.777686628383922e-07,
+      "loss": 0.8436,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.818703855619361e-07,
+      "loss": 0.7988,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.8597210828548e-07,
+      "loss": 0.8595,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.900738310090239e-07,
+      "loss": 0.8195,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.941755537325678e-07,
+      "loss": 0.9543,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.982772764561117e-07,
+      "loss": 1.1762,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.023789991796556e-07,
+      "loss": 0.9503,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.064807219031995e-07,
+      "loss": 0.9083,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.105824446267434e-07,
+      "loss": 1.1094,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.146841673502873e-07,
+      "loss": 0.8462,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.187858900738312e-07,
+      "loss": 1.0523,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.22887612797375e-07,
+      "loss": 0.8812,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.26989335520919e-07,
+      "loss": 0.8695,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.310910582444628e-07,
+      "loss": 1.0079,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.351927809680067e-07,
+      "loss": 0.8059,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.392945036915505e-07,
+      "loss": 0.9682,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.433962264150944e-07,
+      "loss": 1.2368,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.474979491386382e-07,
+      "loss": 0.8925,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.515996718621821e-07,
+      "loss": 0.9541,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.55701394585726e-07,
+      "loss": 0.882,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.598031173092698e-07,
+      "loss": 0.9685,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.639048400328138e-07,
+      "loss": 0.9528,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.680065627563576e-07,
+      "loss": 0.8603,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.721082854799016e-07,
+      "loss": 0.8719,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.762100082034454e-07,
+      "loss": 0.9105,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.803117309269894e-07,
+      "loss": 0.7887,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.844134536505332e-07,
+      "loss": 0.8925,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.885151763740772e-07,
+      "loss": 0.8129,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.92616899097621e-07,
+      "loss": 0.8554,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.96718621821165e-07,
+      "loss": 0.9318,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0008203445447088e-06,
+      "loss": 1.1055,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0049220672682528e-06,
+      "loss": 0.9736,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0090237899917966e-06,
+      "loss": 1.1502,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0131255127153406e-06,
+      "loss": 1.0187,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0172272354388844e-06,
+      "loss": 0.8673,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0213289581624284e-06,
+      "loss": 0.7492,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0254306808859722e-06,
+      "loss": 0.9333,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0295324036095162e-06,
+      "loss": 1.0638,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.03363412633306e-06,
+      "loss": 0.86,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.037735849056604e-06,
+      "loss": 0.8525,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0418375717801478e-06,
+      "loss": 1.0821,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0459392945036918e-06,
+      "loss": 1.2564,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0500410172272356e-06,
+      "loss": 0.8373,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0541427399507796e-06,
+      "loss": 0.7815,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0582444626743232e-06,
+      "loss": 0.9642,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0623461853978672e-06,
+      "loss": 0.8964,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.066447908121411e-06,
+      "loss": 0.8875,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.070549630844955e-06,
+      "loss": 0.7975,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0746513535684988e-06,
+      "loss": 0.8773,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0787530762920428e-06,
+      "loss": 0.8071,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0828547990155865e-06,
+      "loss": 0.8387,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0869565217391306e-06,
+      "loss": 0.8525,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0910582444626743e-06,
+      "loss": 0.8665,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0951599671862183e-06,
+      "loss": 0.9339,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0992616899097621e-06,
+      "loss": 0.9459,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1033634126333061e-06,
+      "loss": 0.855,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.10746513535685e-06,
+      "loss": 0.9238,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.111566858080394e-06,
+      "loss": 0.9075,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1156685808039377e-06,
+      "loss": 0.9518,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1197703035274817e-06,
+      "loss": 0.8971,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1238720262510255e-06,
+      "loss": 0.7884,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1279737489745695e-06,
+      "loss": 0.7877,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1320754716981133e-06,
+      "loss": 0.9186,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1361771944216571e-06,
+      "loss": 0.8665,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1402789171452011e-06,
+      "loss": 0.8078,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.144380639868745e-06,
+      "loss": 1.0643,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.148482362592289e-06,
+      "loss": 0.8401,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1525840853158327e-06,
+      "loss": 0.8229,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1566858080393767e-06,
+      "loss": 1.0039,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1607875307629205e-06,
+      "loss": 0.9245,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1648892534864645e-06,
+      "loss": 0.9582,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1689909762100083e-06,
+      "loss": 0.8707,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.173092698933552e-06,
+      "loss": 0.886,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.177194421657096e-06,
+      "loss": 0.836,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1812961443806399e-06,
+      "loss": 0.9504,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1853978671041839e-06,
+      "loss": 0.8514,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1894995898277277e-06,
+      "loss": 1.0041,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1936013125512717e-06,
+      "loss": 0.8374,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1977030352748155e-06,
+      "loss": 0.8669,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2018047579983595e-06,
+      "loss": 1.0131,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2059064807219033e-06,
+      "loss": 0.9804,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.210008203445447e-06,
+      "loss": 0.8844,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.214109926168991e-06,
+      "loss": 0.8492,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2182116488925349e-06,
+      "loss": 0.9015,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2223133716160789e-06,
+      "loss": 0.8706,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2264150943396227e-06,
+      "loss": 0.7735,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2305168170631667e-06,
+      "loss": 0.8567,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2346185397867105e-06,
+      "loss": 0.9108,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2387202625102545e-06,
+      "loss": 0.9915,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2428219852337983e-06,
+      "loss": 0.8455,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2469237079573423e-06,
+      "loss": 0.8078,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.251025430680886e-06,
+      "loss": 0.839,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2551271534044298e-06,
+      "loss": 0.776,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2592288761279738e-06,
+      "loss": 0.9851,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2633305988515176e-06,
+      "loss": 0.8481,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2674323215750616e-06,
+      "loss": 0.8506,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2715340442986054e-06,
+      "loss": 0.7888,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2756357670221494e-06,
+      "loss": 0.9718,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2797374897456932e-06,
+      "loss": 1.0,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2838392124692372e-06,
+      "loss": 0.8818,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.287940935192781e-06,
+      "loss": 0.8359,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.292042657916325e-06,
+      "loss": 0.7879,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2961443806398688e-06,
+      "loss": 0.9378,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3002461033634128e-06,
+      "loss": 1.0903,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3043478260869566e-06,
+      "loss": 0.8423,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3084495488105006e-06,
+      "loss": 0.8392,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3125512715340444e-06,
+      "loss": 0.901,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3166529942575884e-06,
+      "loss": 0.8134,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3207547169811322e-06,
+      "loss": 0.9616,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3248564397046762e-06,
+      "loss": 1.133,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.32895816242822e-06,
+      "loss": 0.7703,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.333059885151764e-06,
+      "loss": 1.1563,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3371616078753078e-06,
+      "loss": 0.8425,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3412633305988516e-06,
+      "loss": 0.9972,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3453650533223956e-06,
+      "loss": 0.9503,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3494667760459394e-06,
+      "loss": 0.9367,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3535684987694834e-06,
+      "loss": 0.8963,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.357670221493027e-06,
+      "loss": 0.8185,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3617719442165712e-06,
+      "loss": 0.7717,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3658736669401148e-06,
+      "loss": 0.8107,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.369975389663659e-06,
+      "loss": 0.7839,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3740771123872026e-06,
+      "loss": 0.7808,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3781788351107468e-06,
+      "loss": 0.8647,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3822805578342904e-06,
+      "loss": 1.179,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3863822805578344e-06,
+      "loss": 0.8567,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3904840032813782e-06,
+      "loss": 0.9283,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3945857260049222e-06,
+      "loss": 0.837,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.398687448728466e-06,
+      "loss": 1.1187,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.40278917145201e-06,
+      "loss": 0.7709,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4068908941755538e-06,
+      "loss": 0.9058,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4109926168990978e-06,
+      "loss": 0.9658,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4150943396226415e-06,
+      "loss": 0.9108,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4191960623461856e-06,
+      "loss": 0.8593,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4232977850697293e-06,
+      "loss": 0.8122,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4273995077932733e-06,
+      "loss": 0.9247,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4315012305168171e-06,
+      "loss": 0.9478,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4356029532403611e-06,
+      "loss": 0.8526,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.439704675963905e-06,
+      "loss": 0.7907,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.443806398687449e-06,
+      "loss": 0.8749,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4479081214109927e-06,
+      "loss": 0.943,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4520098441345365e-06,
+      "loss": 0.8999,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4561115668580805e-06,
+      "loss": 0.9161,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4602132895816243e-06,
+      "loss": 1.053,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4643150123051683e-06,
+      "loss": 0.9217,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4684167350287121e-06,
+      "loss": 0.8271,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4725184577522561e-06,
+      "loss": 0.8124,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4766201804758e-06,
+      "loss": 0.8674,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.480721903199344e-06,
+      "loss": 0.821,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4848236259228877e-06,
+      "loss": 0.8858,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4889253486464317e-06,
+      "loss": 0.9096,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4930270713699755e-06,
+      "loss": 0.9773,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4971287940935195e-06,
+      "loss": 0.931,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5012305168170633e-06,
+      "loss": 0.9918,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5053322395406073e-06,
+      "loss": 0.8167,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 0.8059,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.513535684987695e-06,
+      "loss": 0.8754,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.517637407711239e-06,
+      "loss": 0.9429,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.521739130434783e-06,
+      "loss": 0.9681,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5258408531583265e-06,
+      "loss": 0.8618,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5299425758818707e-06,
+      "loss": 0.9634,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5340442986054143e-06,
+      "loss": 0.9949,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5381460213289585e-06,
+      "loss": 0.9644,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.542247744052502e-06,
+      "loss": 0.8903,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5463494667760463e-06,
+      "loss": 0.7402,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5504511894995899e-06,
+      "loss": 0.9298,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.554552912223134e-06,
+      "loss": 0.7665,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5586546349466777e-06,
+      "loss": 0.879,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5627563576702215e-06,
+      "loss": 0.8164,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5668580803937655e-06,
+      "loss": 0.8475,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5709598031173093e-06,
+      "loss": 0.6958,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5750615258408533e-06,
+      "loss": 0.7463,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.579163248564397e-06,
+      "loss": 0.9915,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.583264971287941e-06,
+      "loss": 0.8215,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5873666940114848e-06,
+      "loss": 0.8101,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5914684167350288e-06,
+      "loss": 1.0532,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5955701394585726e-06,
+      "loss": 0.8612,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5996718621821166e-06,
+      "loss": 0.8966,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6037735849056604e-06,
+      "loss": 0.8084,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6078753076292044e-06,
+      "loss": 0.9976,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6119770303527482e-06,
+      "loss": 1.1481,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6160787530762922e-06,
+      "loss": 0.9398,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.620180475799836e-06,
+      "loss": 0.8777,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.62428219852338e-06,
+      "loss": 0.8297,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6283839212469238e-06,
+      "loss": 0.8016,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6324856439704678e-06,
+      "loss": 0.8282,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6365873666940116e-06,
+      "loss": 0.8112,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6406890894175556e-06,
+      "loss": 0.8095,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6447908121410994e-06,
+      "loss": 0.931,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6488925348646434e-06,
+      "loss": 0.8087,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6529942575881872e-06,
+      "loss": 0.881,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6570959803117312e-06,
+      "loss": 0.7451,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.661197703035275e-06,
+      "loss": 0.8778,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.665299425758819e-06,
+      "loss": 0.8297,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6694011484823628e-06,
+      "loss": 1.1081,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6735028712059064e-06,
+      "loss": 0.7848,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6776045939294506e-06,
+      "loss": 0.8313,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6817063166529942e-06,
+      "loss": 0.7161,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6858080393765384e-06,
+      "loss": 0.8232,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.689909762100082e-06,
+      "loss": 0.9164,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6940114848236262e-06,
+      "loss": 0.7784,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6981132075471698e-06,
+      "loss": 0.7534,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7022149302707138e-06,
+      "loss": 0.9157,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7063166529942576e-06,
+      "loss": 0.8987,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7104183757178016e-06,
+      "loss": 0.8513,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7145200984413454e-06,
+      "loss": 1.0181,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7186218211648894e-06,
+      "loss": 0.9694,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7227235438884332e-06,
+      "loss": 0.8009,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7268252666119772e-06,
+      "loss": 0.8284,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.730926989335521e-06,
+      "loss": 0.8536,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.735028712059065e-06,
+      "loss": 0.8441,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 0.7493,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7432321575061528e-06,
+      "loss": 0.8618,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7473338802296966e-06,
+      "loss": 0.7625,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7514356029532406e-06,
+      "loss": 0.8095,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7555373256767843e-06,
+      "loss": 0.8357,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7596390484003284e-06,
+      "loss": 0.7923,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7637407711238721e-06,
+      "loss": 0.9923,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7678424938474161e-06,
+      "loss": 0.9502,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.77194421657096e-06,
+      "loss": 0.9296,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.776045939294504e-06,
+      "loss": 0.9339,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7801476620180477e-06,
+      "loss": 0.8722,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7842493847415915e-06,
+      "loss": 0.8844,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7883511074651355e-06,
+      "loss": 0.9355,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7924528301886793e-06,
+      "loss": 0.7417,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7965545529122233e-06,
+      "loss": 0.7402,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8006562756357671e-06,
+      "loss": 1.0325,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8047579983593111e-06,
+      "loss": 0.773,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.808859721082855e-06,
+      "loss": 0.9078,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.812961443806399e-06,
+      "loss": 0.8329,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8170631665299427e-06,
+      "loss": 0.714,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8211648892534867e-06,
+      "loss": 0.7802,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8252666119770305e-06,
+      "loss": 0.7841,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8293683347005745e-06,
+      "loss": 0.8756,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8334700574241183e-06,
+      "loss": 0.8355,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8375717801476623e-06,
+      "loss": 0.786,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.841673502871206e-06,
+      "loss": 0.9489,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.84577522559475e-06,
+      "loss": 0.9413,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8498769483182937e-06,
+      "loss": 0.9557,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.853978671041838e-06,
+      "loss": 0.9505,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8580803937653815e-06,
+      "loss": 1.0049,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8621821164889257e-06,
+      "loss": 0.87,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8662838392124693e-06,
+      "loss": 0.7507,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8703855619360135e-06,
+      "loss": 0.9126,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.874487284659557e-06,
+      "loss": 0.8376,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.878589007383101e-06,
+      "loss": 0.9591,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8826907301066449e-06,
+      "loss": 0.7601,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8867924528301889e-06,
+      "loss": 0.9274,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8908941755537327e-06,
+      "loss": 0.7646,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8949958982772765e-06,
+      "loss": 0.8836,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8990976210008205e-06,
+      "loss": 0.8729,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9031993437243643e-06,
+      "loss": 0.8407,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9073010664479083e-06,
+      "loss": 0.919,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.911402789171452e-06,
+      "loss": 0.7759,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9155045118949963e-06,
+      "loss": 0.8695,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9196062346185396e-06,
+      "loss": 0.7405,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.923707957342084e-06,
+      "loss": 0.7672,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9278096800656276e-06,
+      "loss": 0.872,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.931911402789172e-06,
+      "loss": 0.7824,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9360131255127152e-06,
+      "loss": 0.7743,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9401148482362594e-06,
+      "loss": 0.8598,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9442165709598032e-06,
+      "loss": 0.812,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9483182936833475e-06,
+      "loss": 0.7607,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.952420016406891e-06,
+      "loss": 0.7493,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.956521739130435e-06,
+      "loss": 0.9313,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.960623461853979e-06,
+      "loss": 0.7717,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9647251845775226e-06,
+      "loss": 0.7905,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9688269073010664e-06,
+      "loss": 0.779,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9729286300246106e-06,
+      "loss": 0.8146,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9770303527481544e-06,
+      "loss": 0.843,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.981132075471698e-06,
+      "loss": 0.6805,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.985233798195242e-06,
+      "loss": 0.6956,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9893355209187862e-06,
+      "loss": 0.7838,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99343724364233e-06,
+      "loss": 0.762,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.997538966365874e-06,
+      "loss": 0.7583,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0016406890894176e-06,
+      "loss": 0.8007,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0057424118129614e-06,
+      "loss": 0.7524,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0098441345365056e-06,
+      "loss": 0.821,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0139458572600494e-06,
+      "loss": 0.7973,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.018047579983593e-06,
+      "loss": 0.7306,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.022149302707137e-06,
+      "loss": 0.9543,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.026251025430681e-06,
+      "loss": 0.8914,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.030352748154225e-06,
+      "loss": 0.7931,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0344544708777688e-06,
+      "loss": 0.7785,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0385561936013126e-06,
+      "loss": 0.8293,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0426579163248568e-06,
+      "loss": 0.9693,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0467596390484006e-06,
+      "loss": 0.7602,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0508613617719444e-06,
+      "loss": 0.8416,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.054963084495488e-06,
+      "loss": 0.8233,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0590648072190324e-06,
+      "loss": 0.8728,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.063166529942576e-06,
+      "loss": 0.7971,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.06726825266612e-06,
+      "loss": 0.8091,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0713699753896638e-06,
+      "loss": 0.8141,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.075471698113208e-06,
+      "loss": 0.8194,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0795734208367518e-06,
+      "loss": 0.7724,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0836751435602956e-06,
+      "loss": 0.6685,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0877768662838393e-06,
+      "loss": 0.943,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0918785890073836e-06,
+      "loss": 0.7861,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.095980311730927e-06,
+      "loss": 0.9948,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.100082034454471e-06,
+      "loss": 0.7682,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.104183757178015e-06,
+      "loss": 0.7713,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.108285479901559e-06,
+      "loss": 0.8592,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1123872026251025e-06,
+      "loss": 0.9458,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1164889253486463e-06,
+      "loss": 0.865,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1205906480721905e-06,
+      "loss": 0.7463,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1246923707957343e-06,
+      "loss": 0.678,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.128794093519278e-06,
+      "loss": 0.8087,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.132895816242822e-06,
+      "loss": 0.7516,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.136997538966366e-06,
+      "loss": 0.8609,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.14109926168991e-06,
+      "loss": 0.7626,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1452009844134537e-06,
+      "loss": 0.7754,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1493027071369975e-06,
+      "loss": 0.9565,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1534044298605417e-06,
+      "loss": 0.8455,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1575061525840855e-06,
+      "loss": 0.8204,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1616078753076293e-06,
+      "loss": 0.7583,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.165709598031173e-06,
+      "loss": 0.8604,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1698113207547173e-06,
+      "loss": 0.7703,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 0.8392,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.178014766201805e-06,
+      "loss": 0.8565,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1821164889253487e-06,
+      "loss": 0.8445,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.186218211648893e-06,
+      "loss": 0.9167,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1903199343724367e-06,
+      "loss": 0.7774,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1944216570959805e-06,
+      "loss": 0.7646,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1985233798195243e-06,
+      "loss": 1.0403,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2026251025430685e-06,
+      "loss": 0.8554,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2067268252666123e-06,
+      "loss": 0.8797,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.210828547990156e-06,
+      "loss": 0.856,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2149302707137e-06,
+      "loss": 0.8432,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.219031993437244e-06,
+      "loss": 0.6793,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.223133716160788e-06,
+      "loss": 0.8558,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2272354388843317e-06,
+      "loss": 0.8125,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2313371616078755e-06,
+      "loss": 0.7587,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2354388843314193e-06,
+      "loss": 0.7079,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2395406070549635e-06,
+      "loss": 0.9113,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.243642329778507e-06,
+      "loss": 0.7003,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.247744052502051e-06,
+      "loss": 0.8225,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.251845775225595e-06,
+      "loss": 0.7381,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.255947497949139e-06,
+      "loss": 0.7775,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2600492206726824e-06,
+      "loss": 0.8505,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2641509433962266e-06,
+      "loss": 0.8463,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2682526661197704e-06,
+      "loss": 0.8938,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2723543888433142e-06,
+      "loss": 0.8085,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.276456111566858e-06,
+      "loss": 0.8931,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2805578342904022e-06,
+      "loss": 0.7215,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.284659557013946e-06,
+      "loss": 0.7083,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.28876127973749e-06,
+      "loss": 0.773,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2928630024610336e-06,
+      "loss": 0.8948,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.296964725184578e-06,
+      "loss": 0.9089,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3010664479081216e-06,
+      "loss": 0.7681,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3051681706316654e-06,
+      "loss": 0.8152,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.309269893355209e-06,
+      "loss": 0.8559,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3133716160787534e-06,
+      "loss": 0.6848,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3174733388022972e-06,
+      "loss": 0.8108,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.321575061525841e-06,
+      "loss": 0.8105,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.325676784249385e-06,
+      "loss": 0.7106,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.329778506972929e-06,
+      "loss": 0.9129,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.333880229696473e-06,
+      "loss": 0.8431,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3379819524200166e-06,
+      "loss": 0.9223,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3420836751435604e-06,
+      "loss": 0.7397,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.346185397867104e-06,
+      "loss": 0.7651,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3502871205906484e-06,
+      "loss": 0.762,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.354388843314192e-06,
+      "loss": 0.9835,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.358490566037736e-06,
+      "loss": 0.9057,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3625922887612798e-06,
+      "loss": 0.7739,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.366694011484824e-06,
+      "loss": 0.9075,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3707957342083678e-06,
+      "loss": 0.7493,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3748974569319116e-06,
+      "loss": 0.7296,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3789991796554554e-06,
+      "loss": 0.733,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3831009023789996e-06,
+      "loss": 0.8903,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3872026251025434e-06,
+      "loss": 0.7888,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.391304347826087e-06,
+      "loss": 0.8824,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.395406070549631e-06,
+      "loss": 0.8723,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.399507793273175e-06,
+      "loss": 0.7305,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.403609515996719e-06,
+      "loss": 1.1614,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4077112387202628e-06,
+      "loss": 0.7827,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4118129614438066e-06,
+      "loss": 0.8801,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4159146841673508e-06,
+      "loss": 0.802,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.420016406890894e-06,
+      "loss": 0.8576,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4241181296144384e-06,
+      "loss": 0.7518,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.428219852337982e-06,
+      "loss": 0.8609,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4323215750615264e-06,
+      "loss": 0.724,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4364232977850697e-06,
+      "loss": 0.8324,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.440525020508614e-06,
+      "loss": 0.7612,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4446267432321577e-06,
+      "loss": 0.734,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4487284659557015e-06,
+      "loss": 1.1064,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4528301886792453e-06,
+      "loss": 0.6643,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.456931911402789e-06,
+      "loss": 0.8958,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4610336341263333e-06,
+      "loss": 0.9561,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.465135356849877e-06,
+      "loss": 0.7568,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.469237079573421e-06,
+      "loss": 0.8164,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4733388022969647e-06,
+      "loss": 0.7378,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.477440525020509e-06,
+      "loss": 0.8766,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4815422477440527e-06,
+      "loss": 0.7171,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4856439704675965e-06,
+      "loss": 0.883,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4897456931911403e-06,
+      "loss": 0.8972,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4938474159146845e-06,
+      "loss": 0.848,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4979491386382283e-06,
+      "loss": 0.7832,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 0.884433388710022,
+      "eval_runtime": 440.2001,
+      "eval_samples_per_second": 6.813,
+      "eval_steps_per_second": 0.852,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.502050861361772e-06,
+      "loss": 0.806,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5061525840853163e-06,
+      "loss": 0.9231,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5102543068088597e-06,
+      "loss": 0.7833,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.514356029532404e-06,
+      "loss": 0.9024,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5184577522559477e-06,
+      "loss": 0.8011,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.522559474979492e-06,
+      "loss": 0.8202,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5266611977030353e-06,
+      "loss": 0.8176,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5307629204265795e-06,
+      "loss": 0.8091,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5348646431501233e-06,
+      "loss": 0.7938,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5389663658736675e-06,
+      "loss": 0.7742,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.543068088597211e-06,
+      "loss": 0.8763,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.547169811320755e-06,
+      "loss": 0.8214,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.551271534044299e-06,
+      "loss": 0.9874,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.555373256767843e-06,
+      "loss": 0.7762,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5594749794913865e-06,
+      "loss": 0.823,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5635767022149307e-06,
+      "loss": 0.8849,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5676784249384745e-06,
+      "loss": 0.7898,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.571780147662018e-06,
+      "loss": 0.7526,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.575881870385562e-06,
+      "loss": 0.9254,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5799835931091063e-06,
+      "loss": 0.8613,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.58408531583265e-06,
+      "loss": 0.8033,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5881870385561934e-06,
+      "loss": 0.7831,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5922887612797376e-06,
+      "loss": 0.7537,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5963904840032814e-06,
+      "loss": 0.7981,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6004922067268257e-06,
+      "loss": 0.7872,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.604593929450369e-06,
+      "loss": 0.8137,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 0.8693,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.612797374897457e-06,
+      "loss": 0.8122,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6168990976210012e-06,
+      "loss": 0.7644,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6210008203445446e-06,
+      "loss": 0.815,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.625102543068089e-06,
+      "loss": 0.7273,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6292042657916326e-06,
+      "loss": 0.8854,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.633305988515177e-06,
+      "loss": 0.8789,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.63740771123872e-06,
+      "loss": 0.8527,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6415094339622644e-06,
+      "loss": 0.9532,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6456111566858082e-06,
+      "loss": 0.715,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6497128794093524e-06,
+      "loss": 0.7815,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.653814602132896e-06,
+      "loss": 0.9145,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.65791632485644e-06,
+      "loss": 0.7296,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.662018047579984e-06,
+      "loss": 0.7851,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.666119770303528e-06,
+      "loss": 0.8053,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6702214930270714e-06,
+      "loss": 0.7901,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6743232157506156e-06,
+      "loss": 0.798,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6784249384741594e-06,
+      "loss": 0.8482,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.682526661197703e-06,
+      "loss": 0.7666,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.686628383921247e-06,
+      "loss": 0.9033,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.690730106644791e-06,
+      "loss": 0.7119,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.694831829368335e-06,
+      "loss": 0.7814,
+      "step": 657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6989335520918788e-06,
+      "loss": 0.7491,
+      "step": 658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7030352748154226e-06,
+      "loss": 0.8098,
+      "step": 659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.707136997538967e-06,
+      "loss": 0.7839,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7112387202625106e-06,
+      "loss": 0.7615,
+      "step": 661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.715340442986054e-06,
+      "loss": 0.9067,
+      "step": 662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.719442165709598e-06,
+      "loss": 0.8262,
+      "step": 663
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7235438884331424e-06,
+      "loss": 0.7225,
+      "step": 664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.727645611156686e-06,
+      "loss": 0.7687,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7317473338802295e-06,
+      "loss": 0.7399,
+      "step": 666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7358490566037738e-06,
+      "loss": 0.8765,
+      "step": 667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.739950779327318e-06,
+      "loss": 0.8252,
+      "step": 668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7440525020508618e-06,
+      "loss": 0.9317,
+      "step": 669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.748154224774405e-06,
+      "loss": 0.9674,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7522559474979494e-06,
+      "loss": 0.7559,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7563576702214936e-06,
+      "loss": 0.9651,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7604593929450374e-06,
+      "loss": 0.8063,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7645611156685807e-06,
+      "loss": 0.7369,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.768662838392125e-06,
+      "loss": 0.7372,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7727645611156687e-06,
+      "loss": 0.884,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.776866283839213e-06,
+      "loss": 0.7597,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7809680065627563e-06,
+      "loss": 0.7177,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7850697292863005e-06,
+      "loss": 0.7156,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7891714520098443e-06,
+      "loss": 0.8908,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.793273174733388e-06,
+      "loss": 0.6994,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.797374897456932e-06,
+      "loss": 0.949,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.801476620180476e-06,
+      "loss": 0.8452,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.80557834290402e-06,
+      "loss": 0.7698,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8096800656275637e-06,
+      "loss": 0.904,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8137817883511075e-06,
+      "loss": 0.8348,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8178835110746517e-06,
+      "loss": 0.6756,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8219852337981955e-06,
+      "loss": 0.811,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8260869565217393e-06,
+      "loss": 0.845,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 0.9186,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8342904019688273e-06,
+      "loss": 0.7303,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.838392124692371e-06,
+      "loss": 0.7592,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.842493847415915e-06,
+      "loss": 0.8516,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8465955701394587e-06,
+      "loss": 0.7212,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.850697292863003e-06,
+      "loss": 0.8124,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8547990155865467e-06,
+      "loss": 0.7835,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8589007383100905e-06,
+      "loss": 0.7341,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8630024610336343e-06,
+      "loss": 0.7788,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8671041837571785e-06,
+      "loss": 0.9666,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8712059064807223e-06,
+      "loss": 0.7514,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8753076292042657e-06,
+      "loss": 0.8744,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.87940935192781e-06,
+      "loss": 0.7396,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.883511074651354e-06,
+      "loss": 0.7539,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.887612797374898e-06,
+      "loss": 0.7658,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8917145200984413e-06,
+      "loss": 0.8101,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8958162428219855e-06,
+      "loss": 0.837,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8999179655455297e-06,
+      "loss": 0.8961,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.904019688269073e-06,
+      "loss": 0.8602,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.908121410992617e-06,
+      "loss": 0.7117,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.912223133716161e-06,
+      "loss": 0.8355,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9163248564397053e-06,
+      "loss": 0.8571,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9204265791632486e-06,
+      "loss": 0.7831,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9245283018867924e-06,
+      "loss": 0.8214,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9286300246103367e-06,
+      "loss": 0.8567,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.932731747333881e-06,
+      "loss": 0.7187,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9368334700574242e-06,
+      "loss": 0.733,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.940935192780968e-06,
+      "loss": 0.8883,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9450369155045122e-06,
+      "loss": 0.7581,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.949138638228056e-06,
+      "loss": 0.7586,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9532403609516e-06,
+      "loss": 0.8062,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9573420836751436e-06,
+      "loss": 0.8191,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.961443806398688e-06,
+      "loss": 0.8754,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9655455291222316e-06,
+      "loss": 0.8375,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9696472518457754e-06,
+      "loss": 0.7994,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9737489745693192e-06,
+      "loss": 0.7421,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9778506972928634e-06,
+      "loss": 0.8878,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9819524200164072e-06,
+      "loss": 0.7038,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.986054142739951e-06,
+      "loss": 0.8935,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.990155865463495e-06,
+      "loss": 0.9341,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.994257588187039e-06,
+      "loss": 0.751,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.998359310910583e-06,
+      "loss": 0.7959,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0024610336341266e-06,
+      "loss": 0.7852,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0065627563576704e-06,
+      "loss": 0.8308,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0106644790812146e-06,
+      "loss": 0.889,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.014766201804758e-06,
+      "loss": 0.7955,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.018867924528302e-06,
+      "loss": 0.7493,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.022969647251846e-06,
+      "loss": 0.7568,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.02707136997539e-06,
+      "loss": 0.6831,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0311730926989336e-06,
+      "loss": 0.6899,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.035274815422478e-06,
+      "loss": 0.8336,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0393765381460216e-06,
+      "loss": 0.7631,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.043478260869566e-06,
+      "loss": 0.6875,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.047579983593109e-06,
+      "loss": 0.8439,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.051681706316653e-06,
+      "loss": 0.7767,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.055783429040197e-06,
+      "loss": 0.8132,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0598851517637414e-06,
+      "loss": 0.7453,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0639868744872848e-06,
+      "loss": 0.645,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0680885972108286e-06,
+      "loss": 0.7817,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0721903199343728e-06,
+      "loss": 0.7719,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.076292042657917e-06,
+      "loss": 0.7754,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0803937653814604e-06,
+      "loss": 0.8808,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.084495488105004e-06,
+      "loss": 0.6695,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0885972108285484e-06,
+      "loss": 0.7808,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0926989335520926e-06,
+      "loss": 0.7673,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.096800656275636e-06,
+      "loss": 0.8108,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1009023789991797e-06,
+      "loss": 0.7911,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.105004101722724e-06,
+      "loss": 0.7617,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.109105824446268e-06,
+      "loss": 0.7847,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1132075471698115e-06,
+      "loss": 0.9657,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1173092698933553e-06,
+      "loss": 0.6915,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1214109926168995e-06,
+      "loss": 0.8365,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.125512715340443e-06,
+      "loss": 0.7283,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.129614438063987e-06,
+      "loss": 0.7287,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.133716160787531e-06,
+      "loss": 0.9432,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.137817883511075e-06,
+      "loss": 0.7233,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1419196062346185e-06,
+      "loss": 0.7509,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1460213289581627e-06,
+      "loss": 0.758,
+      "step": 767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1501230516817065e-06,
+      "loss": 0.7034,
+      "step": 768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1542247744052507e-06,
+      "loss": 0.8628,
+      "step": 769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.158326497128794e-06,
+      "loss": 0.7814,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1624282198523383e-06,
+      "loss": 0.7502,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.166529942575882e-06,
+      "loss": 0.8353,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1706316652994263e-06,
+      "loss": 0.7589,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1747333880229697e-06,
+      "loss": 0.9174,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.178835110746514e-06,
+      "loss": 0.7388,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1829368334700577e-06,
+      "loss": 0.801,
+      "step": 776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.187038556193602e-06,
+      "loss": 0.8423,
+      "step": 777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1911402789171453e-06,
+      "loss": 0.8629,
+      "step": 778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1952420016406895e-06,
+      "loss": 0.7773,
+      "step": 779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1993437243642333e-06,
+      "loss": 0.684,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.2034454470877775e-06,
+      "loss": 0.8176,
+      "step": 781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.207547169811321e-06,
+      "loss": 0.79,
+      "step": 782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.211648892534865e-06,
+      "loss": 0.8279,
+      "step": 783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.215750615258409e-06,
+      "loss": 0.8012,
+      "step": 784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.219852337981953e-06,
+      "loss": 0.8059,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.2239540607054965e-06,
+      "loss": 0.7633,
+      "step": 786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.2280557834290403e-06,
+      "loss": 0.7974,
+      "step": 787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.2321575061525845e-06,
+      "loss": 0.6691,
+      "step": 788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.236259228876128e-06,
+      "loss": 0.7929,
+      "step": 789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.240360951599672e-06,
+      "loss": 0.725,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.244462674323216e-06,
+      "loss": 0.7089,
+      "step": 791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.24856439704676e-06,
+      "loss": 0.7675,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2526661197703034e-06,
+      "loss": 0.757,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2567678424938477e-06,
+      "loss": 0.8159,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2608695652173914e-06,
+      "loss": 0.9464,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2649712879409357e-06,
+      "loss": 0.8589,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.269073010664479e-06,
+      "loss": 0.8614,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2731747333880232e-06,
+      "loss": 0.7441,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.277276456111567e-06,
+      "loss": 0.8766,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2813781788351113e-06,
+      "loss": 0.8536,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2854799015586546e-06,
+      "loss": 0.8176,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.289581624282199e-06,
+      "loss": 0.7437,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2936833470057426e-06,
+      "loss": 1.0374,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.297785069729287e-06,
+      "loss": 0.8471,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.30188679245283e-06,
+      "loss": 0.7942,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3059885151763744e-06,
+      "loss": 0.7432,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3100902378999182e-06,
+      "loss": 0.8793,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3141919606234624e-06,
+      "loss": 0.8161,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.318293683347006e-06,
+      "loss": 0.7289,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.32239540607055e-06,
+      "loss": 0.7483,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.326497128794094e-06,
+      "loss": 0.8186,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.330598851517638e-06,
+      "loss": 0.8165,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3347005742411814e-06,
+      "loss": 0.8631,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3388022969647256e-06,
+      "loss": 0.8129,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3429040196882694e-06,
+      "loss": 0.7441,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3470057424118128e-06,
+      "loss": 0.9126,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.351107465135357e-06,
+      "loss": 0.8468,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.355209187858901e-06,
+      "loss": 0.7752,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.359310910582445e-06,
+      "loss": 0.687,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3634126333059884e-06,
+      "loss": 0.8039,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3675143560295326e-06,
+      "loss": 0.7826,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.371616078753077e-06,
+      "loss": 0.7032,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3757178014766206e-06,
+      "loss": 0.8989,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.379819524200164e-06,
+      "loss": 0.9546,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.383921246923708e-06,
+      "loss": 0.8825,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3880229696472524e-06,
+      "loss": 0.7533,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.392124692370796e-06,
+      "loss": 0.8317,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3962264150943395e-06,
+      "loss": 0.7939,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4003281378178838e-06,
+      "loss": 0.9894,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4044298605414276e-06,
+      "loss": 0.684,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4085315832649718e-06,
+      "loss": 0.7845,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.412633305988515e-06,
+      "loss": 0.8003,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4167350287120594e-06,
+      "loss": 1.0026,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.420836751435603e-06,
+      "loss": 0.8011,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4249384741591474e-06,
+      "loss": 0.8452,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4290401968826907e-06,
+      "loss": 0.787,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.433141919606235e-06,
+      "loss": 0.7748,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4372436423297787e-06,
+      "loss": 0.8828,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.441345365053323e-06,
+      "loss": 0.6669,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4454470877768663e-06,
+      "loss": 0.8493,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4495488105004105e-06,
+      "loss": 0.7613,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4536505332239543e-06,
+      "loss": 0.7728,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.457752255947498e-06,
+      "loss": 0.8617,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.461853978671042e-06,
+      "loss": 0.8435,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.465955701394586e-06,
+      "loss": 0.8174,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.47005742411813e-06,
+      "loss": 0.7491,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4741591468416737e-06,
+      "loss": 0.786,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 0.7922,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4823625922887617e-06,
+      "loss": 0.9157,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4864643150123055e-06,
+      "loss": 0.7679,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4905660377358493e-06,
+      "loss": 0.8246,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.494667760459393e-06,
+      "loss": 0.835,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4987694831829373e-06,
+      "loss": 0.8213,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.502871205906481e-06,
+      "loss": 0.7668,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.506972928630025e-06,
+      "loss": 0.7879,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5110746513535687e-06,
+      "loss": 0.7856,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.515176374077113e-06,
+      "loss": 0.7129,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5192780968006567e-06,
+      "loss": 0.9594,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5233798195242e-06,
+      "loss": 0.8083,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5274815422477443e-06,
+      "loss": 0.8883,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5315832649712885e-06,
+      "loss": 0.7681,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5356849876948323e-06,
+      "loss": 0.6984,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5397867104183757e-06,
+      "loss": 0.7316,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.54388843314192e-06,
+      "loss": 0.8497,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.547990155865464e-06,
+      "loss": 0.8107,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.552091878589008e-06,
+      "loss": 0.8516,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5561936013125513e-06,
+      "loss": 0.7219,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5602953240360955e-06,
+      "loss": 0.7721,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5643970467596397e-06,
+      "loss": 0.8178,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.568498769483183e-06,
+      "loss": 0.8318,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.572600492206727e-06,
+      "loss": 0.7596,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.576702214930271e-06,
+      "loss": 0.8787,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.580803937653815e-06,
+      "loss": 0.7625,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5849056603773586e-06,
+      "loss": 0.962,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5890073831009024e-06,
+      "loss": 0.7894,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5931091058244467e-06,
+      "loss": 0.7978,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5972108285479904e-06,
+      "loss": 0.7968,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6013125512715342e-06,
+      "loss": 0.7076,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.605414273995078e-06,
+      "loss": 0.8766,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6095159967186222e-06,
+      "loss": 0.7414,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.613617719442166e-06,
+      "loss": 0.6526,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.61771944216571e-06,
+      "loss": 0.8893,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6218211648892536e-06,
+      "loss": 0.7797,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.625922887612798e-06,
+      "loss": 0.8177,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6300246103363416e-06,
+      "loss": 0.8375,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6341263330598854e-06,
+      "loss": 0.7322,
+      "step": 886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6382280557834292e-06,
+      "loss": 0.8188,
+      "step": 887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6423297785069734e-06,
+      "loss": 0.9255,
+      "step": 888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6464315012305172e-06,
+      "loss": 0.7856,
+      "step": 889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.650533223954061e-06,
+      "loss": 0.8732,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.654634946677605e-06,
+      "loss": 0.8857,
+      "step": 891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.658736669401149e-06,
+      "loss": 0.7637,
+      "step": 892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.662838392124693e-06,
+      "loss": 0.748,
+      "step": 893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6669401148482366e-06,
+      "loss": 0.6899,
+      "step": 894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6710418375717804e-06,
+      "loss": 0.7467,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6751435602953246e-06,
+      "loss": 0.7317,
+      "step": 896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.679245283018868e-06,
+      "loss": 0.7247,
+      "step": 897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.683347005742412e-06,
+      "loss": 0.7443,
+      "step": 898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.687448728465956e-06,
+      "loss": 0.7008,
+      "step": 899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6915504511895e-06,
+      "loss": 0.8931,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6956521739130436e-06,
+      "loss": 0.6771,
+      "step": 901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6997538966365874e-06,
+      "loss": 0.6794,
+      "step": 902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7038556193601316e-06,
+      "loss": 0.7544,
+      "step": 903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.707957342083676e-06,
+      "loss": 0.7795,
+      "step": 904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.712059064807219e-06,
+      "loss": 0.669,
+      "step": 905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.716160787530763e-06,
+      "loss": 0.7383,
+      "step": 906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.720262510254307e-06,
+      "loss": 0.6825,
+      "step": 907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7243642329778514e-06,
+      "loss": 0.8593,
+      "step": 908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7284659557013948e-06,
+      "loss": 0.8097,
+      "step": 909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7325676784249386e-06,
+      "loss": 0.7653,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7366694011484828e-06,
+      "loss": 0.7387,
+      "step": 911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.740771123872027e-06,
+      "loss": 0.723,
+      "step": 912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7448728465955704e-06,
+      "loss": 0.7061,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.748974569319114e-06,
+      "loss": 0.6868,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7530762920426584e-06,
+      "loss": 0.7385,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.757178014766202e-06,
+      "loss": 0.7878,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.761279737489746e-06,
+      "loss": 0.6709,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7653814602132897e-06,
+      "loss": 0.7702,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.769483182936834e-06,
+      "loss": 0.8824,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7735849056603777e-06,
+      "loss": 0.6653,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7776866283839215e-06,
+      "loss": 0.7765,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7817883511074653e-06,
+      "loss": 0.7945,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7858900738310095e-06,
+      "loss": 0.7067,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.789991796554553e-06,
+      "loss": 0.7882,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.794093519278097e-06,
+      "loss": 0.9232,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.798195242001641e-06,
+      "loss": 0.7408,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.802296964725185e-06,
+      "loss": 0.7114,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8063986874487285e-06,
+      "loss": 0.7327,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8105004101722727e-06,
+      "loss": 0.7152,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8146021328958165e-06,
+      "loss": 0.7188,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.81870385561936e-06,
+      "loss": 0.7174,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.822805578342904e-06,
+      "loss": 0.6291,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.826907301066448e-06,
+      "loss": 0.7759,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8310090237899925e-06,
+      "loss": 0.899,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.835110746513536e-06,
+      "loss": 0.828,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.839212469237079e-06,
+      "loss": 0.8987,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.843314191960624e-06,
+      "loss": 0.9064,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.847415914684168e-06,
+      "loss": 0.8331,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8515176374077115e-06,
+      "loss": 0.7707,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.855619360131255e-06,
+      "loss": 0.7473,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.859721082854799e-06,
+      "loss": 0.9548,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.863822805578344e-06,
+      "loss": 0.7221,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8679245283018875e-06,
+      "loss": 0.7773,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8720262510254305e-06,
+      "loss": 0.8338,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.876127973748975e-06,
+      "loss": 0.7257,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.880229696472519e-06,
+      "loss": 0.7582,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.884331419196063e-06,
+      "loss": 0.8185,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8884331419196065e-06,
+      "loss": 0.7498,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.89253486464315e-06,
+      "loss": 0.7814,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.896636587366695e-06,
+      "loss": 0.7388,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.900738310090238e-06,
+      "loss": 0.8025,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.904840032813782e-06,
+      "loss": 0.9571,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.908941755537326e-06,
+      "loss": 0.7862,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.91304347826087e-06,
+      "loss": 0.7188,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.917145200984414e-06,
+      "loss": 0.7236,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.921246923707958e-06,
+      "loss": 0.6863,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9253486464315014e-06,
+      "loss": 0.8272,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.929450369155045e-06,
+      "loss": 0.7601,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.933552091878589e-06,
+      "loss": 0.7147,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.937653814602133e-06,
+      "loss": 0.8307,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9417555373256775e-06,
+      "loss": 0.729,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.945857260049221e-06,
+      "loss": 0.6672,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.949958982772765e-06,
+      "loss": 0.8494,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.954060705496309e-06,
+      "loss": 0.7878,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.958162428219853e-06,
+      "loss": 0.7353,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.962264150943396e-06,
+      "loss": 0.8107,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.96636587366694e-06,
+      "loss": 0.7696,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970467596390484e-06,
+      "loss": 0.977,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974569319114029e-06,
+      "loss": 0.7448,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9786710418375724e-06,
+      "loss": 0.7387,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.982772764561116e-06,
+      "loss": 0.8371,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.98687448728466e-06,
+      "loss": 0.812,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.990976210008204e-06,
+      "loss": 0.8242,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.995077932731748e-06,
+      "loss": 0.7632,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.999179655455291e-06,
+      "loss": 0.7557,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.003281378178835e-06,
+      "loss": 0.8066,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.00738310090238e-06,
+      "loss": 0.8468,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.011484823625923e-06,
+      "loss": 0.7443,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.0155865463494666e-06,
+      "loss": 0.856,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.019688269073011e-06,
+      "loss": 0.7157,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.023789991796555e-06,
+      "loss": 0.8317,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.027891714520099e-06,
+      "loss": 0.7752,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.031993437243643e-06,
+      "loss": 0.7251,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.036095159967186e-06,
+      "loss": 0.8645,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.040196882690731e-06,
+      "loss": 0.9182,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.044298605414274e-06,
+      "loss": 0.8799,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.048400328137818e-06,
+      "loss": 0.9647,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.052502050861362e-06,
+      "loss": 0.7491,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.056603773584906e-06,
+      "loss": 0.7728,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.06070549630845e-06,
+      "loss": 0.7912,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.064807219031994e-06,
+      "loss": 0.7597,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.0689089417555376e-06,
+      "loss": 0.7125,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.073010664479082e-06,
+      "loss": 0.9142,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.077112387202625e-06,
+      "loss": 0.7854,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.081214109926169e-06,
+      "loss": 0.7481,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.0853158326497136e-06,
+      "loss": 0.8236,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.089417555373257e-06,
+      "loss": 0.7838,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.093519278096801e-06,
+      "loss": 0.8016,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.097621000820345e-06,
+      "loss": 0.7838,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.101722723543889e-06,
+      "loss": 0.8424,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.1058244462674325e-06,
+      "loss": 0.854,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.109926168990976e-06,
+      "loss": 0.7042,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.11402789171452e-06,
+      "loss": 0.8874,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.118129614438065e-06,
+      "loss": 0.9304,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.122231337161608e-06,
+      "loss": 0.7167,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.126333059885152e-06,
+      "loss": 0.8442,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.130434782608696e-06,
+      "loss": 0.7194,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.13453650533224e-06,
+      "loss": 0.9087,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.138638228055784e-06,
+      "loss": 0.8759,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.1427399507793275e-06,
+      "loss": 0.7208,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.146841673502871e-06,
+      "loss": 0.7122,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.150943396226416e-06,
+      "loss": 0.8029,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.155045118949959e-06,
+      "loss": 0.9185,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.1591468416735035e-06,
+      "loss": 0.8026,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.163248564397047e-06,
+      "loss": 0.812,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.167350287120591e-06,
+      "loss": 0.7886,
+      "step": 1016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.171452009844135e-06,
+      "loss": 0.8831,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.175553732567679e-06,
+      "loss": 0.9334,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.1796554552912225e-06,
+      "loss": 0.7033,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.183757178014767e-06,
+      "loss": 0.8052,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.18785890073831e-06,
+      "loss": 0.8142,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.191960623461854e-06,
+      "loss": 0.8057,
+      "step": 1022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.1960623461853985e-06,
+      "loss": 0.8371,
+      "step": 1023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.200164068908942e-06,
+      "loss": 0.867,
+      "step": 1024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.204265791632486e-06,
+      "loss": 0.7535,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.20836751435603e-06,
+      "loss": 0.8278,
+      "step": 1026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.212469237079574e-06,
+      "loss": 0.7496,
+      "step": 1027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.216570959803118e-06,
+      "loss": 0.7575,
+      "step": 1028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.220672682526661e-06,
+      "loss": 0.7035,
+      "step": 1029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.224774405250205e-06,
+      "loss": 0.8901,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.22887612797375e-06,
+      "loss": 0.7357,
+      "step": 1031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.232977850697293e-06,
+      "loss": 0.8149,
+      "step": 1032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.237079573420837e-06,
+      "loss": 0.8377,
+      "step": 1033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.241181296144381e-06,
+      "loss": 0.834,
+      "step": 1034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.245283018867925e-06,
+      "loss": 0.8165,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.249384741591469e-06,
+      "loss": 0.7327,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.2534864643150124e-06,
+      "loss": 0.7386,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.257588187038556e-06,
+      "loss": 0.7813,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.261689909762101e-06,
+      "loss": 0.7286,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.265791632485644e-06,
+      "loss": 0.8529,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.2698933552091885e-06,
+      "loss": 0.9245,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.273995077932732e-06,
+      "loss": 0.7386,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.278096800656276e-06,
+      "loss": 0.775,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.28219852337982e-06,
+      "loss": 0.769,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.286300246103364e-06,
+      "loss": 0.7174,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.290401968826907e-06,
+      "loss": 0.8616,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.294503691550452e-06,
+      "loss": 0.7557,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.298605414273995e-06,
+      "loss": 0.664,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.30270713699754e-06,
+      "loss": 0.7876,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.3068088597210834e-06,
+      "loss": 0.6988,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.310910582444627e-06,
+      "loss": 0.7048,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.315012305168171e-06,
+      "loss": 0.8038,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.319114027891715e-06,
+      "loss": 0.8489,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.323215750615259e-06,
+      "loss": 0.9102,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.327317473338803e-06,
+      "loss": 0.8066,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.331419196062346e-06,
+      "loss": 0.7356,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.335520918785891e-06,
+      "loss": 0.7896,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.339622641509435e-06,
+      "loss": 0.8935,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.3437243642329776e-06,
+      "loss": 0.8609,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 0.8426,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.351927809680066e-06,
+      "loss": 0.7338,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.35602953240361e-06,
+      "loss": 0.7638,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.360131255127154e-06,
+      "loss": 0.7022,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.364232977850697e-06,
+      "loss": 0.7373,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.368334700574241e-06,
+      "loss": 0.8834,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.372436423297786e-06,
+      "loss": 0.8189,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.376538146021329e-06,
+      "loss": 0.8296,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.380639868744873e-06,
+      "loss": 0.9432,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.384741591468417e-06,
+      "loss": 0.7136,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.388843314191961e-06,
+      "loss": 0.7176,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.392945036915505e-06,
+      "loss": 0.7686,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.3970467596390486e-06,
+      "loss": 0.8399,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.401148482362592e-06,
+      "loss": 0.7652,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.405250205086137e-06,
+      "loss": 0.8609,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.40935192780968e-06,
+      "loss": 0.7557,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4134536505332246e-06,
+      "loss": 0.791,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.417555373256768e-06,
+      "loss": 0.7713,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.421657095980312e-06,
+      "loss": 0.6733,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.425758818703856e-06,
+      "loss": 0.9115,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4298605414274e-06,
+      "loss": 0.8486,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4339622641509435e-06,
+      "loss": 0.8326,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.438063986874488e-06,
+      "loss": 0.7071,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.442165709598031e-06,
+      "loss": 0.7948,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.446267432321576e-06,
+      "loss": 0.7141,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4503691550451196e-06,
+      "loss": 0.8117,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.454470877768663e-06,
+      "loss": 0.6933,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.458572600492207e-06,
+      "loss": 0.9112,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.462674323215751e-06,
+      "loss": 0.8143,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.466776045939295e-06,
+      "loss": 0.7261,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4708777686628385e-06,
+      "loss": 0.8145,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.474979491386382e-06,
+      "loss": 0.7096,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.479081214109927e-06,
+      "loss": 0.886,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.483182936833471e-06,
+      "loss": 0.7208,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.487284659557014e-06,
+      "loss": 0.6928,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.491386382280558e-06,
+      "loss": 0.738,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.495488105004102e-06,
+      "loss": 0.7008,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.499589827727646e-06,
+      "loss": 0.697,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.50369155045119e-06,
+      "loss": 0.6777,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5077932731747335e-06,
+      "loss": 0.7405,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.511894995898278e-06,
+      "loss": 0.7844,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.515996718621822e-06,
+      "loss": 1.0399,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.520098441345365e-06,
+      "loss": 0.8416,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5242001640689095e-06,
+      "loss": 0.8865,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.528301886792453e-06,
+      "loss": 0.8626,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.532403609515997e-06,
+      "loss": 0.818,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.536505332239541e-06,
+      "loss": 0.7254,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.540607054963085e-06,
+      "loss": 0.8394,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5447087776866285e-06,
+      "loss": 0.7861,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.548810500410173e-06,
+      "loss": 0.7563,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.552912223133716e-06,
+      "loss": 0.8122,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.557013945857261e-06,
+      "loss": 0.7436,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5611156685808045e-06,
+      "loss": 0.8175,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.565217391304348e-06,
+      "loss": 0.9333,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.569319114027892e-06,
+      "loss": 0.774,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.573420836751436e-06,
+      "loss": 0.7328,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.57752255947498e-06,
+      "loss": 0.7824,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5816242821985234e-06,
+      "loss": 0.6774,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.585726004922067e-06,
+      "loss": 0.7214,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.589827727645612e-06,
+      "loss": 0.8141,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.593929450369156e-06,
+      "loss": 0.7678,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5980311730926995e-06,
+      "loss": 0.816,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.602132895816243e-06,
+      "loss": 0.718,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.606234618539787e-06,
+      "loss": 0.8484,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.610336341263331e-06,
+      "loss": 0.76,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.614438063986875e-06,
+      "loss": 0.7826,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.618539786710418e-06,
+      "loss": 0.9182,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.622641509433963e-06,
+      "loss": 0.8494,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.626743232157507e-06,
+      "loss": 0.7741,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.630844954881051e-06,
+      "loss": 0.7598,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6349466776045944e-06,
+      "loss": 0.671,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.639048400328138e-06,
+      "loss": 0.7507,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.643150123051682e-06,
+      "loss": 0.8138,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.647251845775226e-06,
+      "loss": 0.8393,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.65135356849877e-06,
+      "loss": 0.7482,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.655455291222314e-06,
+      "loss": 0.7074,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.659557013945858e-06,
+      "loss": 0.7653,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.663658736669401e-06,
+      "loss": 1.0065,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.667760459392946e-06,
+      "loss": 0.7215,
+      "step": 1138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.671862182116489e-06,
+      "loss": 0.7602,
+      "step": 1139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.675963904840033e-06,
+      "loss": 0.7707,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.680065627563577e-06,
+      "loss": 0.8843,
+      "step": 1141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.684167350287121e-06,
+      "loss": 0.9455,
+      "step": 1142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6882690730106654e-06,
+      "loss": 0.699,
+      "step": 1143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.692370795734208e-06,
+      "loss": 0.7572,
+      "step": 1144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.696472518457752e-06,
+      "loss": 0.7181,
+      "step": 1145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.700574241181297e-06,
+      "loss": 0.8608,
+      "step": 1146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.704675963904841e-06,
+      "loss": 0.7148,
+      "step": 1147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.708777686628384e-06,
+      "loss": 0.8489,
+      "step": 1148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.712879409351928e-06,
+      "loss": 0.8302,
+      "step": 1149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.716981132075472e-06,
+      "loss": 0.8361,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.721082854799016e-06,
+      "loss": 0.7997,
+      "step": 1151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7251845775225596e-06,
+      "loss": 0.7428,
+      "step": 1152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.729286300246103e-06,
+      "loss": 0.7884,
+      "step": 1153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.733388022969648e-06,
+      "loss": 0.7308,
+      "step": 1154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.737489745693192e-06,
+      "loss": 0.8123,
+      "step": 1155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7415914684167356e-06,
+      "loss": 0.6981,
+      "step": 1156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.745693191140279e-06,
+      "loss": 0.7132,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.749794913863823e-06,
+      "loss": 0.7539,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.753896636587367e-06,
+      "loss": 0.8701,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.757998359310911e-06,
+      "loss": 0.8608,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7621000820344545e-06,
+      "loss": 0.8206,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.766201804757999e-06,
+      "loss": 0.7658,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.770303527481543e-06,
+      "loss": 0.795,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.774405250205087e-06,
+      "loss": 0.7335,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7785069729286305e-06,
+      "loss": 0.855,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.782608695652174e-06,
+      "loss": 0.8392,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.786710418375718e-06,
+      "loss": 0.8312,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.790812141099262e-06,
+      "loss": 0.7318,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.794913863822806e-06,
+      "loss": 0.7644,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.79901558654635e-06,
+      "loss": 0.8277,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.803117309269893e-06,
+      "loss": 0.769,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.807219031993438e-06,
+      "loss": 0.7435,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.811320754716982e-06,
+      "loss": 0.7865,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8154224774405255e-06,
+      "loss": 0.793,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.819524200164069e-06,
+      "loss": 0.8449,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.823625922887613e-06,
+      "loss": 0.9182,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.827727645611157e-06,
+      "loss": 0.8116,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8318293683347015e-06,
+      "loss": 0.7453,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8359310910582445e-06,
+      "loss": 0.7666,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840032813781788e-06,
+      "loss": 0.7688,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844134536505333e-06,
+      "loss": 0.8473,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.848236259228877e-06,
+      "loss": 0.9108,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8523379819524205e-06,
+      "loss": 0.8957,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.856439704675964e-06,
+      "loss": 0.7309,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.860541427399508e-06,
+      "loss": 0.8436,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.864643150123053e-06,
+      "loss": 0.8732,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.868744872846596e-06,
+      "loss": 0.7477,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8728465955701395e-06,
+      "loss": 0.8696,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.876948318293684e-06,
+      "loss": 0.7478,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881050041017228e-06,
+      "loss": 0.673,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885151763740772e-06,
+      "loss": 0.7486,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8892534864643155e-06,
+      "loss": 0.8131,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.893355209187859e-06,
+      "loss": 0.8487,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.897456931911403e-06,
+      "loss": 0.715,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.901558654634947e-06,
+      "loss": 0.6864,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.905660377358491e-06,
+      "loss": 0.7267,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.909762100082035e-06,
+      "loss": 0.7392,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913863822805578e-06,
+      "loss": 0.6826,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917965545529123e-06,
+      "loss": 0.7054,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.922067268252667e-06,
+      "loss": 0.8528,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9261689909762105e-06,
+      "loss": 0.83,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930270713699754e-06,
+      "loss": 0.7626,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.934372436423298e-06,
+      "loss": 0.7808,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.938474159146842e-06,
+      "loss": 0.7742,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9425758818703865e-06,
+      "loss": 0.756,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.946677604593929e-06,
+      "loss": 0.7519,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.950779327317474e-06,
+      "loss": 0.777,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.954881050041018e-06,
+      "loss": 0.7663,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.958982772764562e-06,
+      "loss": 0.6831,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9630844954881054e-06,
+      "loss": 0.6944,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.967186218211649e-06,
+      "loss": 0.7299,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.971287940935193e-06,
+      "loss": 0.826,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.975389663658738e-06,
+      "loss": 0.7802,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.979491386382281e-06,
+      "loss": 0.776,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983593109105825e-06,
+      "loss": 0.9116,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.987694831829369e-06,
+      "loss": 0.8172,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.991796554552913e-06,
+      "loss": 0.6605,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.995898277276457e-06,
+      "loss": 0.6904,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 0.8698498606681824,
+      "eval_runtime": 434.0495,
+      "eval_samples_per_second": 6.909,
+      "eval_steps_per_second": 0.864,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 0.7212,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999999897408053e-06,
+      "loss": 0.7313,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999999589632221e-06,
+      "loss": 0.8731,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999999076672527e-06,
+      "loss": 0.7508,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999998358529016e-06,
+      "loss": 0.7264,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999997435201744e-06,
+      "loss": 0.7783,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.99999630669079e-06,
+      "loss": 0.9493,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999994972996244e-06,
+      "loss": 0.6744,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999993434118216e-06,
+      "loss": 0.853,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999991690056833e-06,
+      "loss": 0.6466,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9999897408122386e-06,
+      "loss": 0.7224,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999987586384591e-06,
+      "loss": 0.7886,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999985226774069e-06,
+      "loss": 0.7632,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999982661980866e-06,
+      "loss": 0.722,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.99997989200519e-06,
+      "loss": 0.9718,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999976916847271e-06,
+      "loss": 0.7868,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9999737365073516e-06,
+      "loss": 0.7849,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999970350985695e-06,
+      "loss": 0.7452,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999966760282577e-06,
+      "loss": 0.841,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999962964398293e-06,
+      "loss": 0.8452,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999958963333154e-06,
+      "loss": 0.784,
+      "step": 1239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.99995475708749e-06,
+      "loss": 0.9473,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999950345661644e-06,
+      "loss": 0.727,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999945729055979e-06,
+      "loss": 1.0332,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999940907270874e-06,
+      "loss": 0.902,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999935880306726e-06,
+      "loss": 0.6599,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999930648163945e-06,
+      "loss": 0.8543,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999925210842962e-06,
+      "loss": 0.7948,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999919568344224e-06,
+      "loss": 0.7764,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999913720668192e-06,
+      "loss": 0.967,
+      "step": 1248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999907667815348e-06,
+      "loss": 0.7698,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999901409786188e-06,
+      "loss": 0.7596,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999894946581225e-06,
+      "loss": 0.7361,
+      "step": 1251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999888278200989e-06,
+      "loss": 0.7331,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9998814046460296e-06,
+      "loss": 0.8318,
+      "step": 1253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999874325916909e-06,
+      "loss": 0.7381,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999867042014209e-06,
+      "loss": 0.8214,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999859552938528e-06,
+      "loss": 0.722,
+      "step": 1256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999851858690478e-06,
+      "loss": 0.8291,
+      "step": 1257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999843959270694e-06,
+      "loss": 0.8362,
+      "step": 1258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999835854679821e-06,
+      "loss": 0.6631,
+      "step": 1259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999827544918527e-06,
+      "loss": 0.7338,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9998190299874925e-06,
+      "loss": 0.7183,
+      "step": 1261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999810309887416e-06,
+      "loss": 0.6901,
+      "step": 1262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999801384619015e-06,
+      "loss": 0.818,
+      "step": 1263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999792254183021e-06,
+      "loss": 0.802,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999782918580183e-06,
+      "loss": 0.7802,
+      "step": 1265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999773377811266e-06,
+      "loss": 0.7534,
+      "step": 1266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999763631877057e-06,
+      "loss": 0.7361,
+      "step": 1267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999753680778352e-06,
+      "loss": 0.9718,
+      "step": 1268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999743524515969e-06,
+      "loss": 0.8507,
+      "step": 1269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999733163090742e-06,
+      "loss": 0.6842,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999722596503521e-06,
+      "loss": 0.7253,
+      "step": 1271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999711824755173e-06,
+      "loss": 0.6831,
+      "step": 1272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999700847846583e-06,
+      "loss": 0.8409,
+      "step": 1273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999689665778651e-06,
+      "loss": 0.9769,
+      "step": 1274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999678278552296e-06,
+      "loss": 0.7245,
+      "step": 1275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.99966668616845e-06,
+      "loss": 0.8215,
+      "step": 1276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999654888628067e-06,
+      "loss": 0.7096,
+      "step": 1277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999642885932115e-06,
+      "loss": 0.9656,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999630678081577e-06,
+      "loss": 0.8564,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999618265077457e-06,
+      "loss": 0.7319,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999605646920773e-06,
+      "loss": 0.7242,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999592823612561e-06,
+      "loss": 0.7951,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999579795153874e-06,
+      "loss": 0.9002,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99956656154578e-06,
+      "loss": 0.7736,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999553122789366e-06,
+      "loss": 0.7502,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9995394788857335e-06,
+      "loss": 0.7279,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999525629836004e-06,
+      "loss": 0.7414,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9995115756413135e-06,
+      "loss": 0.8465,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999497316302816e-06,
+      "loss": 0.999,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999482851821681e-06,
+      "loss": 0.849,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999468182199096e-06,
+      "loss": 0.8178,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9994533074362655e-06,
+      "loss": 0.7093,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99943822753441e-06,
+      "loss": 0.826,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999422942494766e-06,
+      "loss": 0.7663,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99940745231859e-06,
+      "loss": 0.7541,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999391757007152e-06,
+      "loss": 0.6966,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999375856561739e-06,
+      "loss": 0.7009,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999359750983659e-06,
+      "loss": 0.7376,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999343440274233e-06,
+      "loss": 0.8381,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999326924434798e-06,
+      "loss": 1.0381,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9993102034667105e-06,
+      "loss": 0.7961,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999293277371343e-06,
+      "loss": 0.7628,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999276146150085e-06,
+      "loss": 0.6671,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999258809804341e-06,
+      "loss": 0.7036,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9992412683355366e-06,
+      "loss": 0.7014,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999223521745109e-06,
+      "loss": 0.7956,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999205570034516e-06,
+      "loss": 0.7455,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99918741320523e-06,
+      "loss": 0.883,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999169051258741e-06,
+      "loss": 0.9308,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999150484196558e-06,
+      "loss": 0.9363,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999131712020203e-06,
+      "loss": 0.7792,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999112734731217e-06,
+      "loss": 0.8848,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999093552331159e-06,
+      "loss": 0.7427,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999074164821602e-06,
+      "loss": 0.768,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999054572204136e-06,
+      "loss": 0.7668,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999034774480371e-06,
+      "loss": 0.756,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.999014771651932e-06,
+      "loss": 0.7872,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99899456372046e-06,
+      "loss": 0.78,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998974150687613e-06,
+      "loss": 0.6832,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998953532555068e-06,
+      "loss": 0.7904,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998932709324516e-06,
+      "loss": 0.8392,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998911680997665e-06,
+      "loss": 0.8608,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998890447576242e-06,
+      "loss": 0.8689,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99886900906199e-06,
+      "loss": 0.7378,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9988473654566694e-06,
+      "loss": 0.6822,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9988255167620545e-06,
+      "loss": 0.7955,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9988034629799396e-06,
+      "loss": 0.6725,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998781204112136e-06,
+      "loss": 0.7858,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998758740160466e-06,
+      "loss": 0.9575,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998736071126779e-06,
+      "loss": 0.7876,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998713197012932e-06,
+      "loss": 0.7071,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9986901178208034e-06,
+      "loss": 0.7423,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998666833552288e-06,
+      "loss": 0.7554,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9986433442092955e-06,
+      "loss": 0.9263,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998619649793754e-06,
+      "loss": 0.7284,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99859575030761e-06,
+      "loss": 0.8734,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998571645752822e-06,
+      "loss": 0.8888,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998547336131371e-06,
+      "loss": 0.7466,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998522821445251e-06,
+      "loss": 0.895,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998498101696474e-06,
+      "loss": 0.7435,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9984731768870686e-06,
+      "loss": 0.7384,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998448047019081e-06,
+      "loss": 0.8022,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9984227120945734e-06,
+      "loss": 0.8882,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998397172115626e-06,
+      "loss": 0.799,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998371427084333e-06,
+      "loss": 0.8708,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998345477002809e-06,
+      "loss": 0.6804,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998319321873184e-06,
+      "loss": 0.7557,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998292961697605e-06,
+      "loss": 0.7702,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998266396478233e-06,
+      "loss": 0.6796,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99823962621725e-06,
+      "loss": 0.8993,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998212650916854e-06,
+      "loss": 0.9099,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998185470579257e-06,
+      "loss": 0.7129,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998158085206691e-06,
+      "loss": 0.8343,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998130494801402e-06,
+      "loss": 0.7336,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9981026993656576e-06,
+      "loss": 0.7053,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998074698901736e-06,
+      "loss": 0.6878,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998046493411938e-06,
+      "loss": 0.6412,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.998018082898576e-06,
+      "loss": 0.736,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997989467363983e-06,
+      "loss": 0.8771,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997960646810507e-06,
+      "loss": 1.0634,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997931621240514e-06,
+      "loss": 0.6839,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997902390656386e-06,
+      "loss": 0.7298,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997872955060523e-06,
+      "loss": 0.7621,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997843314455339e-06,
+      "loss": 0.7267,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997813468843268e-06,
+      "loss": 0.7446,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99778341822676e-06,
+      "loss": 0.7151,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997753162608279e-06,
+      "loss": 0.7173,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997722701990311e-06,
+      "loss": 0.8528,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997692036375354e-06,
+      "loss": 0.6694,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997661165765925e-06,
+      "loss": 0.662,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997630090164561e-06,
+      "loss": 0.7524,
+      "step": 1371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997598809573807e-06,
+      "loss": 0.7947,
+      "step": 1372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997567323996235e-06,
+      "loss": 0.7185,
+      "step": 1373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997535633434426e-06,
+      "loss": 0.7997,
+      "step": 1374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9975037378909824e-06,
+      "loss": 0.7668,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9974716373685215e-06,
+      "loss": 0.7771,
+      "step": 1376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997439331869679e-06,
+      "loss": 0.6796,
+      "step": 1377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997406821397105e-06,
+      "loss": 0.7286,
+      "step": 1378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997374105953468e-06,
+      "loss": 0.8529,
+      "step": 1379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997341185541453e-06,
+      "loss": 0.7194,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997308060163763e-06,
+      "loss": 0.7252,
+      "step": 1381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997274729823116e-06,
+      "loss": 0.8199,
+      "step": 1382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997241194522246e-06,
+      "loss": 0.7343,
+      "step": 1383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997207454263908e-06,
+      "loss": 0.6707,
+      "step": 1384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997173509050871e-06,
+      "loss": 0.9205,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997139358885918e-06,
+      "loss": 0.7171,
+      "step": 1386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997105003771855e-06,
+      "loss": 0.7935,
+      "step": 1387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9970704437115005e-06,
+      "loss": 0.841,
+      "step": 1388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99703567870769e-06,
+      "loss": 0.7675,
+      "step": 1389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9970007087632785e-06,
+      "loss": 0.8336,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.996965533881135e-06,
+      "loss": 0.9474,
+      "step": 1391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.996930154064147e-06,
+      "loss": 0.6691,
+      "step": 1392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.996894569315218e-06,
+      "loss": 0.7346,
+      "step": 1393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.996858779637269e-06,
+      "loss": 0.7202,
+      "step": 1394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.996822785033236e-06,
+      "loss": 0.7622,
+      "step": 1395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.996786585506075e-06,
+      "loss": 0.8002,
+      "step": 1396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.996750181058755e-06,
+      "loss": 0.6995,
+      "step": 1397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9967135716942665e-06,
+      "loss": 0.7057,
+      "step": 1398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.996676757415611e-06,
+      "loss": 0.7965,
+      "step": 1399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9966397382258136e-06,
+      "loss": 0.9417,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.99660251412791e-06,
+      "loss": 0.7005,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996565085124956e-06,
+      "loss": 0.877,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996527451220023e-06,
+      "loss": 0.7906,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9964896124162e-06,
+      "loss": 0.7856,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996451568716594e-06,
+      "loss": 0.7614,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996413320124326e-06,
+      "loss": 0.9918,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996374866642535e-06,
+      "loss": 0.752,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996336208274377e-06,
+      "loss": 0.731,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996297345023026e-06,
+      "loss": 0.6333,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9962582768916705e-06,
+      "loss": 0.7967,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996219003883517e-06,
+      "loss": 0.6688,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99617952600179e-06,
+      "loss": 0.7591,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996139843249728e-06,
+      "loss": 0.8266,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996099955630589e-06,
+      "loss": 0.7214,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996059863147646e-06,
+      "loss": 0.7789,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.996019565804191e-06,
+      "loss": 0.7571,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995979063603529e-06,
+      "loss": 0.891,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995938356548986e-06,
+      "loss": 0.8313,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995897444643902e-06,
+      "loss": 0.8359,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995856327891635e-06,
+      "loss": 0.8526,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99581500629556e-06,
+      "loss": 0.7985,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995773479859068e-06,
+      "loss": 0.7751,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995731748585568e-06,
+      "loss": 0.7233,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995689812478484e-06,
+      "loss": 0.961,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995647671541258e-06,
+      "loss": 0.7589,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99560532577735e-06,
+      "loss": 0.77,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995562775190233e-06,
+      "loss": 0.8654,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995520019783401e-06,
+      "loss": 0.9,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995477059560363e-06,
+      "loss": 0.7195,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995433894524645e-06,
+      "loss": 0.673,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995390524679789e-06,
+      "loss": 0.6736,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995346950029355e-06,
+      "loss": 0.7952,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995303170576918e-06,
+      "loss": 0.7886,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995259186326074e-06,
+      "loss": 0.7603,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995214997280431e-06,
+      "loss": 0.8436,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995170603443615e-06,
+      "loss": 0.9063,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995126004819272e-06,
+      "loss": 0.7882,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99508120141106e-06,
+      "loss": 0.7563,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995036193222658e-06,
+      "loss": 0.8602,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9949909802577576e-06,
+      "loss": 0.8004,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9949455625200724e-06,
+      "loss": 0.9135,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994899940013328e-06,
+      "loss": 0.7095,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99485411274127e-06,
+      "loss": 0.8403,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9948080807076584e-06,
+      "loss": 0.8136,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994761843916273e-06,
+      "loss": 0.7982,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994715402370907e-06,
+      "loss": 0.7607,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994668756075372e-06,
+      "loss": 0.7481,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994621905033497e-06,
+      "loss": 0.7611,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994574849249128e-06,
+      "loss": 0.7202,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994527588726126e-06,
+      "loss": 0.8378,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9944801234683705e-06,
+      "loss": 0.8118,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994432453479756e-06,
+      "loss": 0.7673,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994384578764196e-06,
+      "loss": 0.7475,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994336499325619e-06,
+      "loss": 0.7065,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9942882151679725e-06,
+      "loss": 0.6958,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994239726295217e-06,
+      "loss": 0.6841,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994191032711335e-06,
+      "loss": 0.8273,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99414213442032e-06,
+      "loss": 0.7858,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994093031426188e-06,
+      "loss": 0.8144,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994043723732967e-06,
+      "loss": 0.7882,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9939942113447045e-06,
+      "loss": 0.7719,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993944494265465e-06,
+      "loss": 0.8353,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993894572499327e-06,
+      "loss": 0.7303,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99384444605039e-06,
+      "loss": 0.7774,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993794114922766e-06,
+      "loss": 0.7169,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993743579120589e-06,
+      "loss": 0.9257,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993692838648003e-06,
+      "loss": 0.7512,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993641893509174e-06,
+      "loss": 0.7561,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993590743708283e-06,
+      "loss": 0.7368,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993539389249529e-06,
+      "loss": 0.7514,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993487830137126e-06,
+      "loss": 0.9049,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9934360663753056e-06,
+      "loss": 0.6864,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993384097968317e-06,
+      "loss": 0.7743,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9933319249204236e-06,
+      "loss": 0.6992,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993279547235909e-06,
+      "loss": 0.8004,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993226964919074e-06,
+      "loss": 0.751,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993174177974229e-06,
+      "loss": 0.8155,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99312118640571e-06,
+      "loss": 0.7265,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993067990217866e-06,
+      "loss": 0.922,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993014589415063e-06,
+      "loss": 0.6905,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9929609840016824e-06,
+      "loss": 0.7042,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992907173982125e-06,
+      "loss": 0.7551,
+      "step": 1482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992853159360807e-06,
+      "loss": 0.7689,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992798940142162e-06,
+      "loss": 0.7752,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992744516330639e-06,
+      "loss": 1.0482,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992689887930705e-06,
+      "loss": 0.9347,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992635054946844e-06,
+      "loss": 0.8194,
+      "step": 1487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992580017383556e-06,
+      "loss": 0.7833,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992524775245357e-06,
+      "loss": 0.8272,
+      "step": 1489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992469328536784e-06,
+      "loss": 0.8322,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992413677262386e-06,
+      "loss": 0.7951,
+      "step": 1491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9923578214267285e-06,
+      "loss": 0.8219,
+      "step": 1492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992301761034398e-06,
+      "loss": 0.8847,
+      "step": 1493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9922454960899965e-06,
+      "loss": 0.7257,
+      "step": 1494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99218902659814e-06,
+      "loss": 0.916,
+      "step": 1495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992132352563464e-06,
+      "loss": 0.7405,
+      "step": 1496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992075473990619e-06,
+      "loss": 0.8453,
+      "step": 1497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992018390884273e-06,
+      "loss": 0.7685,
+      "step": 1498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991961103249113e-06,
+      "loss": 0.7268,
+      "step": 1499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99190361108984e-06,
+      "loss": 0.9198,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991845914411172e-06,
+      "loss": 0.7132,
+      "step": 1501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991788013217844e-06,
+      "loss": 0.7469,
+      "step": 1502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99172990751461e-06,
+      "loss": 0.7507,
+      "step": 1503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9916715973062365e-06,
+      "loss": 0.7292,
+      "step": 1504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991613082597511e-06,
+      "loss": 0.7862,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991554363393235e-06,
+      "loss": 0.849,
+      "step": 1506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991495439698229e-06,
+      "loss": 0.7249,
+      "step": 1507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991436311517329e-06,
+      "loss": 0.7034,
+      "step": 1508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991376978855386e-06,
+      "loss": 0.8161,
+      "step": 1509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991317441717272e-06,
+      "loss": 0.8129,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991257700107871e-06,
+      "loss": 0.7181,
+      "step": 1511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991197754032088e-06,
+      "loss": 0.6722,
+      "step": 1512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991137603494842e-06,
+      "loss": 0.7084,
+      "step": 1513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991077248501071e-06,
+      "loss": 0.7228,
+      "step": 1514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.991016689055728e-06,
+      "loss": 0.8081,
+      "step": 1515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.990955925163782e-06,
+      "loss": 0.7133,
+      "step": 1516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.990894956830222e-06,
+      "loss": 0.7694,
+      "step": 1517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.990833784060051e-06,
+      "loss": 0.7618,
+      "step": 1518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.99077240685829e-06,
+      "loss": 0.7954,
+      "step": 1519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9907108252299755e-06,
+      "loss": 0.6683,
+      "step": 1520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.990649039180163e-06,
+      "loss": 0.7773,
+      "step": 1521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.990587048713923e-06,
+      "loss": 0.7998,
+      "step": 1522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.990524853836343e-06,
+      "loss": 0.9575,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.990462454552528e-06,
+      "loss": 0.743,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.990399850867599e-06,
+      "loss": 0.6961,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9903370427866935e-06,
+      "loss": 0.7498,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.990274030314969e-06,
+      "loss": 0.6916,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.990210813457594e-06,
+      "loss": 0.8133,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.990147392219758e-06,
+      "loss": 0.8167,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.990083766606667e-06,
+      "loss": 0.7036,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.990019936623542e-06,
+      "loss": 0.7235,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989955902275622e-06,
+      "loss": 0.7177,
+      "step": 1532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989891663568163e-06,
+      "loss": 0.7387,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989827220506437e-06,
+      "loss": 0.7427,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989762573095733e-06,
+      "loss": 0.7542,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989697721341356e-06,
+      "loss": 0.7019,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989632665248631e-06,
+      "loss": 0.8058,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989567404822895e-06,
+      "loss": 0.7533,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989501940069506e-06,
+      "loss": 0.7892,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989436270993835e-06,
+      "loss": 0.8102,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989370397601272e-06,
+      "loss": 0.8222,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9893043198972255e-06,
+      "loss": 0.8379,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9892380378871165e-06,
+      "loss": 0.7665,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989171551576386e-06,
+      "loss": 0.8425,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989104860970491e-06,
+      "loss": 0.9374,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989037966074905e-06,
+      "loss": 0.6966,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.988970866895117e-06,
+      "loss": 0.715,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9889035634366354e-06,
+      "loss": 0.7931,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9888360557049834e-06,
+      "loss": 0.7092,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9887683437057026e-06,
+      "loss": 0.6889,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9887004274443486e-06,
+      "loss": 0.7774,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.988632306926497e-06,
+      "loss": 0.8783,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.988563982157739e-06,
+      "loss": 0.8164,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.988495453143681e-06,
+      "loss": 0.7468,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9884267198899475e-06,
+      "loss": 0.7902,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98835778240218e-06,
+      "loss": 0.9727,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.988288640686037e-06,
+      "loss": 0.7484,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.988219294747193e-06,
+      "loss": 0.7128,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.988149744591339e-06,
+      "loss": 0.8489,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.988079990224183e-06,
+      "loss": 0.6821,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.988010031651451e-06,
+      "loss": 0.7329,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987939868878883e-06,
+      "loss": 0.6793,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9878695019122404e-06,
+      "loss": 0.7653,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987798930757296e-06,
+      "loss": 0.7334,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987728155419843e-06,
+      "loss": 0.7867,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987657175905689e-06,
+      "loss": 0.7212,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987585992220662e-06,
+      "loss": 0.6784,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9875146043706e-06,
+      "loss": 0.8699,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987443012361367e-06,
+      "loss": 0.7436,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9873712161988356e-06,
+      "loss": 1.0319,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987299215888899e-06,
+      "loss": 0.6812,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987227011437466e-06,
+      "loss": 0.9122,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987154602850465e-06,
+      "loss": 0.6718,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9870819901338365e-06,
+      "loss": 0.8094,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.987009173293541e-06,
+      "loss": 0.7778,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.986936152335555e-06,
+      "loss": 0.7515,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9868629272658705e-06,
+      "loss": 0.8362,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.986789498090499e-06,
+      "loss": 0.7024,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.986715864815466e-06,
+      "loss": 0.7493,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.986642027446815e-06,
+      "loss": 0.775,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.986567985990606e-06,
+      "loss": 0.8595,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9864937404529155e-06,
+      "loss": 0.7793,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.986419290839839e-06,
+      "loss": 0.7926,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.986344637157484e-06,
+      "loss": 0.798,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9862697794119805e-06,
+      "loss": 0.7426,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.986194717609469e-06,
+      "loss": 0.8047,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.986119451756114e-06,
+      "loss": 0.7626,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98604398185809e-06,
+      "loss": 0.8079,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.985968307921592e-06,
+      "loss": 0.8776,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98589242995283e-06,
+      "loss": 0.773,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.985816347958032e-06,
+      "loss": 0.9095,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.985740061943444e-06,
+      "loss": 0.7739,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9856635719153255e-06,
+      "loss": 0.6677,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.985586877879954e-06,
+      "loss": 0.757,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.985509979843624e-06,
+      "loss": 0.7558,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.985432877812648e-06,
+      "loss": 0.8849,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9853555717933535e-06,
+      "loss": 0.6941,
+      "step": 1597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9852780617920845e-06,
+      "loss": 0.8047,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.985200347815205e-06,
+      "loss": 0.7089,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9851224298690896e-06,
+      "loss": 0.8149,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.985044307960135e-06,
+      "loss": 1.0036,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9849659820947535e-06,
+      "loss": 0.7287,
+      "step": 1602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.984887452279374e-06,
+      "loss": 0.9006,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9848087185204405e-06,
+      "loss": 0.8385,
+      "step": 1604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9847297808244146e-06,
+      "loss": 0.6174,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.984650639197776e-06,
+      "loss": 0.8025,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.984571293647021e-06,
+      "loss": 0.7017,
+      "step": 1607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98449174417866e-06,
+      "loss": 0.8933,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.984411990799222e-06,
+      "loss": 0.7519,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.984332033515253e-06,
+      "loss": 0.8229,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9842518723333166e-06,
+      "loss": 0.6953,
+      "step": 1611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9841715072599906e-06,
+      "loss": 0.7959,
+      "step": 1612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.984090938301871e-06,
+      "loss": 0.7914,
+      "step": 1613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98401016546557e-06,
+      "loss": 0.6959,
+      "step": 1614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.983929188757719e-06,
+      "loss": 0.8151,
+      "step": 1615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.983848008184961e-06,
+      "loss": 0.7689,
+      "step": 1616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98376662375396e-06,
+      "loss": 0.771,
+      "step": 1617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.983685035471397e-06,
+      "loss": 0.692,
+      "step": 1618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9836032433439666e-06,
+      "loss": 0.7763,
+      "step": 1619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.983521247378383e-06,
+      "loss": 0.8598,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.983439047581373e-06,
+      "loss": 0.8364,
+      "step": 1621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.983356643959687e-06,
+      "loss": 0.8403,
+      "step": 1622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.983274036520085e-06,
+      "loss": 0.8001,
+      "step": 1623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9831912252693495e-06,
+      "loss": 0.7468,
+      "step": 1624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9831082102142745e-06,
+      "loss": 0.7855,
+      "step": 1625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9830249913616756e-06,
+      "loss": 0.822,
+      "step": 1626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9829415687183815e-06,
+      "loss": 0.8527,
+      "step": 1627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982857942291239e-06,
+      "loss": 0.6744,
+      "step": 1628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982774112087112e-06,
+      "loss": 0.6809,
+      "step": 1629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98269007811288e-06,
+      "loss": 0.715,
+      "step": 1630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982605840375443e-06,
+      "loss": 0.7825,
+      "step": 1631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982521398881711e-06,
+      "loss": 0.8382,
+      "step": 1632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982436753638616e-06,
+      "loss": 0.6836,
+      "step": 1633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982351904653105e-06,
+      "loss": 0.767,
+      "step": 1634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982266851932142e-06,
+      "loss": 0.7999,
+      "step": 1635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982181595482706e-06,
+      "loss": 0.8002,
+      "step": 1636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982096135311797e-06,
+      "loss": 0.7478,
+      "step": 1637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.982010471426427e-06,
+      "loss": 0.7276,
+      "step": 1638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.981924603833629e-06,
+      "loss": 0.7846,
+      "step": 1639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.981838532540448e-06,
+      "loss": 0.82,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.981752257553948e-06,
+      "loss": 0.7141,
+      "step": 1641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.981665778881212e-06,
+      "loss": 0.6097,
+      "step": 1642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.981579096529337e-06,
+      "loss": 0.7392,
+      "step": 1643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9814922105054356e-06,
+      "loss": 0.7473,
+      "step": 1644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.981405120816642e-06,
+      "loss": 0.8136,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9813178274701e-06,
+      "loss": 0.8578,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.981230330472977e-06,
+      "loss": 0.7157,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.981142629832454e-06,
+      "loss": 0.7703,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.981054725555728e-06,
+      "loss": 0.9497,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980966617650014e-06,
+      "loss": 0.7267,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980878306122543e-06,
+      "loss": 0.7813,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980789790980563e-06,
+      "loss": 0.7896,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.98070107223134e-06,
+      "loss": 0.7959,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980612149882154e-06,
+      "loss": 0.859,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980523023940303e-06,
+      "loss": 0.6817,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9804336944131025e-06,
+      "loss": 0.7383,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9803441613078855e-06,
+      "loss": 0.8777,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980254424631998e-06,
+      "loss": 0.7021,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.980164484392806e-06,
+      "loss": 0.7557,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9800743405976915e-06,
+      "loss": 0.7136,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979983993254052e-06,
+      "loss": 0.8074,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979893442369304e-06,
+      "loss": 0.8457,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979802687950878e-06,
+      "loss": 0.6806,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979711730006223e-06,
+      "loss": 0.7218,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979620568542805e-06,
+      "loss": 0.8184,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979529203568104e-06,
+      "loss": 0.8077,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9794376350896216e-06,
+      "loss": 0.7558,
+      "step": 1667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97934586311487e-06,
+      "loss": 0.7239,
+      "step": 1668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979253887651384e-06,
+      "loss": 0.6959,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97916170870671e-06,
+      "loss": 0.7521,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.979069326288415e-06,
+      "loss": 0.8613,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9789767404040805e-06,
+      "loss": 0.7376,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.978883951061306e-06,
+      "loss": 0.7591,
+      "step": 1673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9787909582677066e-06,
+      "loss": 0.7706,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.978697762030914e-06,
+      "loss": 0.7443,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9786043623585774e-06,
+      "loss": 0.7741,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.978510759258363e-06,
+      "loss": 0.7553,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.978416952737954e-06,
+      "loss": 0.7575,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.978322942805047e-06,
+      "loss": 0.8377,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97822872946736e-06,
+      "loss": 0.7416,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.978134312732624e-06,
+      "loss": 0.6996,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.978039692608588e-06,
+      "loss": 0.7376,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977944869103019e-06,
+      "loss": 0.9042,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977849842223699e-06,
+      "loss": 0.8075,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977754611978427e-06,
+      "loss": 0.8476,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9776591783750185e-06,
+      "loss": 0.7587,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977563541421307e-06,
+      "loss": 0.9127,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977467701125141e-06,
+      "loss": 0.6962,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977371657494388e-06,
+      "loss": 0.9059,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9772754105369275e-06,
+      "loss": 0.6687,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977178960260661e-06,
+      "loss": 0.7005,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977082306673505e-06,
+      "loss": 0.7497,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976985449783392e-06,
+      "loss": 0.8421,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97688838959827e-06,
+      "loss": 0.7634,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976791126126106e-06,
+      "loss": 0.7411,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9766936593748825e-06,
+      "loss": 0.7274,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976595989352599e-06,
+      "loss": 0.9439,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976498116067273e-06,
+      "loss": 0.7254,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976400039526935e-06,
+      "loss": 0.7502,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976301759739635e-06,
+      "loss": 0.8129,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976203276713442e-06,
+      "loss": 0.8382,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976104590456434e-06,
+      "loss": 0.7518,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976005700976715e-06,
+      "loss": 0.8898,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975906608282398e-06,
+      "loss": 0.8323,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975807312381617e-06,
+      "loss": 0.6868,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9757078132825235e-06,
+      "loss": 0.6637,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975608110993281e-06,
+      "loss": 0.7059,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975508205522073e-06,
+      "loss": 0.6473,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9754080968771005e-06,
+      "loss": 0.7479,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975307785066578e-06,
+      "loss": 0.7966,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97520727009874e-06,
+      "loss": 0.8759,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975106551981835e-06,
+      "loss": 0.7839,
+      "step": 1712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975005630724129e-06,
+      "loss": 0.7528,
+      "step": 1713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974904506333906e-06,
+      "loss": 0.7172,
+      "step": 1714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974803178819466e-06,
+      "loss": 0.8037,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974701648189124e-06,
+      "loss": 0.7247,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974599914451214e-06,
+      "loss": 0.9611,
+      "step": 1717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974497977614085e-06,
+      "loss": 0.728,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974395837686102e-06,
+      "loss": 0.8469,
+      "step": 1719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974293494675651e-06,
+      "loss": 0.7428,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974190948591131e-06,
+      "loss": 0.8014,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9740881994409565e-06,
+      "loss": 0.7509,
+      "step": 1722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973985247233561e-06,
+      "loss": 0.7857,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973882091977396e-06,
+      "loss": 0.7071,
+      "step": 1724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973778733680925e-06,
+      "loss": 0.7259,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973675172352633e-06,
+      "loss": 0.7326,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97357140800102e-06,
+      "loss": 0.6904,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973467440634601e-06,
+      "loss": 0.7889,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973363270261908e-06,
+      "loss": 0.7848,
+      "step": 1729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973258896891494e-06,
+      "loss": 0.7215,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973154320531922e-06,
+      "loss": 0.7866,
+      "step": 1731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973049541191776e-06,
+      "loss": 0.8608,
+      "step": 1732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.972944558879657e-06,
+      "loss": 0.7058,
+      "step": 1733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97283937360418e-06,
+      "loss": 0.8613,
+      "step": 1734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9727339853739775e-06,
+      "loss": 0.745,
+      "step": 1735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.972628394197699e-06,
+      "loss": 0.7896,
+      "step": 1736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.972522600084013e-06,
+      "loss": 0.7827,
+      "step": 1737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9724166030416e-06,
+      "loss": 0.7039,
+      "step": 1738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97231040307916e-06,
+      "loss": 0.7351,
+      "step": 1739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97220400020541e-06,
+      "loss": 0.6808,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.972097394429083e-06,
+      "loss": 0.6568,
+      "step": 1741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971990585758927e-06,
+      "loss": 0.6536,
+      "step": 1742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9718835742037094e-06,
+      "loss": 0.8235,
+      "step": 1743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971776359772213e-06,
+      "loss": 0.8414,
+      "step": 1744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971668942473237e-06,
+      "loss": 0.6827,
+      "step": 1745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971561322315597e-06,
+      "loss": 0.7622,
+      "step": 1746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971453499308127e-06,
+      "loss": 0.7469,
+      "step": 1747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971345473459674e-06,
+      "loss": 0.8708,
+      "step": 1748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971237244779108e-06,
+      "loss": 0.7085,
+      "step": 1749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9711288132753085e-06,
+      "loss": 0.8281,
+      "step": 1750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971020178957176e-06,
+      "loss": 0.8285,
+      "step": 1751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.970911341833626e-06,
+      "loss": 0.695,
+      "step": 1752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.970802301913592e-06,
+      "loss": 0.7387,
+      "step": 1753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.970693059206022e-06,
+      "loss": 0.752,
+      "step": 1754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.970583613719883e-06,
+      "loss": 0.8824,
+      "step": 1755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9704739654641575e-06,
+      "loss": 0.9182,
+      "step": 1756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.970364114447845e-06,
+      "loss": 0.6456,
+      "step": 1757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97025406067996e-06,
+      "loss": 0.6969,
+      "step": 1758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.970143804169536e-06,
+      "loss": 0.7118,
+      "step": 1759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.970033344925623e-06,
+      "loss": 0.709,
+      "step": 1760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.969922682957286e-06,
+      "loss": 0.9539,
+      "step": 1761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.969811818273606e-06,
+      "loss": 0.7119,
+      "step": 1762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.969700750883684e-06,
+      "loss": 0.6659,
+      "step": 1763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.969589480796633e-06,
+      "loss": 0.8907,
+      "step": 1764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9694780080215895e-06,
+      "loss": 0.6979,
+      "step": 1765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.969366332567701e-06,
+      "loss": 0.7455,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.969254454444131e-06,
+      "loss": 0.7847,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.969142373660063e-06,
+      "loss": 0.7136,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9690300902246965e-06,
+      "loss": 0.6631,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968917604147246e-06,
+      "loss": 0.7752,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968804915436945e-06,
+      "loss": 0.8098,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968692024103041e-06,
+      "loss": 0.809,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9685789301548e-06,
+      "loss": 0.8147,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968465633601504e-06,
+      "loss": 0.7378,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968352134452451e-06,
+      "loss": 0.8626,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968238432716957e-06,
+      "loss": 0.8234,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968124528404353e-06,
+      "loss": 0.7264,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968010421523989e-06,
+      "loss": 0.8086,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967896112085229e-06,
+      "loss": 0.7909,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967781600097456e-06,
+      "loss": 0.7495,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967666885570066e-06,
+      "loss": 0.7185,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967551968512476e-06,
+      "loss": 0.7295,
+      "step": 1782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967436848934117e-06,
+      "loss": 0.7616,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967321526844439e-06,
+      "loss": 0.7683,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967206002252904e-06,
+      "loss": 0.829,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.967090275168996e-06,
+      "loss": 0.8036,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.96697434560221e-06,
+      "loss": 0.7652,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9668582135620646e-06,
+      "loss": 0.7301,
+      "step": 1788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.966741879058089e-06,
+      "loss": 0.75,
+      "step": 1789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.966625342099831e-06,
+      "loss": 0.7714,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.966508602696857e-06,
+      "loss": 0.7516,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.966391660858746e-06,
+      "loss": 0.7742,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.966274516595096e-06,
+      "loss": 0.6937,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.966157169915523e-06,
+      "loss": 0.713,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.966039620829658e-06,
+      "loss": 0.8123,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.965921869347147e-06,
+      "loss": 0.7962,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.965803915477656e-06,
+      "loss": 0.7203,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.965685759230865e-06,
+      "loss": 0.8743,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.965567400616471e-06,
+      "loss": 0.762,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9654488396441895e-06,
+      "loss": 0.8892,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.96533007632375e-06,
+      "loss": 0.7433,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.965211110664901e-06,
+      "loss": 0.7579,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.965091942677404e-06,
+      "loss": 0.7352,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.964972572371043e-06,
+      "loss": 0.6766,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9648529997556135e-06,
+      "loss": 0.7705,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.964733224840929e-06,
+      "loss": 0.6952,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.964613247636819e-06,
+      "loss": 0.7115,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.964493068153132e-06,
+      "loss": 0.7399,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9643726863997315e-06,
+      "loss": 0.7245,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.964252102386497e-06,
+      "loss": 0.8682,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.964131316123325e-06,
+      "loss": 0.7648,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.96401032762013e-06,
+      "loss": 0.8105,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963889136886841e-06,
+      "loss": 0.8424,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963767743933405e-06,
+      "loss": 0.6703,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963646148769785e-06,
+      "loss": 0.7586,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963524351405961e-06,
+      "loss": 0.8106,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963402351851928e-06,
+      "loss": 0.6686,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963280150117701e-06,
+      "loss": 0.95,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.963157746213308e-06,
+      "loss": 0.8885,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9630351401487955e-06,
+      "loss": 0.7267,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.962912331934228e-06,
+      "loss": 0.7997,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.962789321579681e-06,
+      "loss": 0.8217,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.962666109095254e-06,
+      "loss": 0.9095,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.962542694491057e-06,
+      "loss": 0.6937,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.962419077777221e-06,
+      "loss": 0.6828,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.962295258963891e-06,
+      "loss": 0.7796,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.962171238061228e-06,
+      "loss": 0.8195,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 0.85826575756073,
+      "eval_runtime": 438.2944,
+      "eval_samples_per_second": 6.842,
+      "eval_steps_per_second": 0.856,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.962047015079413e-06,
+      "loss": 0.8644,
+      "step": 1828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.961922590028641e-06,
+      "loss": 0.8123,
+      "step": 1829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.961797962919122e-06,
+      "loss": 0.8473,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9616731337610855e-06,
+      "loss": 0.7183,
+      "step": 1831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.961548102564778e-06,
+      "loss": 0.7267,
+      "step": 1832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.96142286934046e-06,
+      "loss": 0.8488,
+      "step": 1833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.961297434098411e-06,
+      "loss": 0.7639,
+      "step": 1834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.961171796848924e-06,
+      "loss": 0.6537,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.961045957602313e-06,
+      "loss": 0.7939,
+      "step": 1836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.960919916368902e-06,
+      "loss": 0.724,
+      "step": 1837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.960793673159041e-06,
+      "loss": 0.7587,
+      "step": 1838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.960667227983088e-06,
+      "loss": 0.8169,
+      "step": 1839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.960540580851421e-06,
+      "loss": 0.7942,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9604137317744345e-06,
+      "loss": 0.8082,
+      "step": 1841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.96028668076254e-06,
+      "loss": 0.7235,
+      "step": 1842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9601594278261644e-06,
+      "loss": 0.8284,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.960031972975753e-06,
+      "loss": 0.7931,
+      "step": 1844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.959904316221765e-06,
+      "loss": 0.8723,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.959776457574678e-06,
+      "loss": 0.7462,
+      "step": 1846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.959648397044986e-06,
+      "loss": 0.7167,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9595201346432e-06,
+      "loss": 0.7611,
+      "step": 1848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.959391670379846e-06,
+      "loss": 0.7867,
+      "step": 1849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.959263004265468e-06,
+      "loss": 0.7602,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.959134136310626e-06,
+      "loss": 0.8231,
+      "step": 1851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.959005066525897e-06,
+      "loss": 0.7016,
+      "step": 1852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9588757949218735e-06,
+      "loss": 0.6961,
+      "step": 1853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.958746321509166e-06,
+      "loss": 0.8002,
+      "step": 1854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9586166462984e-06,
+      "loss": 0.7593,
+      "step": 1855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.95848676930022e-06,
+      "loss": 0.7408,
+      "step": 1856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9583566905252825e-06,
+      "loss": 0.8093,
+      "step": 1857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.958226409984267e-06,
+      "loss": 0.8098,
+      "step": 1858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.958095927687864e-06,
+      "loss": 0.7961,
+      "step": 1859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9579652436467825e-06,
+      "loss": 0.7973,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.95783435787175e-06,
+      "loss": 0.699,
+      "step": 1861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.957703270373506e-06,
+      "loss": 0.8089,
+      "step": 1862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.957571981162812e-06,
+      "loss": 0.7085,
+      "step": 1863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9574404902504425e-06,
+      "loss": 0.6982,
+      "step": 1864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.957308797647189e-06,
+      "loss": 0.8625,
+      "step": 1865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.95717690336386e-06,
+      "loss": 0.7342,
+      "step": 1866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.957044807411281e-06,
+      "loss": 0.6396,
+      "step": 1867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.956912509800292e-06,
+      "loss": 0.7371,
+      "step": 1868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9567800105417545e-06,
+      "loss": 0.7528,
+      "step": 1869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.95664730964654e-06,
+      "loss": 0.765,
+      "step": 1870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9565144071255415e-06,
+      "loss": 0.6883,
+      "step": 1871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9563813029896655e-06,
+      "loss": 0.763,
+      "step": 1872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.956247997249837e-06,
+      "loss": 0.7608,
+      "step": 1873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.956114489916996e-06,
+      "loss": 0.949,
+      "step": 1874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9559807810021025e-06,
+      "loss": 0.7008,
+      "step": 1875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.955846870516128e-06,
+      "loss": 0.9049,
+      "step": 1876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.955712758470064e-06,
+      "loss": 0.744,
+      "step": 1877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.955578444874917e-06,
+      "loss": 0.6848,
+      "step": 1878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.955443929741711e-06,
+      "loss": 0.8799,
+      "step": 1879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.955309213081486e-06,
+      "loss": 0.6637,
+      "step": 1880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.955174294905298e-06,
+      "loss": 0.7047,
+      "step": 1881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.955039175224222e-06,
+      "loss": 0.7276,
+      "step": 1882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.954903854049347e-06,
+      "loss": 0.7695,
+      "step": 1883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.954768331391778e-06,
+      "loss": 0.7364,
+      "step": 1884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.954632607262639e-06,
+      "loss": 0.8041,
+      "step": 1885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.954496681673069e-06,
+      "loss": 0.7185,
+      "step": 1886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.954360554634225e-06,
+      "loss": 0.7404,
+      "step": 1887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9542242261572774e-06,
+      "loss": 0.8471,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.954087696253417e-06,
+      "loss": 0.7263,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.953950964933848e-06,
+      "loss": 0.7254,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.953814032209793e-06,
+      "loss": 0.7307,
+      "step": 1891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9536768980924904e-06,
+      "loss": 0.7848,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.953539562593196e-06,
+      "loss": 0.8135,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9534020257231805e-06,
+      "loss": 0.7748,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.953264287493732e-06,
+      "loss": 0.6732,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.953126347916156e-06,
+      "loss": 0.7616,
+      "step": 1896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.952988207001773e-06,
+      "loss": 0.7559,
+      "step": 1897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.952849864761921e-06,
+      "loss": 0.7627,
+      "step": 1898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.952711321207954e-06,
+      "loss": 0.6824,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.952572576351244e-06,
+      "loss": 0.7515,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9524336302031766e-06,
+      "loss": 0.7114,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9522944827751555e-06,
+      "loss": 0.7496,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.952155134078602e-06,
+      "loss": 0.7044,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.952015584124953e-06,
+      "loss": 0.718,
+      "step": 1904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.951875832925661e-06,
+      "loss": 0.6688,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.951735880492197e-06,
+      "loss": 0.6411,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.951595726836047e-06,
+      "loss": 0.6996,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.951455371968713e-06,
+      "loss": 0.702,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.951314815901717e-06,
+      "loss": 0.8828,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.951174058646591e-06,
+      "loss": 0.7999,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9510331002148905e-06,
+      "loss": 0.7855,
+      "step": 1911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.950891940618184e-06,
+      "loss": 0.707,
+      "step": 1912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.950750579868055e-06,
+      "loss": 0.8735,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.950609017976108e-06,
+      "loss": 0.7094,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.95046725495396e-06,
+      "loss": 0.8747,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.950325290813246e-06,
+      "loss": 0.8813,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.950183125565619e-06,
+      "loss": 0.8558,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.950040759222745e-06,
+      "loss": 0.9098,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.949898191796311e-06,
+      "loss": 0.707,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.949755423298015e-06,
+      "loss": 0.7744,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.949612453739576e-06,
+      "loss": 0.8304,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.949469283132729e-06,
+      "loss": 0.7126,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.949325911489223e-06,
+      "loss": 0.6693,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.949182338820826e-06,
+      "loss": 0.7787,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.94903856513932e-06,
+      "loss": 0.7207,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.948894590456507e-06,
+      "loss": 0.8013,
+      "step": 1926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9487504147842024e-06,
+      "loss": 0.767,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.948606038134239e-06,
+      "loss": 0.8084,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.948461460518467e-06,
+      "loss": 0.8234,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.948316681948753e-06,
+      "loss": 0.6655,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.948171702436977e-06,
+      "loss": 0.6686,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.948026521995041e-06,
+      "loss": 0.7341,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.947881140634858e-06,
+      "loss": 0.7528,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.947735558368362e-06,
+      "loss": 0.6561,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9475897752075e-06,
+      "loss": 0.7103,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9474437911642376e-06,
+      "loss": 0.6768,
+      "step": 1936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.947297606250556e-06,
+      "loss": 0.8063,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.947151220478454e-06,
+      "loss": 0.7777,
+      "step": 1938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.947004633859943e-06,
+      "loss": 0.7671,
+      "step": 1939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9468578464070576e-06,
+      "loss": 0.7453,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.946710858131844e-06,
+      "loss": 0.7351,
+      "step": 1941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9465636690463645e-06,
+      "loss": 0.6478,
+      "step": 1942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9464162791627015e-06,
+      "loss": 0.7349,
+      "step": 1943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.946268688492951e-06,
+      "loss": 0.7242,
+      "step": 1944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.946120897049226e-06,
+      "loss": 0.7488,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945972904843656e-06,
+      "loss": 0.8038,
+      "step": 1946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945824711888388e-06,
+      "loss": 0.794,
+      "step": 1947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945676318195585e-06,
+      "loss": 0.7974,
+      "step": 1948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945527723777425e-06,
+      "loss": 0.7914,
+      "step": 1949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945378928646105e-06,
+      "loss": 0.6875,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945229932813836e-06,
+      "loss": 0.855,
+      "step": 1951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945080736292847e-06,
+      "loss": 0.7857,
+      "step": 1952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.944931339095384e-06,
+      "loss": 0.728,
+      "step": 1953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.944781741233707e-06,
+      "loss": 0.7241,
+      "step": 1954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.944631942720094e-06,
+      "loss": 0.7012,
+      "step": 1955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.944481943566842e-06,
+      "loss": 0.7416,
+      "step": 1956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.944331743786259e-06,
+      "loss": 0.8208,
+      "step": 1957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.944181343390674e-06,
+      "loss": 0.6837,
+      "step": 1958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.944030742392431e-06,
+      "loss": 0.7813,
+      "step": 1959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.943879940803889e-06,
+      "loss": 0.7021,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.943728938637427e-06,
+      "loss": 0.7103,
+      "step": 1961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9435777359054365e-06,
+      "loss": 0.7083,
+      "step": 1962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.943426332620327e-06,
+      "loss": 0.8268,
+      "step": 1963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9432747287945275e-06,
+      "loss": 0.9143,
+      "step": 1964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9431229244404765e-06,
+      "loss": 0.7352,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.942970919570636e-06,
+      "loss": 0.7972,
+      "step": 1966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.942818714197481e-06,
+      "loss": 0.7271,
+      "step": 1967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9426663083335034e-06,
+      "loss": 0.712,
+      "step": 1968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.942513701991212e-06,
+      "loss": 0.8189,
+      "step": 1969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.942360895183131e-06,
+      "loss": 0.7628,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.942207887921802e-06,
+      "loss": 0.5967,
+      "step": 1971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9420546802197835e-06,
+      "loss": 0.7313,
+      "step": 1972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.941901272089649e-06,
+      "loss": 0.8652,
+      "step": 1973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.94174766354399e-06,
+      "loss": 0.723,
+      "step": 1974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.941593854595413e-06,
+      "loss": 0.7401,
+      "step": 1975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.941439845256542e-06,
+      "loss": 0.694,
+      "step": 1976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9412856355400175e-06,
+      "loss": 0.764,
+      "step": 1977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.941131225458496e-06,
+      "loss": 0.7008,
+      "step": 1978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.940976615024649e-06,
+      "loss": 0.7984,
+      "step": 1979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9408218042511665e-06,
+      "loss": 0.7673,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.940666793150756e-06,
+      "loss": 0.7707,
+      "step": 1981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9405115817361385e-06,
+      "loss": 0.8276,
+      "step": 1982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.940356170020053e-06,
+      "loss": 0.9174,
+      "step": 1983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9402005580152545e-06,
+      "loss": 0.7866,
+      "step": 1984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.940044745734516e-06,
+      "loss": 0.7689,
+      "step": 1985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.939888733190623e-06,
+      "loss": 0.8438,
+      "step": 1986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9397325203963815e-06,
+      "loss": 0.731,
+      "step": 1987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.939576107364612e-06,
+      "loss": 0.7497,
+      "step": 1988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.939419494108153e-06,
+      "loss": 0.8353,
+      "step": 1989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.939262680639857e-06,
+      "loss": 0.7548,
+      "step": 1990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9391056669725944e-06,
+      "loss": 0.8367,
+      "step": 1991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9389484531192535e-06,
+      "loss": 0.8397,
+      "step": 1992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.938791039092734e-06,
+      "loss": 0.8178,
+      "step": 1993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.938633424905959e-06,
+      "loss": 0.8548,
+      "step": 1994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.938475610571863e-06,
+      "loss": 0.9715,
+      "step": 1995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.938317596103398e-06,
+      "loss": 0.6286,
+      "step": 1996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.938159381513533e-06,
+      "loss": 0.7595,
+      "step": 1997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.938000966815254e-06,
+      "loss": 0.8444,
+      "step": 1998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.937842352021561e-06,
+      "loss": 0.9097,
+      "step": 1999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.937683537145475e-06,
+      "loss": 0.7314,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.937524522200026e-06,
+      "loss": 0.7473,
+      "step": 2001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.93736530719827e-06,
+      "loss": 0.7039,
+      "step": 2002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.93720589215327e-06,
+      "loss": 0.8671,
+      "step": 2003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.937046277078113e-06,
+      "loss": 0.8273,
+      "step": 2004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.936886461985897e-06,
+      "loss": 0.8026,
+      "step": 2005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9367264468897405e-06,
+      "loss": 0.8692,
+      "step": 2006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.936566231802775e-06,
+      "loss": 0.7398,
+      "step": 2007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.936405816738149e-06,
+      "loss": 0.9101,
+      "step": 2008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.936245201709031e-06,
+      "loss": 0.7171,
+      "step": 2009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9360843867286024e-06,
+      "loss": 0.7979,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.935923371810061e-06,
+      "loss": 0.8363,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.935762156966621e-06,
+      "loss": 0.7121,
+      "step": 2012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.935600742211517e-06,
+      "loss": 0.8812,
+      "step": 2013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9354391275579935e-06,
+      "loss": 0.836,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.935277313019318e-06,
+      "loss": 0.7491,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.935115298608768e-06,
+      "loss": 0.7088,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.934953084339643e-06,
+      "loss": 0.8809,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.934790670225256e-06,
+      "loss": 0.7924,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.934628056278935e-06,
+      "loss": 0.757,
+      "step": 2019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9344652425140286e-06,
+      "loss": 0.6738,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.934302228943899e-06,
+      "loss": 0.897,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.934139015581925e-06,
+      "loss": 0.8355,
+      "step": 2022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.933975602441502e-06,
+      "loss": 0.7114,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.933811989536042e-06,
+      "loss": 0.9,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.933648176878974e-06,
+      "loss": 0.839,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9334841644837405e-06,
+      "loss": 0.8583,
+      "step": 2026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.933319952363805e-06,
+      "loss": 0.7144,
+      "step": 2027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.933155540532644e-06,
+      "loss": 0.7574,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.932990929003751e-06,
+      "loss": 0.7135,
+      "step": 2029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.932826117790637e-06,
+      "loss": 0.6692,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.932661106906829e-06,
+      "loss": 0.7802,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.932495896365868e-06,
+      "loss": 0.7011,
+      "step": 2032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.932330486181316e-06,
+      "loss": 1.042,
+      "step": 2033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9321648763667475e-06,
+      "loss": 0.7812,
+      "step": 2034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.931999066935754e-06,
+      "loss": 0.6975,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.931833057901946e-06,
+      "loss": 0.677,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.931666849278946e-06,
+      "loss": 0.8234,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9315004410803965e-06,
+      "loss": 0.7529,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9313338333199565e-06,
+      "loss": 0.7261,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9311670260112985e-06,
+      "loss": 0.9676,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.931000019168113e-06,
+      "loss": 0.8758,
+      "step": 2041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.930832812804108e-06,
+      "loss": 0.8337,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.930665406933005e-06,
+      "loss": 0.8561,
+      "step": 2043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.930497801568546e-06,
+      "loss": 0.7933,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.930329996724484e-06,
+      "loss": 0.7894,
+      "step": 2045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.930161992414594e-06,
+      "loss": 0.7219,
+      "step": 2046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.929993788652663e-06,
+      "loss": 0.7048,
+      "step": 2047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.929825385452496e-06,
+      "loss": 0.7987,
+      "step": 2048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.929656782827916e-06,
+      "loss": 0.727,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.92948798079276e-06,
+      "loss": 0.66,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.929318979360882e-06,
+      "loss": 0.7932,
+      "step": 2051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9291497785461515e-06,
+      "loss": 0.8504,
+      "step": 2052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.928980378362458e-06,
+      "loss": 0.7295,
+      "step": 2053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.928810778823702e-06,
+      "loss": 0.6558,
+      "step": 2054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.928640979943806e-06,
+      "loss": 0.7551,
+      "step": 2055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9284709817367035e-06,
+      "loss": 0.6968,
+      "step": 2056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9283007842163485e-06,
+      "loss": 0.7918,
+      "step": 2057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.928130387396709e-06,
+      "loss": 0.7106,
+      "step": 2058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927959791291769e-06,
+      "loss": 0.7634,
+      "step": 2059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927788995915532e-06,
+      "loss": 0.818,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927618001282015e-06,
+      "loss": 0.7366,
+      "step": 2061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9274468074052524e-06,
+      "loss": 0.6664,
+      "step": 2062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927275414299293e-06,
+      "loss": 0.6553,
+      "step": 2063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.927103821978206e-06,
+      "loss": 0.6945,
+      "step": 2064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926932030456073e-06,
+      "loss": 0.7459,
+      "step": 2065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926760039746994e-06,
+      "loss": 0.7366,
+      "step": 2066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926587849865085e-06,
+      "loss": 0.6641,
+      "step": 2067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926415460824479e-06,
+      "loss": 0.8362,
+      "step": 2068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926242872639322e-06,
+      "loss": 1.0034,
+      "step": 2069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.926070085323782e-06,
+      "loss": 0.7062,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925897098892038e-06,
+      "loss": 0.7596,
+      "step": 2071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925723913358289e-06,
+      "loss": 0.8078,
+      "step": 2072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925550528736749e-06,
+      "loss": 0.8184,
+      "step": 2073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925376945041648e-06,
+      "loss": 0.7159,
+      "step": 2074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925203162287231e-06,
+      "loss": 0.8005,
+      "step": 2075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.925029180487764e-06,
+      "loss": 0.7169,
+      "step": 2076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.924854999657524e-06,
+      "loss": 0.718,
+      "step": 2077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.924680619810807e-06,
+      "loss": 0.8588,
+      "step": 2078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.924506040961925e-06,
+      "loss": 0.6507,
+      "step": 2079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9243312631252084e-06,
+      "loss": 0.8142,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.924156286314998e-06,
+      "loss": 0.7204,
+      "step": 2081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.923981110545658e-06,
+      "loss": 0.8376,
+      "step": 2082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.923805735831565e-06,
+      "loss": 0.8191,
+      "step": 2083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.923630162187111e-06,
+      "loss": 0.7231,
+      "step": 2084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.923454389626708e-06,
+      "loss": 0.801,
+      "step": 2085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.923278418164781e-06,
+      "loss": 0.8063,
+      "step": 2086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.923102247815773e-06,
+      "loss": 0.7623,
+      "step": 2087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.922925878594143e-06,
+      "loss": 0.7807,
+      "step": 2088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.922749310514366e-06,
+      "loss": 0.6473,
+      "step": 2089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9225725435909335e-06,
+      "loss": 0.6957,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.922395577838355e-06,
+      "loss": 0.6719,
+      "step": 2091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.922218413271151e-06,
+      "loss": 0.8973,
+      "step": 2092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.922041049903866e-06,
+      "loss": 0.7787,
+      "step": 2093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921863487751056e-06,
+      "loss": 0.7362,
+      "step": 2094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921685726827292e-06,
+      "loss": 0.7595,
+      "step": 2095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921507767147165e-06,
+      "loss": 0.6982,
+      "step": 2096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921329608725281e-06,
+      "loss": 0.6938,
+      "step": 2097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921151251576262e-06,
+      "loss": 0.6834,
+      "step": 2098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920972695714746e-06,
+      "loss": 0.7997,
+      "step": 2099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920793941155387e-06,
+      "loss": 0.6741,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920614987912857e-06,
+      "loss": 0.7299,
+      "step": 2101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920435836001844e-06,
+      "loss": 0.7611,
+      "step": 2102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.92025648543705e-06,
+      "loss": 0.8548,
+      "step": 2103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9200769362331965e-06,
+      "loss": 0.8063,
+      "step": 2104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919897188405019e-06,
+      "loss": 0.6828,
+      "step": 2105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.91971724196727e-06,
+      "loss": 0.7119,
+      "step": 2106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919537096934718e-06,
+      "loss": 0.7965,
+      "step": 2107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919356753322149e-06,
+      "loss": 0.667,
+      "step": 2108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919176211144364e-06,
+      "loss": 0.7874,
+      "step": 2109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.91899547041618e-06,
+      "loss": 0.8018,
+      "step": 2110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918814531152432e-06,
+      "loss": 0.7194,
+      "step": 2111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918633393367971e-06,
+      "loss": 0.8023,
+      "step": 2112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9184520570776615e-06,
+      "loss": 0.8183,
+      "step": 2113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918270522296388e-06,
+      "loss": 0.6942,
+      "step": 2114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.91808878903905e-06,
+      "loss": 0.7859,
+      "step": 2115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917906857320561e-06,
+      "loss": 0.7886,
+      "step": 2116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917724727155856e-06,
+      "loss": 0.6159,
+      "step": 2117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.91754239855988e-06,
+      "loss": 0.9275,
+      "step": 2118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917359871547599e-06,
+      "loss": 0.7812,
+      "step": 2119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917177146133992e-06,
+      "loss": 0.627,
+      "step": 2120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.916994222334058e-06,
+      "loss": 0.7917,
+      "step": 2121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.916811100162809e-06,
+      "loss": 0.6757,
+      "step": 2122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.916627779635276e-06,
+      "loss": 0.8349,
+      "step": 2123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.916444260766502e-06,
+      "loss": 0.724,
+      "step": 2124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.916260543571552e-06,
+      "loss": 0.6546,
+      "step": 2125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.916076628065502e-06,
+      "loss": 0.9355,
+      "step": 2126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.915892514263447e-06,
+      "loss": 0.7217,
+      "step": 2127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9157082021804994e-06,
+      "loss": 0.7686,
+      "step": 2128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.915523691831786e-06,
+      "loss": 0.8127,
+      "step": 2129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9153389832324485e-06,
+      "loss": 0.8181,
+      "step": 2130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.915154076397647e-06,
+      "loss": 0.7905,
+      "step": 2131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.914968971342559e-06,
+      "loss": 0.7495,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914783668082377e-06,
+      "loss": 0.7722,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9145981666323065e-06,
+      "loss": 0.7172,
+      "step": 2134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914412467007575e-06,
+      "loss": 0.7567,
+      "step": 2135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914226569223422e-06,
+      "loss": 0.6764,
+      "step": 2136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914040473295107e-06,
+      "loss": 0.8972,
+      "step": 2137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9138541792379e-06,
+      "loss": 0.7046,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913667687067094e-06,
+      "loss": 0.7164,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913480996797994e-06,
+      "loss": 0.7625,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913294108445922e-06,
+      "loss": 0.8134,
+      "step": 2141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913107022026216e-06,
+      "loss": 0.797,
+      "step": 2142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912919737554233e-06,
+      "loss": 0.7352,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912732255045343e-06,
+      "loss": 0.6622,
+      "step": 2144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912544574514932e-06,
+      "loss": 0.7401,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912356695978405e-06,
+      "loss": 0.7588,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912168619451181e-06,
+      "loss": 0.8167,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.911980344948698e-06,
+      "loss": 0.6462,
+      "step": 2148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9117918724864064e-06,
+      "loss": 0.717,
+      "step": 2149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.911603202079775e-06,
+      "loss": 0.731,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91141433374429e-06,
+      "loss": 0.762,
+      "step": 2151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9112252674954515e-06,
+      "loss": 0.7568,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.911036003348777e-06,
+      "loss": 0.7294,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9108465413198e-06,
+      "loss": 0.7772,
+      "step": 2154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91065688142407e-06,
+      "loss": 0.8668,
+      "step": 2155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.910467023677153e-06,
+      "loss": 0.7739,
+      "step": 2156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.910276968094633e-06,
+      "loss": 0.7336,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.910086714692106e-06,
+      "loss": 0.6966,
+      "step": 2158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.909896263485189e-06,
+      "loss": 0.7734,
+      "step": 2159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.909705614489511e-06,
+      "loss": 0.704,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.909514767720721e-06,
+      "loss": 0.7709,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.909323723194482e-06,
+      "loss": 0.7962,
+      "step": 2162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.909132480926473e-06,
+      "loss": 0.6891,
+      "step": 2163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9089410409323905e-06,
+      "loss": 0.7508,
+      "step": 2164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9087494032279474e-06,
+      "loss": 0.7617,
+      "step": 2165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.90855756782887e-06,
+      "loss": 0.7261,
+      "step": 2166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.908365534750905e-06,
+      "loss": 0.78,
+      "step": 2167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.908173304009811e-06,
+      "loss": 0.7548,
+      "step": 2168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.907980875621368e-06,
+      "loss": 0.7756,
+      "step": 2169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.907788249601367e-06,
+      "loss": 0.7161,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9075954259656185e-06,
+      "loss": 0.7708,
+      "step": 2171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.907402404729948e-06,
+      "loss": 0.9603,
+      "step": 2172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.907209185910198e-06,
+      "loss": 0.7226,
+      "step": 2173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.907015769522225e-06,
+      "loss": 0.6807,
+      "step": 2174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.906822155581905e-06,
+      "loss": 0.7691,
+      "step": 2175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.906628344105128e-06,
+      "loss": 0.8517,
+      "step": 2176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9064343351078e-06,
+      "loss": 0.792,
+      "step": 2177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.906240128605846e-06,
+      "loss": 0.7022,
+      "step": 2178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.906045724615204e-06,
+      "loss": 0.7788,
+      "step": 2179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.905851123151829e-06,
+      "loss": 0.7473,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.905656324231693e-06,
+      "loss": 0.7807,
+      "step": 2181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9054613278707845e-06,
+      "loss": 0.6579,
+      "step": 2182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.905266134085106e-06,
+      "loss": 0.8617,
+      "step": 2183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9050707428906795e-06,
+      "loss": 0.746,
+      "step": 2184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.90487515430354e-06,
+      "loss": 0.7668,
+      "step": 2185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.904679368339742e-06,
+      "loss": 0.7224,
+      "step": 2186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.904483385015352e-06,
+      "loss": 0.7381,
+      "step": 2187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.904287204346457e-06,
+      "loss": 0.7789,
+      "step": 2188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.904090826349157e-06,
+      "loss": 0.7624,
+      "step": 2189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.90389425103957e-06,
+      "loss": 0.7256,
+      "step": 2190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9036974784338305e-06,
+      "loss": 0.7102,
+      "step": 2191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9035005085480865e-06,
+      "loss": 0.8812,
+      "step": 2192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9033033413985056e-06,
+      "loss": 0.8372,
+      "step": 2193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9031059770012685e-06,
+      "loss": 0.6853,
+      "step": 2194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.902908415372575e-06,
+      "loss": 0.796,
+      "step": 2195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9027106565286385e-06,
+      "loss": 0.7487,
+      "step": 2196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.90251270048569e-06,
+      "loss": 0.7905,
+      "step": 2197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.902314547259978e-06,
+      "loss": 0.8984,
+      "step": 2198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.902116196867764e-06,
+      "loss": 0.8245,
+      "step": 2199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.901917649325328e-06,
+      "loss": 0.7842,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.901718904648965e-06,
+      "loss": 0.7836,
+      "step": 2201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.901519962854987e-06,
+      "loss": 0.6277,
+      "step": 2202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.901320823959722e-06,
+      "loss": 0.8107,
+      "step": 2203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.901121487979513e-06,
+      "loss": 0.7562,
+      "step": 2204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.900921954930722e-06,
+      "loss": 0.7935,
+      "step": 2205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.900722224829725e-06,
+      "loss": 0.7084,
+      "step": 2206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.900522297692913e-06,
+      "loss": 0.7784,
+      "step": 2207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.900322173536697e-06,
+      "loss": 0.7881,
+      "step": 2208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9001218523775e-06,
+      "loss": 0.72,
+      "step": 2209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8999213342317634e-06,
+      "loss": 0.8703,
+      "step": 2210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.899720619115944e-06,
+      "loss": 0.63,
+      "step": 2211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.899519707046517e-06,
+      "loss": 0.7031,
+      "step": 2212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.899318598039971e-06,
+      "loss": 0.7233,
+      "step": 2213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.899117292112812e-06,
+      "loss": 0.7107,
+      "step": 2214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.898915789281561e-06,
+      "loss": 0.7516,
+      "step": 2215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.898714089562757e-06,
+      "loss": 0.7268,
+      "step": 2216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.898512192972953e-06,
+      "loss": 0.6835,
+      "step": 2217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.898310099528721e-06,
+      "loss": 0.768,
+      "step": 2218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.898107809246646e-06,
+      "loss": 0.7501,
+      "step": 2219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.897905322143333e-06,
+      "loss": 0.799,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.897702638235397e-06,
+      "loss": 0.7369,
+      "step": 2221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.897499757539478e-06,
+      "loss": 0.6938,
+      "step": 2222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.897296680072222e-06,
+      "loss": 0.688,
+      "step": 2223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8970934058503e-06,
+      "loss": 0.799,
+      "step": 2224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.896889934890394e-06,
+      "loss": 0.7298,
+      "step": 2225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.896686267209203e-06,
+      "loss": 0.7292,
+      "step": 2226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.896482402823444e-06,
+      "loss": 0.7454,
+      "step": 2227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.896278341749848e-06,
+      "loss": 0.6892,
+      "step": 2228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8960740840051635e-06,
+      "loss": 0.7966,
+      "step": 2229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.895869629606154e-06,
+      "loss": 0.8244,
+      "step": 2230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.895664978569601e-06,
+      "loss": 0.762,
+      "step": 2231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.895460130912301e-06,
+      "loss": 0.7786,
+      "step": 2232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.895255086651064e-06,
+      "loss": 0.7578,
+      "step": 2233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.895049845802722e-06,
+      "loss": 0.7278,
+      "step": 2234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.894844408384117e-06,
+      "loss": 0.7392,
+      "step": 2235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.894638774412113e-06,
+      "loss": 0.7456,
+      "step": 2236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.894432943903584e-06,
+      "loss": 0.7855,
+      "step": 2237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.894226916875425e-06,
+      "loss": 0.7349,
+      "step": 2238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8940206933445455e-06,
+      "loss": 0.8122,
+      "step": 2239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.89381427332787e-06,
+      "loss": 0.8264,
+      "step": 2240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.893607656842341e-06,
+      "loss": 0.7355,
+      "step": 2241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.893400843904916e-06,
+      "loss": 0.6743,
+      "step": 2242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.893193834532569e-06,
+      "loss": 0.6953,
+      "step": 2243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.89298662874229e-06,
+      "loss": 0.8208,
+      "step": 2244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8927792265510845e-06,
+      "loss": 0.7995,
+      "step": 2245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.892571627975975e-06,
+      "loss": 0.7329,
+      "step": 2246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.892363833034e-06,
+      "loss": 0.7494,
+      "step": 2247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.892155841742214e-06,
+      "loss": 0.8407,
+      "step": 2248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8919476541176884e-06,
+      "loss": 0.7668,
+      "step": 2249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.891739270177508e-06,
+      "loss": 0.8492,
+      "step": 2250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.891530689938778e-06,
+      "loss": 0.6982,
+      "step": 2251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.891321913418615e-06,
+      "loss": 0.7701,
+      "step": 2252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.891112940634155e-06,
+      "loss": 0.7514,
+      "step": 2253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.89090377160255e-06,
+      "loss": 0.7367,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.890694406340965e-06,
+      "loss": 0.883,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8904848448665855e-06,
+      "loss": 0.6976,
+      "step": 2256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.890275087196611e-06,
+      "loss": 0.8294,
+      "step": 2257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8900651333482555e-06,
+      "loss": 0.8709,
+      "step": 2258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.889854983338751e-06,
+      "loss": 0.7188,
+      "step": 2259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.889644637185347e-06,
+      "loss": 0.8665,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.889434094905305e-06,
+      "loss": 0.6767,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.889223356515907e-06,
+      "loss": 0.7882,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.889012422034447e-06,
+      "loss": 0.8075,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.888801291478238e-06,
+      "loss": 0.7756,
+      "step": 2264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.888589964864609e-06,
+      "loss": 0.7757,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8883784422109044e-06,
+      "loss": 0.7323,
+      "step": 2266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8881667235344834e-06,
+      "loss": 0.8579,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8879548088527225e-06,
+      "loss": 0.6735,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.887742698183017e-06,
+      "loss": 0.7947,
+      "step": 2269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.887530391542771e-06,
+      "loss": 0.7538,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.887317888949413e-06,
+      "loss": 0.8726,
+      "step": 2271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.887105190420382e-06,
+      "loss": 0.7334,
+      "step": 2272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.886892295973136e-06,
+      "loss": 0.7055,
+      "step": 2273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.886679205625148e-06,
+      "loss": 0.8214,
+      "step": 2274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.886465919393905e-06,
+      "loss": 0.7142,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.886252437296915e-06,
+      "loss": 0.7012,
+      "step": 2276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.886038759351697e-06,
+      "loss": 0.8365,
+      "step": 2277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.88582488557579e-06,
+      "loss": 0.7626,
+      "step": 2278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.885610815986747e-06,
+      "loss": 0.8121,
+      "step": 2279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.885396550602137e-06,
+      "loss": 0.6896,
+      "step": 2280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.885182089439545e-06,
+      "loss": 0.7505,
+      "step": 2281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884967432516574e-06,
+      "loss": 0.7116,
+      "step": 2282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884752579850841e-06,
+      "loss": 0.7079,
+      "step": 2283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884537531459978e-06,
+      "loss": 0.6987,
+      "step": 2284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884322287361638e-06,
+      "loss": 0.7118,
+      "step": 2285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884106847573485e-06,
+      "loss": 0.7428,
+      "step": 2286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883891212113201e-06,
+      "loss": 0.8293,
+      "step": 2287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883675380998484e-06,
+      "loss": 0.7711,
+      "step": 2288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883459354247048e-06,
+      "loss": 0.7849,
+      "step": 2289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883243131876624e-06,
+      "loss": 0.6371,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.883026713904956e-06,
+      "loss": 0.7151,
+      "step": 2291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.882810100349808e-06,
+      "loss": 0.6948,
+      "step": 2292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.882593291228959e-06,
+      "loss": 0.6959,
+      "step": 2293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.882376286560201e-06,
+      "loss": 0.9193,
+      "step": 2294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8821590863613464e-06,
+      "loss": 0.7551,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.88194169065022e-06,
+      "loss": 0.6834,
+      "step": 2296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.881724099444665e-06,
+      "loss": 0.6802,
+      "step": 2297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8815063127625395e-06,
+      "loss": 0.7454,
+      "step": 2298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.881288330621719e-06,
+      "loss": 0.6949,
+      "step": 2299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.881070153040093e-06,
+      "loss": 0.809,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.880851780035568e-06,
+      "loss": 0.8886,
+      "step": 2301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.880633211626067e-06,
+      "loss": 0.7668,
+      "step": 2302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.880414447829529e-06,
+      "loss": 0.8653,
+      "step": 2303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8801954886639095e-06,
+      "loss": 0.7553,
+      "step": 2304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.879976334147178e-06,
+      "loss": 0.8109,
+      "step": 2305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.879756984297321e-06,
+      "loss": 0.7595,
+      "step": 2306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8795374391323426e-06,
+      "loss": 0.6238,
+      "step": 2307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.87931769867026e-06,
+      "loss": 0.7518,
+      "step": 2308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.87909776292911e-06,
+      "loss": 0.8746,
+      "step": 2309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.878877631926942e-06,
+      "loss": 0.7209,
+      "step": 2310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.878657305681823e-06,
+      "loss": 0.6825,
+      "step": 2311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.878436784211838e-06,
+      "loss": 0.7562,
+      "step": 2312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8782160675350835e-06,
+      "loss": 0.8415,
+      "step": 2313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.877995155669676e-06,
+      "loss": 0.7101,
+      "step": 2314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.877774048633745e-06,
+      "loss": 0.7589,
+      "step": 2315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8775527464454385e-06,
+      "loss": 0.6974,
+      "step": 2316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.877331249122921e-06,
+      "loss": 0.8353,
+      "step": 2317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8771095566843685e-06,
+      "loss": 0.6907,
+      "step": 2318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.876887669147978e-06,
+      "loss": 0.8244,
+      "step": 2319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.876665586531961e-06,
+      "loss": 0.721,
+      "step": 2320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.876443308854544e-06,
+      "loss": 0.8875,
+      "step": 2321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.876220836133969e-06,
+      "loss": 0.9087,
+      "step": 2322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.875998168388497e-06,
+      "loss": 0.7045,
+      "step": 2323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.875775305636403e-06,
+      "loss": 0.7343,
+      "step": 2324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.875552247895976e-06,
+      "loss": 0.6923,
+      "step": 2325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.875328995185526e-06,
+      "loss": 0.7249,
+      "step": 2326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.875105547523373e-06,
+      "loss": 0.8033,
+      "step": 2327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8748819049278595e-06,
+      "loss": 0.9029,
+      "step": 2328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.874658067417338e-06,
+      "loss": 0.676,
+      "step": 2329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.874434035010181e-06,
+      "loss": 0.6744,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.874209807724776e-06,
+      "loss": 0.8562,
+      "step": 2331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.873985385579524e-06,
+      "loss": 0.7077,
+      "step": 2332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.873760768592847e-06,
+      "loss": 0.672,
+      "step": 2333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.873535956783177e-06,
+      "loss": 0.9142,
+      "step": 2334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8733109501689675e-06,
+      "loss": 0.8843,
+      "step": 2335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.873085748768685e-06,
+      "loss": 0.7628,
+      "step": 2336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.872860352600814e-06,
+      "loss": 0.6926,
+      "step": 2337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8726347616838494e-06,
+      "loss": 0.6692,
+      "step": 2338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.87240897603631e-06,
+      "loss": 0.7089,
+      "step": 2339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.872182995676727e-06,
+      "loss": 0.7855,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.871956820623644e-06,
+      "loss": 0.6991,
+      "step": 2341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.871730450895627e-06,
+      "loss": 0.8948,
+      "step": 2342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8715038865112545e-06,
+      "loss": 0.6321,
+      "step": 2343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.87127712748912e-06,
+      "loss": 0.763,
+      "step": 2344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.871050173847836e-06,
+      "loss": 0.7697,
+      "step": 2345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.870823025606028e-06,
+      "loss": 0.6862,
+      "step": 2346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8705956827823415e-06,
+      "loss": 0.7223,
+      "step": 2347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.870368145395432e-06,
+      "loss": 0.7006,
+      "step": 2348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.870140413463976e-06,
+      "loss": 0.9352,
+      "step": 2349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.869912487006665e-06,
+      "loss": 0.6699,
+      "step": 2350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.869684366042205e-06,
+      "loss": 0.6455,
+      "step": 2351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.869456050589317e-06,
+      "loss": 0.8539,
+      "step": 2352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.869227540666742e-06,
+      "loss": 0.754,
+      "step": 2353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.868998836293234e-06,
+      "loss": 0.7564,
+      "step": 2354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.868769937487563e-06,
+      "loss": 0.6817,
+      "step": 2355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.868540844268516e-06,
+      "loss": 0.7427,
+      "step": 2356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.868311556654896e-06,
+      "loss": 0.672,
+      "step": 2357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.86808207466552e-06,
+      "loss": 0.8667,
+      "step": 2358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.867852398319224e-06,
+      "loss": 0.7355,
+      "step": 2359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.867622527634858e-06,
+      "loss": 0.7661,
+      "step": 2360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.867392462631287e-06,
+      "loss": 0.8538,
+      "step": 2361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.867162203327394e-06,
+      "loss": 0.7086,
+      "step": 2362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.866931749742078e-06,
+      "loss": 0.9462,
+      "step": 2363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8667011018942525e-06,
+      "loss": 0.6726,
+      "step": 2364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.866470259802848e-06,
+      "loss": 0.6118,
+      "step": 2365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.866239223486809e-06,
+      "loss": 0.8025,
+      "step": 2366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8660079929651e-06,
+      "loss": 0.7056,
+      "step": 2367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.865776568256697e-06,
+      "loss": 0.779,
+      "step": 2368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.865544949380594e-06,
+      "loss": 0.7297,
+      "step": 2369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.865313136355801e-06,
+      "loss": 0.6811,
+      "step": 2370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.865081129201345e-06,
+      "loss": 0.7393,
+      "step": 2371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.864848927936266e-06,
+      "loss": 0.8272,
+      "step": 2372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.864616532579622e-06,
+      "loss": 0.6943,
+      "step": 2373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.864383943150487e-06,
+      "loss": 0.7192,
+      "step": 2374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.864151159667949e-06,
+      "loss": 0.7607,
+      "step": 2375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.863918182151116e-06,
+      "loss": 0.7045,
+      "step": 2376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.863685010619107e-06,
+      "loss": 0.7416,
+      "step": 2377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.863451645091061e-06,
+      "loss": 0.7268,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.863218085586129e-06,
+      "loss": 0.7352,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.862984332123481e-06,
+      "loss": 0.7141,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.862750384722302e-06,
+      "loss": 0.8946,
+      "step": 2381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.862516243401794e-06,
+      "loss": 0.6685,
+      "step": 2382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.862281908181172e-06,
+      "loss": 1.0052,
+      "step": 2383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8620473790796705e-06,
+      "loss": 0.7018,
+      "step": 2384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.861812656116537e-06,
+      "loss": 0.735,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.861577739311036e-06,
+      "loss": 0.7449,
+      "step": 2386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.861342628682448e-06,
+      "loss": 0.748,
+      "step": 2387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.861107324250069e-06,
+      "loss": 0.8039,
+      "step": 2388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.860871826033213e-06,
+      "loss": 0.922,
+      "step": 2389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.860636134051207e-06,
+      "loss": 0.7037,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.860400248323395e-06,
+      "loss": 0.6676,
+      "step": 2391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.860164168869137e-06,
+      "loss": 0.6826,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.859927895707809e-06,
+      "loss": 0.9982,
+      "step": 2393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8596914288588025e-06,
+      "loss": 0.8631,
+      "step": 2394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8594547683415254e-06,
+      "loss": 0.9066,
+      "step": 2395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8592179141754014e-06,
+      "loss": 0.7519,
+      "step": 2396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.85898086637987e-06,
+      "loss": 0.8491,
+      "step": 2397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8587436249743865e-06,
+      "loss": 0.7002,
+      "step": 2398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.858506189978423e-06,
+      "loss": 0.6688,
+      "step": 2399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8582685614114635e-06,
+      "loss": 0.7137,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.858030739293015e-06,
+      "loss": 0.737,
+      "step": 2401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.857792723642593e-06,
+      "loss": 0.8016,
+      "step": 2402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.857554514479736e-06,
+      "loss": 0.8172,
+      "step": 2403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.857316111823992e-06,
+      "loss": 0.7489,
+      "step": 2404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.857077515694927e-06,
+      "loss": 0.8225,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.856838726112126e-06,
+      "loss": 0.7596,
+      "step": 2406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.856599743095185e-06,
+      "loss": 0.8572,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.856360566663719e-06,
+      "loss": 0.7318,
+      "step": 2408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.856121196837359e-06,
+      "loss": 0.7399,
+      "step": 2409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.855881633635749e-06,
+      "loss": 0.7024,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.855641877078553e-06,
+      "loss": 0.7783,
+      "step": 2411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.855401927185447e-06,
+      "loss": 0.7071,
+      "step": 2412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.855161783976125e-06,
+      "loss": 0.8572,
+      "step": 2413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.854921447470297e-06,
+      "loss": 0.7607,
+      "step": 2414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.854680917687688e-06,
+      "loss": 0.6944,
+      "step": 2415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.854440194648037e-06,
+      "loss": 0.7919,
+      "step": 2416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.854199278371105e-06,
+      "loss": 0.7274,
+      "step": 2417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.853958168876662e-06,
+      "loss": 0.7953,
+      "step": 2418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.853716866184497e-06,
+      "loss": 0.7993,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.853475370314416e-06,
+      "loss": 0.8162,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.853233681286238e-06,
+      "loss": 0.766,
+      "step": 2421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8529917991198e-06,
+      "loss": 0.6703,
+      "step": 2422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.852749723834953e-06,
+      "loss": 0.8163,
+      "step": 2423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.852507455451567e-06,
+      "loss": 0.6621,
+      "step": 2424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.852264993989524e-06,
+      "loss": 0.613,
+      "step": 2425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.852022339468725e-06,
+      "loss": 0.7267,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.851779491909084e-06,
+      "loss": 0.7782,
+      "step": 2427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.851536451330534e-06,
+      "loss": 0.745,
+      "step": 2428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.851293217753021e-06,
+      "loss": 0.7059,
+      "step": 2429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.851049791196508e-06,
+      "loss": 0.7991,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.850806171680975e-06,
+      "loss": 0.7143,
+      "step": 2431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.850562359226417e-06,
+      "loss": 0.8892,
+      "step": 2432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.850318353852843e-06,
+      "loss": 0.7931,
+      "step": 2433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.85007415558028e-06,
+      "loss": 0.8116,
+      "step": 2434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8498297644287695e-06,
+      "loss": 0.7421,
+      "step": 2435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.849585180418372e-06,
+      "loss": 0.7463,
+      "step": 2436
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 0.8474946022033691,
+      "eval_runtime": 433.9764,
+      "eval_samples_per_second": 6.911,
+      "eval_steps_per_second": 0.864,
+      "step": 2436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.849340403569158e-06,
+      "loss": 0.8756,
+      "step": 2437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.84909543390122e-06,
+      "loss": 0.7329,
+      "step": 2438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.848850271434662e-06,
+      "loss": 0.6534,
+      "step": 2439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.848604916189605e-06,
+      "loss": 0.7345,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.848359368186189e-06,
+      "loss": 0.6808,
+      "step": 2441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8481136274445625e-06,
+      "loss": 0.7127,
+      "step": 2442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.847867693984899e-06,
+      "loss": 0.6707,
+      "step": 2443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.84762156782738e-06,
+      "loss": 0.8546,
+      "step": 2444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.847375248992206e-06,
+      "loss": 0.8012,
+      "step": 2445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.847128737499595e-06,
+      "loss": 0.6596,
+      "step": 2446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.846882033369779e-06,
+      "loss": 0.7578,
+      "step": 2447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.846635136623004e-06,
+      "loss": 1.0282,
+      "step": 2448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.846388047279535e-06,
+      "loss": 0.6552,
+      "step": 2449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8461407653596515e-06,
+      "loss": 0.7101,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.845893290883648e-06,
+      "loss": 0.8004,
+      "step": 2451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.845645623871837e-06,
+      "loss": 0.6849,
+      "step": 2452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.845397764344544e-06,
+      "loss": 0.7439,
+      "step": 2453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.845149712322113e-06,
+      "loss": 0.6775,
+      "step": 2454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.844901467824902e-06,
+      "loss": 0.6241,
+      "step": 2455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.844653030873284e-06,
+      "loss": 0.8295,
+      "step": 2456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.844404401487651e-06,
+      "loss": 0.888,
+      "step": 2457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.844155579688408e-06,
+      "loss": 0.8984,
+      "step": 2458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.843906565495977e-06,
+      "loss": 0.7342,
+      "step": 2459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.843657358930796e-06,
+      "loss": 0.7567,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.843407960013316e-06,
+      "loss": 0.7901,
+      "step": 2461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.843158368764009e-06,
+      "loss": 0.7159,
+      "step": 2462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.842908585203357e-06,
+      "loss": 0.8145,
+      "step": 2463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.842658609351864e-06,
+      "loss": 0.7074,
+      "step": 2464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.842408441230043e-06,
+      "loss": 0.6906,
+      "step": 2465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.842158080858429e-06,
+      "loss": 0.767,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.841907528257568e-06,
+      "loss": 0.7124,
+      "step": 2467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.841656783448025e-06,
+      "loss": 0.6461,
+      "step": 2468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.841405846450378e-06,
+      "loss": 0.9081,
+      "step": 2469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.841154717285223e-06,
+      "loss": 0.7968,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8409033959731725e-06,
+      "loss": 0.8489,
+      "step": 2471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.840651882534852e-06,
+      "loss": 0.7682,
+      "step": 2472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.840400176990904e-06,
+      "loss": 0.7247,
+      "step": 2473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.840148279361987e-06,
+      "loss": 0.9043,
+      "step": 2474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8398961896687745e-06,
+      "loss": 0.6617,
+      "step": 2475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.839643907931958e-06,
+      "loss": 0.7531,
+      "step": 2476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.839391434172242e-06,
+      "loss": 0.7398,
+      "step": 2477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8391387684103485e-06,
+      "loss": 0.6353,
+      "step": 2478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.838885910667015e-06,
+      "loss": 0.7914,
+      "step": 2479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8386328609629925e-06,
+      "loss": 0.7215,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.838379619319053e-06,
+      "loss": 0.8256,
+      "step": 2481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.838126185755977e-06,
+      "loss": 0.8267,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.837872560294567e-06,
+      "loss": 0.7236,
+      "step": 2483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.837618742955638e-06,
+      "loss": 0.7318,
+      "step": 2484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8373647337600235e-06,
+      "loss": 0.8145,
+      "step": 2485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.83711053272857e-06,
+      "loss": 0.8141,
+      "step": 2486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.83685613988214e-06,
+      "loss": 0.6934,
+      "step": 2487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.836601555241612e-06,
+      "loss": 0.7102,
+      "step": 2488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.836346778827883e-06,
+      "loss": 0.7204,
+      "step": 2489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8360918106618604e-06,
+      "loss": 0.7428,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.835836650764473e-06,
+      "loss": 0.9053,
+      "step": 2491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.835581299156661e-06,
+      "loss": 0.7545,
+      "step": 2492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.835325755859382e-06,
+      "loss": 0.6875,
+      "step": 2493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.835070020893611e-06,
+      "loss": 0.8539,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.834814094280335e-06,
+      "loss": 0.7165,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.834557976040561e-06,
+      "loss": 0.6956,
+      "step": 2496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8343016661953075e-06,
+      "loss": 0.6793,
+      "step": 2497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8340451647656125e-06,
+      "loss": 0.7037,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.833788471772526e-06,
+      "loss": 0.6975,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.833531587237117e-06,
+      "loss": 0.7812,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.833274511180469e-06,
+      "loss": 0.7092,
+      "step": 2501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8330172436236815e-06,
+      "loss": 0.7013,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8327597845878675e-06,
+      "loss": 0.7462,
+      "step": 2503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.832502134094159e-06,
+      "loss": 0.7099,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.832244292163703e-06,
+      "loss": 0.7883,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.831986258817659e-06,
+      "loss": 0.6894,
+      "step": 2506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8317280340772076e-06,
+      "loss": 0.7883,
+      "step": 2507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.83146961796354e-06,
+      "loss": 0.6876,
+      "step": 2508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.831211010497867e-06,
+      "loss": 0.8012,
+      "step": 2509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.830952211701413e-06,
+      "loss": 0.8968,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.830693221595418e-06,
+      "loss": 0.7591,
+      "step": 2511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8304340402011375e-06,
+      "loss": 0.7567,
+      "step": 2512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.830174667539845e-06,
+      "loss": 0.6424,
+      "step": 2513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.829915103632827e-06,
+      "loss": 0.6954,
+      "step": 2514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.829655348501388e-06,
+      "loss": 0.8077,
+      "step": 2515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.829395402166847e-06,
+      "loss": 0.7787,
+      "step": 2516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8291352646505375e-06,
+      "loss": 0.7822,
+      "step": 2517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.828874935973811e-06,
+      "loss": 0.728,
+      "step": 2518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.828614416158033e-06,
+      "loss": 0.6864,
+      "step": 2519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.828353705224585e-06,
+      "loss": 0.7605,
+      "step": 2520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.828092803194865e-06,
+      "loss": 0.8071,
+      "step": 2521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.827831710090286e-06,
+      "loss": 0.6951,
+      "step": 2522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.827570425932277e-06,
+      "loss": 0.8733,
+      "step": 2523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.827308950742283e-06,
+      "loss": 0.8687,
+      "step": 2524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.827047284541763e-06,
+      "loss": 0.6349,
+      "step": 2525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.826785427352194e-06,
+      "loss": 0.8368,
+      "step": 2526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.826523379195066e-06,
+      "loss": 0.8709,
+      "step": 2527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.826261140091888e-06,
+      "loss": 0.7708,
+      "step": 2528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.825998710064183e-06,
+      "loss": 0.8254,
+      "step": 2529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.825736089133487e-06,
+      "loss": 0.7331,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.825473277321357e-06,
+      "loss": 0.7467,
+      "step": 2531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8252102746493615e-06,
+      "loss": 0.7196,
+      "step": 2532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.824947081139086e-06,
+      "loss": 0.7686,
+      "step": 2533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.824683696812133e-06,
+      "loss": 0.6592,
+      "step": 2534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.824420121690119e-06,
+      "loss": 0.8123,
+      "step": 2535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.824156355794675e-06,
+      "loss": 0.7447,
+      "step": 2536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.82389239914745e-06,
+      "loss": 0.7965,
+      "step": 2537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.82362825177011e-06,
+      "loss": 0.7686,
+      "step": 2538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.823363913684332e-06,
+      "loss": 0.7529,
+      "step": 2539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.823099384911811e-06,
+      "loss": 0.7342,
+      "step": 2540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.822834665474259e-06,
+      "loss": 0.6771,
+      "step": 2541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.822569755393403e-06,
+      "loss": 0.7859,
+      "step": 2542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.822304654690983e-06,
+      "loss": 0.9063,
+      "step": 2543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.822039363388759e-06,
+      "loss": 0.7245,
+      "step": 2544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.821773881508502e-06,
+      "loss": 0.8973,
+      "step": 2545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.821508209072004e-06,
+      "loss": 0.6487,
+      "step": 2546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.821242346101067e-06,
+      "loss": 0.7545,
+      "step": 2547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.820976292617513e-06,
+      "loss": 0.8922,
+      "step": 2548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.820710048643177e-06,
+      "loss": 0.7416,
+      "step": 2549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.820443614199911e-06,
+      "loss": 0.9902,
+      "step": 2550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.820176989309583e-06,
+      "loss": 0.692,
+      "step": 2551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.819910173994074e-06,
+      "loss": 0.8268,
+      "step": 2552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.819643168275285e-06,
+      "loss": 0.7853,
+      "step": 2553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.819375972175128e-06,
+      "loss": 0.6768,
+      "step": 2554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.819108585715533e-06,
+      "loss": 0.7722,
+      "step": 2555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8188410089184455e-06,
+      "loss": 0.796,
+      "step": 2556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.818573241805828e-06,
+      "loss": 0.7259,
+      "step": 2557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.818305284399655e-06,
+      "loss": 0.6528,
+      "step": 2558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.818037136721919e-06,
+      "loss": 0.7411,
+      "step": 2559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.81776879879463e-06,
+      "loss": 0.7264,
+      "step": 2560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.817500270639808e-06,
+      "loss": 0.7195,
+      "step": 2561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.817231552279495e-06,
+      "loss": 0.7109,
+      "step": 2562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.816962643735745e-06,
+      "loss": 0.8034,
+      "step": 2563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.816693545030626e-06,
+      "loss": 0.6843,
+      "step": 2564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.816424256186227e-06,
+      "loss": 0.6721,
+      "step": 2565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.816154777224649e-06,
+      "loss": 0.8231,
+      "step": 2566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.815885108168007e-06,
+      "loss": 0.7147,
+      "step": 2567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.815615249038435e-06,
+      "loss": 0.7366,
+      "step": 2568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.815345199858082e-06,
+      "loss": 0.6869,
+      "step": 2569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.815074960649111e-06,
+      "loss": 0.7802,
+      "step": 2570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.814804531433702e-06,
+      "loss": 0.7208,
+      "step": 2571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.814533912234049e-06,
+      "loss": 0.7539,
+      "step": 2572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.814263103072364e-06,
+      "loss": 0.7192,
+      "step": 2573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.813992103970872e-06,
+      "loss": 0.6855,
+      "step": 2574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.813720914951817e-06,
+      "loss": 0.7882,
+      "step": 2575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.813449536037455e-06,
+      "loss": 0.6826,
+      "step": 2576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.813177967250058e-06,
+      "loss": 0.8175,
+      "step": 2577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.812906208611916e-06,
+      "loss": 1.0951,
+      "step": 2578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8126342601453336e-06,
+      "loss": 0.7331,
+      "step": 2579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8123621218726294e-06,
+      "loss": 0.6546,
+      "step": 2580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.812089793816139e-06,
+      "loss": 0.8492,
+      "step": 2581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.811817275998214e-06,
+      "loss": 0.6295,
+      "step": 2582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.811544568441221e-06,
+      "loss": 0.7531,
+      "step": 2583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.81127167116754e-06,
+      "loss": 0.7478,
+      "step": 2584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.810998584199571e-06,
+      "loss": 0.702,
+      "step": 2585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.810725307559727e-06,
+      "loss": 0.7811,
+      "step": 2586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.810451841270436e-06,
+      "loss": 0.7639,
+      "step": 2587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.810178185354142e-06,
+      "loss": 0.716,
+      "step": 2588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.809904339833306e-06,
+      "loss": 0.7403,
+      "step": 2589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.809630304730403e-06,
+      "loss": 0.6951,
+      "step": 2590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.809356080067923e-06,
+      "loss": 0.7801,
+      "step": 2591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.809081665868375e-06,
+      "loss": 0.7409,
+      "step": 2592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.808807062154279e-06,
+      "loss": 0.8183,
+      "step": 2593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8085322689481724e-06,
+      "loss": 0.7048,
+      "step": 2594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.808257286272611e-06,
+      "loss": 0.665,
+      "step": 2595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.807982114150162e-06,
+      "loss": 0.7296,
+      "step": 2596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.807706752603409e-06,
+      "loss": 0.7707,
+      "step": 2597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.807431201654954e-06,
+      "loss": 0.609,
+      "step": 2598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.807155461327411e-06,
+      "loss": 0.8095,
+      "step": 2599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8068795316434105e-06,
+      "loss": 0.6605,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8066034126256e-06,
+      "loss": 0.8928,
+      "step": 2601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.806327104296641e-06,
+      "loss": 0.7197,
+      "step": 2602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.806050606679211e-06,
+      "loss": 0.6799,
+      "step": 2603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.805773919796005e-06,
+      "loss": 0.699,
+      "step": 2604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.80549704366973e-06,
+      "loss": 0.7375,
+      "step": 2605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.805219978323109e-06,
+      "loss": 0.8445,
+      "step": 2606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8049427237788845e-06,
+      "loss": 0.747,
+      "step": 2607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.80466528005981e-06,
+      "loss": 0.6632,
+      "step": 2608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.804387647188657e-06,
+      "loss": 0.6699,
+      "step": 2609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.804109825188211e-06,
+      "loss": 0.6117,
+      "step": 2610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8038318140812746e-06,
+      "loss": 0.6631,
+      "step": 2611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8035536138906655e-06,
+      "loss": 0.8196,
+      "step": 2612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.803275224639215e-06,
+      "loss": 0.861,
+      "step": 2613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.802996646349773e-06,
+      "loss": 0.8163,
+      "step": 2614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8027178790452035e-06,
+      "loss": 0.7132,
+      "step": 2615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8024389227483844e-06,
+      "loss": 0.6609,
+      "step": 2616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.802159777482213e-06,
+      "loss": 0.7334,
+      "step": 2617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.801880443269597e-06,
+      "loss": 0.6775,
+      "step": 2618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.801600920133464e-06,
+      "loss": 0.8215,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.801321208096756e-06,
+      "loss": 0.7843,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.801041307182428e-06,
+      "loss": 0.8369,
+      "step": 2621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.800761217413455e-06,
+      "loss": 0.8191,
+      "step": 2622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.800480938812821e-06,
+      "loss": 0.8713,
+      "step": 2623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.800200471403534e-06,
+      "loss": 0.7235,
+      "step": 2624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.79991981520861e-06,
+      "loss": 0.7568,
+      "step": 2625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.799638970251086e-06,
+      "loss": 0.6491,
+      "step": 2626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.799357936554009e-06,
+      "loss": 0.7373,
+      "step": 2627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.799076714140446e-06,
+      "loss": 0.6819,
+      "step": 2628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.798795303033477e-06,
+      "loss": 0.7635,
+      "step": 2629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7985137032562005e-06,
+      "loss": 0.8214,
+      "step": 2630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.798231914831727e-06,
+      "loss": 0.7874,
+      "step": 2631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.797949937783183e-06,
+      "loss": 0.7246,
+      "step": 2632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.797667772133713e-06,
+      "loss": 0.8379,
+      "step": 2633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7973854179064734e-06,
+      "loss": 0.8057,
+      "step": 2634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.797102875124641e-06,
+      "loss": 0.6047,
+      "step": 2635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.796820143811402e-06,
+      "loss": 0.6789,
+      "step": 2636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.796537223989964e-06,
+      "loss": 0.828,
+      "step": 2637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7962541156835435e-06,
+      "loss": 0.7601,
+      "step": 2638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.795970818915379e-06,
+      "loss": 0.7659,
+      "step": 2639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.795687333708722e-06,
+      "loss": 0.6607,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.795403660086838e-06,
+      "loss": 0.8726,
+      "step": 2641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.795119798073009e-06,
+      "loss": 0.706,
+      "step": 2642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.794835747690533e-06,
+      "loss": 0.7626,
+      "step": 2643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.794551508962723e-06,
+      "loss": 0.8481,
+      "step": 2644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.794267081912908e-06,
+      "loss": 0.8955,
+      "step": 2645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7939824665644305e-06,
+      "loss": 0.7236,
+      "step": 2646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.793697662940651e-06,
+      "loss": 0.6523,
+      "step": 2647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.793412671064944e-06,
+      "loss": 0.7391,
+      "step": 2648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.793127490960699e-06,
+      "loss": 0.7418,
+      "step": 2649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7928421226513245e-06,
+      "loss": 0.8146,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.792556566160239e-06,
+      "loss": 0.7053,
+      "step": 2651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.79227082151088e-06,
+      "loss": 0.6773,
+      "step": 2652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7919848887266995e-06,
+      "loss": 0.7222,
+      "step": 2653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.791698767831164e-06,
+      "loss": 0.7554,
+      "step": 2654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.791412458847759e-06,
+      "loss": 0.7024,
+      "step": 2655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.791125961799981e-06,
+      "loss": 0.7199,
+      "step": 2656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.790839276711344e-06,
+      "loss": 0.7085,
+      "step": 2657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.790552403605378e-06,
+      "loss": 0.8311,
+      "step": 2658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7902653425056265e-06,
+      "loss": 0.6929,
+      "step": 2659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.789978093435651e-06,
+      "loss": 0.6553,
+      "step": 2660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.789690656419026e-06,
+      "loss": 0.7742,
+      "step": 2661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7894030314793435e-06,
+      "loss": 0.7739,
+      "step": 2662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.789115218640209e-06,
+      "loss": 0.6519,
+      "step": 2663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7888272179252435e-06,
+      "loss": 1.0032,
+      "step": 2664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7885390293580866e-06,
+      "loss": 0.7209,
+      "step": 2665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.788250652962389e-06,
+      "loss": 0.6966,
+      "step": 2666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.787962088761821e-06,
+      "loss": 0.7381,
+      "step": 2667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.787673336780063e-06,
+      "loss": 0.6471,
+      "step": 2668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.787384397040815e-06,
+      "loss": 0.6493,
+      "step": 2669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.787095269567793e-06,
+      "loss": 0.7337,
+      "step": 2670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.786805954384726e-06,
+      "loss": 0.8649,
+      "step": 2671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.786516451515357e-06,
+      "loss": 1.0332,
+      "step": 2672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.786226760983449e-06,
+      "loss": 0.7638,
+      "step": 2673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.785936882812777e-06,
+      "loss": 0.8809,
+      "step": 2674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7856468170271324e-06,
+      "loss": 0.842,
+      "step": 2675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7853565636503214e-06,
+      "loss": 0.6967,
+      "step": 2676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.785066122706168e-06,
+      "loss": 0.6315,
+      "step": 2677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.784775494218508e-06,
+      "loss": 0.7815,
+      "step": 2678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.784484678211194e-06,
+      "loss": 0.6613,
+      "step": 2679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.784193674708095e-06,
+      "loss": 0.901,
+      "step": 2680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.783902483733096e-06,
+      "loss": 0.878,
+      "step": 2681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.783611105310094e-06,
+      "loss": 0.7725,
+      "step": 2682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.783319539463004e-06,
+      "loss": 0.7781,
+      "step": 2683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.783027786215756e-06,
+      "loss": 0.7391,
+      "step": 2684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.782735845592296e-06,
+      "loss": 0.6559,
+      "step": 2685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7824437176165835e-06,
+      "loss": 0.6519,
+      "step": 2686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.782151402312594e-06,
+      "loss": 0.7932,
+      "step": 2687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7818588997043214e-06,
+      "loss": 0.742,
+      "step": 2688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7815662098157705e-06,
+      "loss": 0.7026,
+      "step": 2689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.781273332670964e-06,
+      "loss": 0.9024,
+      "step": 2690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.780980268293938e-06,
+      "loss": 0.6756,
+      "step": 2691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.780687016708747e-06,
+      "loss": 0.7447,
+      "step": 2692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7803935779394585e-06,
+      "loss": 0.7268,
+      "step": 2693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.780099952010156e-06,
+      "loss": 0.8884,
+      "step": 2694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.779806138944939e-06,
+      "loss": 0.655,
+      "step": 2695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.779512138767921e-06,
+      "loss": 0.7961,
+      "step": 2696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.779217951503232e-06,
+      "loss": 0.8689,
+      "step": 2697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.778923577175018e-06,
+      "loss": 0.6498,
+      "step": 2698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.778629015807437e-06,
+      "loss": 0.8489,
+      "step": 2699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.778334267424667e-06,
+      "loss": 0.6966,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.778039332050898e-06,
+      "loss": 0.6547,
+      "step": 2701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.777744209710335e-06,
+      "loss": 0.7862,
+      "step": 2702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.777448900427203e-06,
+      "loss": 0.7138,
+      "step": 2703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.777153404225737e-06,
+      "loss": 0.7079,
+      "step": 2704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.77685772113019e-06,
+      "loss": 0.683,
+      "step": 2705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.776561851164829e-06,
+      "loss": 0.6298,
+      "step": 2706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.776265794353938e-06,
+      "loss": 0.6584,
+      "step": 2707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.775969550721815e-06,
+      "loss": 0.7242,
+      "step": 2708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.775673120292773e-06,
+      "loss": 0.7155,
+      "step": 2709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.775376503091143e-06,
+      "loss": 0.8134,
+      "step": 2710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.775079699141268e-06,
+      "loss": 0.697,
+      "step": 2711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.774782708467508e-06,
+      "loss": 0.8585,
+      "step": 2712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.774485531094238e-06,
+      "loss": 0.6679,
+      "step": 2713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.774188167045849e-06,
+      "loss": 0.6632,
+      "step": 2714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7738906163467465e-06,
+      "loss": 0.7149,
+      "step": 2715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.773592879021351e-06,
+      "loss": 0.7638,
+      "step": 2716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7732949550940996e-06,
+      "loss": 0.7008,
+      "step": 2717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.772996844589443e-06,
+      "loss": 0.7832,
+      "step": 2718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.772698547531849e-06,
+      "loss": 0.7285,
+      "step": 2719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7724000639458005e-06,
+      "loss": 0.755,
+      "step": 2720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.772101393855793e-06,
+      "loss": 0.7388,
+      "step": 2721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.771802537286342e-06,
+      "loss": 0.6982,
+      "step": 2722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.771503494261973e-06,
+      "loss": 0.6552,
+      "step": 2723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.771204264807232e-06,
+      "loss": 0.6769,
+      "step": 2724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.770904848946677e-06,
+      "loss": 0.6255,
+      "step": 2725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.770605246704882e-06,
+      "loss": 0.7865,
+      "step": 2726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.770305458106435e-06,
+      "loss": 0.8458,
+      "step": 2727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.770005483175944e-06,
+      "loss": 0.7114,
+      "step": 2728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.769705321938025e-06,
+      "loss": 0.7416,
+      "step": 2729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.769404974417316e-06,
+      "loss": 0.801,
+      "step": 2730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.769104440638468e-06,
+      "loss": 0.6744,
+      "step": 2731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.768803720626145e-06,
+      "loss": 0.7745,
+      "step": 2732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7685028144050296e-06,
+      "loss": 0.7732,
+      "step": 2733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.768201721999817e-06,
+      "loss": 0.8238,
+      "step": 2734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.76790044343522e-06,
+      "loss": 0.9224,
+      "step": 2735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.767598978735964e-06,
+      "loss": 0.7265,
+      "step": 2736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.767297327926794e-06,
+      "loss": 0.7624,
+      "step": 2737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.766995491032466e-06,
+      "loss": 0.6152,
+      "step": 2738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.766693468077752e-06,
+      "loss": 0.8678,
+      "step": 2739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7663912590874404e-06,
+      "loss": 0.7579,
+      "step": 2740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.766088864086336e-06,
+      "loss": 0.6475,
+      "step": 2741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.765786283099257e-06,
+      "loss": 0.6928,
+      "step": 2742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.765483516151036e-06,
+      "loss": 0.8223,
+      "step": 2743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.765180563266524e-06,
+      "loss": 0.7845,
+      "step": 2744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.764877424470583e-06,
+      "loss": 0.5801,
+      "step": 2745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.764574099788095e-06,
+      "loss": 0.9925,
+      "step": 2746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.764270589243953e-06,
+      "loss": 0.7054,
+      "step": 2747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.763966892863069e-06,
+      "loss": 0.8203,
+      "step": 2748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.763663010670368e-06,
+      "loss": 0.7189,
+      "step": 2749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.76335894269079e-06,
+      "loss": 0.7761,
+      "step": 2750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.763054688949291e-06,
+      "loss": 0.7266,
+      "step": 2751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.762750249470843e-06,
+      "loss": 0.7875,
+      "step": 2752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.762445624280432e-06,
+      "loss": 0.8784,
+      "step": 2753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.762140813403059e-06,
+      "loss": 0.6998,
+      "step": 2754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7618358168637426e-06,
+      "loss": 0.6672,
+      "step": 2755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.761530634687512e-06,
+      "loss": 0.7067,
+      "step": 2756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.761225266899418e-06,
+      "loss": 0.7519,
+      "step": 2757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.760919713524521e-06,
+      "loss": 0.7165,
+      "step": 2758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7606139745879e-06,
+      "loss": 0.6975,
+      "step": 2759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.760308050114647e-06,
+      "loss": 0.8228,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.760001940129871e-06,
+      "loss": 0.8515,
+      "step": 2761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.759695644658696e-06,
+      "loss": 0.872,
+      "step": 2762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.75938916372626e-06,
+      "loss": 0.8146,
+      "step": 2763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.759082497357717e-06,
+      "loss": 0.723,
+      "step": 2764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.758775645578236e-06,
+      "loss": 0.6841,
+      "step": 2765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.758468608413003e-06,
+      "loss": 0.728,
+      "step": 2766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.758161385887215e-06,
+      "loss": 0.7508,
+      "step": 2767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.757853978026089e-06,
+      "loss": 0.6609,
+      "step": 2768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.757546384854854e-06,
+      "loss": 0.7063,
+      "step": 2769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.757238606398756e-06,
+      "loss": 0.6374,
+      "step": 2770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.756930642683055e-06,
+      "loss": 0.6698,
+      "step": 2771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.756622493733026e-06,
+      "loss": 0.8354,
+      "step": 2772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.756314159573961e-06,
+      "loss": 0.8782,
+      "step": 2773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.756005640231166e-06,
+      "loss": 0.7831,
+      "step": 2774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7556969357299625e-06,
+      "loss": 0.7139,
+      "step": 2775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.755388046095686e-06,
+      "loss": 0.721,
+      "step": 2776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.755078971353688e-06,
+      "loss": 0.8087,
+      "step": 2777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.754769711529337e-06,
+      "loss": 0.7494,
+      "step": 2778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.754460266648014e-06,
+      "loss": 0.6803,
+      "step": 2779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7541506367351156e-06,
+      "loss": 0.735,
+      "step": 2780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.753840821816055e-06,
+      "loss": 0.6812,
+      "step": 2781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.75353082191626e-06,
+      "loss": 0.7502,
+      "step": 2782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7532206370611734e-06,
+      "loss": 0.6165,
+      "step": 2783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.752910267276253e-06,
+      "loss": 0.7949,
+      "step": 2784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.752599712586971e-06,
+      "loss": 0.7606,
+      "step": 2785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.752288973018817e-06,
+      "loss": 0.8276,
+      "step": 2786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.751978048597294e-06,
+      "loss": 0.7684,
+      "step": 2787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.751666939347921e-06,
+      "loss": 0.7337,
+      "step": 2788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.751355645296232e-06,
+      "loss": 0.6527,
+      "step": 2789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.751044166467775e-06,
+      "loss": 0.7869,
+      "step": 2790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7507325028881145e-06,
+      "loss": 0.8462,
+      "step": 2791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.750420654582831e-06,
+      "loss": 0.8505,
+      "step": 2792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7501086215775175e-06,
+      "loss": 0.7923,
+      "step": 2793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.749796403897784e-06,
+      "loss": 0.8218,
+      "step": 2794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.749484001569257e-06,
+      "loss": 0.7132,
+      "step": 2795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7491714146175735e-06,
+      "loss": 0.7857,
+      "step": 2796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7488586430683915e-06,
+      "loss": 0.6759,
+      "step": 2797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.748545686947379e-06,
+      "loss": 0.6324,
+      "step": 2798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.748232546280223e-06,
+      "loss": 0.7235,
+      "step": 2799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.747919221092623e-06,
+      "loss": 0.7212,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.747605711410295e-06,
+      "loss": 0.9026,
+      "step": 2801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.747292017258971e-06,
+      "loss": 0.7076,
+      "step": 2802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.746978138664395e-06,
+      "loss": 0.937,
+      "step": 2803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.74666407565233e-06,
+      "loss": 0.8519,
+      "step": 2804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.746349828248551e-06,
+      "loss": 0.7338,
+      "step": 2805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.74603539647885e-06,
+      "loss": 0.683,
+      "step": 2806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.745720780369033e-06,
+      "loss": 0.7636,
+      "step": 2807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.745405979944922e-06,
+      "loss": 0.6631,
+      "step": 2808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.745090995232354e-06,
+      "loss": 0.7018,
+      "step": 2809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7447758262571804e-06,
+      "loss": 0.774,
+      "step": 2810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.744460473045269e-06,
+      "loss": 0.797,
+      "step": 2811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.744144935622502e-06,
+      "loss": 0.7654,
+      "step": 2812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.743829214014775e-06,
+      "loss": 0.6872,
+      "step": 2813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.743513308248003e-06,
+      "loss": 0.7386,
+      "step": 2814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7431972183481115e-06,
+      "loss": 0.8027,
+      "step": 2815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.742880944341044e-06,
+      "loss": 0.7747,
+      "step": 2816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7425644862527574e-06,
+      "loss": 0.7085,
+      "step": 2817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.742247844109225e-06,
+      "loss": 0.8794,
+      "step": 2818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.741931017936435e-06,
+      "loss": 0.8248,
+      "step": 2819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.741614007760391e-06,
+      "loss": 0.9262,
+      "step": 2820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.74129681360711e-06,
+      "loss": 0.7454,
+      "step": 2821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.740979435502626e-06,
+      "loss": 0.8795,
+      "step": 2822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.740661873472987e-06,
+      "loss": 0.6715,
+      "step": 2823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.740344127544257e-06,
+      "loss": 0.8149,
+      "step": 2824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.740026197742512e-06,
+      "loss": 0.7115,
+      "step": 2825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.739708084093849e-06,
+      "loss": 0.7353,
+      "step": 2826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.739389786624376e-06,
+      "loss": 0.7691,
+      "step": 2827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.739071305360216e-06,
+      "loss": 0.7635,
+      "step": 2828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.738752640327507e-06,
+      "loss": 0.8011,
+      "step": 2829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.738433791552406e-06,
+      "loss": 0.6807,
+      "step": 2830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.738114759061079e-06,
+      "loss": 0.8368,
+      "step": 2831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7377955428797105e-06,
+      "loss": 0.6369,
+      "step": 2832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.737476143034501e-06,
+      "loss": 0.7126,
+      "step": 2833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.737156559551665e-06,
+      "loss": 0.7173,
+      "step": 2834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7368367924574306e-06,
+      "loss": 0.8975,
+      "step": 2835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.736516841778043e-06,
+      "loss": 0.8765,
+      "step": 2836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.73619670753976e-06,
+      "loss": 0.8009,
+      "step": 2837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.735876389768859e-06,
+      "loss": 0.7437,
+      "step": 2838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.735555888491628e-06,
+      "loss": 0.75,
+      "step": 2839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7352352037343715e-06,
+      "loss": 0.662,
+      "step": 2840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.73491433552341e-06,
+      "loss": 0.7696,
+      "step": 2841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.734593283885078e-06,
+      "loss": 0.7523,
+      "step": 2842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.734272048845725e-06,
+      "loss": 0.7252,
+      "step": 2843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.733950630431716e-06,
+      "loss": 0.7591,
+      "step": 2844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.733629028669431e-06,
+      "loss": 0.795,
+      "step": 2845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.733307243585265e-06,
+      "loss": 0.6837,
+      "step": 2846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.732985275205629e-06,
+      "loss": 0.6191,
+      "step": 2847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.732663123556947e-06,
+      "loss": 0.6492,
+      "step": 2848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.732340788665659e-06,
+      "loss": 0.7419,
+      "step": 2849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.73201827055822e-06,
+      "loss": 0.9334,
+      "step": 2850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7316955692611024e-06,
+      "loss": 0.6759,
+      "step": 2851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.731372684800788e-06,
+      "loss": 0.717,
+      "step": 2852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.73104961720378e-06,
+      "loss": 0.8756,
+      "step": 2853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.730726366496592e-06,
+      "loss": 0.6025,
+      "step": 2854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.730402932705755e-06,
+      "loss": 0.8481,
+      "step": 2855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.730079315857815e-06,
+      "loss": 0.6949,
+      "step": 2856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7297555159793316e-06,
+      "loss": 0.7679,
+      "step": 2857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.72943153309688e-06,
+      "loss": 0.6868,
+      "step": 2858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.72910736723705e-06,
+      "loss": 0.8427,
+      "step": 2859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.728783018426449e-06,
+      "loss": 0.7459,
+      "step": 2860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.728458486691696e-06,
+      "loss": 0.6853,
+      "step": 2861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.728133772059428e-06,
+      "loss": 0.6784,
+      "step": 2862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.727808874556293e-06,
+      "loss": 0.6855,
+      "step": 2863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.727483794208958e-06,
+      "loss": 0.7105,
+      "step": 2864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.727158531044104e-06,
+      "loss": 0.7104,
+      "step": 2865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.726833085088425e-06,
+      "loss": 0.6698,
+      "step": 2866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.726507456368633e-06,
+      "loss": 0.828,
+      "step": 2867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.726181644911453e-06,
+      "loss": 0.7702,
+      "step": 2868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.725855650743625e-06,
+      "loss": 0.6408,
+      "step": 2869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.725529473891904e-06,
+      "loss": 0.6763,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.725203114383063e-06,
+      "loss": 0.8113,
+      "step": 2871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.724876572243885e-06,
+      "loss": 0.7656,
+      "step": 2872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.724549847501172e-06,
+      "loss": 0.7533,
+      "step": 2873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.724222940181739e-06,
+      "loss": 0.6472,
+      "step": 2874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.723895850312415e-06,
+      "loss": 0.7341,
+      "step": 2875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.723568577920047e-06,
+      "loss": 0.8518,
+      "step": 2876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.723241123031495e-06,
+      "loss": 0.8247,
+      "step": 2877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.722913485673636e-06,
+      "loss": 0.8022,
+      "step": 2878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.722585665873357e-06,
+      "loss": 0.7578,
+      "step": 2879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.722257663657567e-06,
+      "loss": 0.7866,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.721929479053183e-06,
+      "loss": 0.7112,
+      "step": 2881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7216011120871415e-06,
+      "loss": 0.7333,
+      "step": 2882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.721272562786394e-06,
+      "loss": 0.6982,
+      "step": 2883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.720943831177905e-06,
+      "loss": 0.7725,
+      "step": 2884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7206149172886535e-06,
+      "loss": 0.7475,
+      "step": 2885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.720285821145636e-06,
+      "loss": 0.735,
+      "step": 2886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.719956542775862e-06,
+      "loss": 0.7516,
+      "step": 2887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.719627082206357e-06,
+      "loss": 0.742,
+      "step": 2888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.71929743946416e-06,
+      "loss": 0.7129,
+      "step": 2889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.718967614576328e-06,
+      "loss": 0.7024,
+      "step": 2890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7186376075699276e-06,
+      "loss": 0.7414,
+      "step": 2891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.718307418472047e-06,
+      "loss": 0.6413,
+      "step": 2892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7179770473097835e-06,
+      "loss": 0.7688,
+      "step": 2893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.717646494110253e-06,
+      "loss": 0.7639,
+      "step": 2894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.717315758900586e-06,
+      "loss": 0.7478,
+      "step": 2895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.716984841707926e-06,
+      "loss": 0.7928,
+      "step": 2896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7166537425594324e-06,
+      "loss": 0.7742,
+      "step": 2897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.71632246148228e-06,
+      "loss": 0.8292,
+      "step": 2898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.715990998503658e-06,
+      "loss": 0.6675,
+      "step": 2899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.715659353650772e-06,
+      "loss": 0.774,
+      "step": 2900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.715327526950839e-06,
+      "loss": 0.8193,
+      "step": 2901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7149955184310955e-06,
+      "loss": 0.7712,
+      "step": 2902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7146633281187895e-06,
+      "loss": 0.7991,
+      "step": 2903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.714330956041184e-06,
+      "loss": 0.7884,
+      "step": 2904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.713998402225561e-06,
+      "loss": 0.7733,
+      "step": 2905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7136656666992105e-06,
+      "loss": 0.6826,
+      "step": 2906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.713332749489444e-06,
+      "loss": 0.6981,
+      "step": 2907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.712999650623585e-06,
+      "loss": 0.9366,
+      "step": 2908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7126663701289705e-06,
+      "loss": 0.7118,
+      "step": 2909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7123329080329565e-06,
+      "loss": 0.7473,
+      "step": 2910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7119992643629085e-06,
+      "loss": 0.7555,
+      "step": 2911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.711665439146212e-06,
+      "loss": 0.7394,
+      "step": 2912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.711331432410264e-06,
+      "loss": 0.7283,
+      "step": 2913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.710997244182479e-06,
+      "loss": 0.8329,
+      "step": 2914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.710662874490284e-06,
+      "loss": 0.7141,
+      "step": 2915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.710328323361122e-06,
+      "loss": 0.7912,
+      "step": 2916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.70999359082245e-06,
+      "loss": 0.7469,
+      "step": 2917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.709658676901742e-06,
+      "loss": 0.7757,
+      "step": 2918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7093235816264874e-06,
+      "loss": 0.6897,
+      "step": 2919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.708988305024183e-06,
+      "loss": 0.6918,
+      "step": 2920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.708652847122352e-06,
+      "loss": 0.818,
+      "step": 2921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.708317207948523e-06,
+      "loss": 0.6169,
+      "step": 2922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.707981387530245e-06,
+      "loss": 0.773,
+      "step": 2923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.707645385895079e-06,
+      "loss": 0.7897,
+      "step": 2924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.707309203070602e-06,
+      "loss": 0.7094,
+      "step": 2925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.706972839084406e-06,
+      "loss": 0.6684,
+      "step": 2926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.706636293964097e-06,
+      "loss": 0.8788,
+      "step": 2927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.706299567737297e-06,
+      "loss": 0.7149,
+      "step": 2928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.705962660431642e-06,
+      "loss": 0.7338,
+      "step": 2929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.705625572074784e-06,
+      "loss": 0.7694,
+      "step": 2930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.705288302694389e-06,
+      "loss": 0.7562,
+      "step": 2931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.704950852318135e-06,
+      "loss": 0.7622,
+      "step": 2932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.704613220973722e-06,
+      "loss": 0.9276,
+      "step": 2933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.704275408688859e-06,
+      "loss": 0.7057,
+      "step": 2934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.70393741549127e-06,
+      "loss": 0.6753,
+      "step": 2935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.703599241408696e-06,
+      "loss": 0.721,
+      "step": 2936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.703260886468894e-06,
+      "loss": 0.7118,
+      "step": 2937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.702922350699632e-06,
+      "loss": 0.7482,
+      "step": 2938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.702583634128695e-06,
+      "loss": 0.743,
+      "step": 2939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.702244736783883e-06,
+      "loss": 0.7312,
+      "step": 2940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.701905658693011e-06,
+      "loss": 0.6494,
+      "step": 2941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.701566399883908e-06,
+      "loss": 0.8683,
+      "step": 2942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.701226960384419e-06,
+      "loss": 0.699,
+      "step": 2943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.700887340222401e-06,
+      "loss": 0.7707,
+      "step": 2944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.70054753942573e-06,
+      "loss": 0.6923,
+      "step": 2945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.700207558022291e-06,
+      "loss": 0.7608,
+      "step": 2946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.699867396039992e-06,
+      "loss": 0.9246,
+      "step": 2947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.69952705350675e-06,
+      "loss": 0.9277,
+      "step": 2948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.699186530450497e-06,
+      "loss": 0.9097,
+      "step": 2949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.698845826899181e-06,
+      "loss": 0.8645,
+      "step": 2950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.698504942880766e-06,
+      "loss": 0.7578,
+      "step": 2951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.698163878423228e-06,
+      "loss": 0.6641,
+      "step": 2952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6978226335545605e-06,
+      "loss": 0.664,
+      "step": 2953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.69748120830277e-06,
+      "loss": 0.736,
+      "step": 2954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.69713960269588e-06,
+      "loss": 0.7193,
+      "step": 2955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.696797816761925e-06,
+      "loss": 0.7984,
+      "step": 2956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.696455850528958e-06,
+      "loss": 0.9167,
+      "step": 2957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.696113704025045e-06,
+      "loss": 0.7019,
+      "step": 2958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.695771377278268e-06,
+      "loss": 0.7455,
+      "step": 2959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.695428870316721e-06,
+      "loss": 0.6984,
+      "step": 2960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.695086183168517e-06,
+      "loss": 0.7114,
+      "step": 2961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.69474331586178e-06,
+      "loss": 0.7881,
+      "step": 2962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.694400268424651e-06,
+      "loss": 0.6954,
+      "step": 2963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.694057040885285e-06,
+      "loss": 0.8181,
+      "step": 2964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.693713633271852e-06,
+      "loss": 0.7039,
+      "step": 2965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6933700456125364e-06,
+      "loss": 0.761,
+      "step": 2966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.693026277935539e-06,
+      "loss": 0.7933,
+      "step": 2967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.692682330269072e-06,
+      "loss": 0.6941,
+      "step": 2968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.692338202641364e-06,
+      "loss": 0.6128,
+      "step": 2969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.691993895080662e-06,
+      "loss": 0.8535,
+      "step": 2970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.691649407615222e-06,
+      "loss": 0.8178,
+      "step": 2971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.691304740273318e-06,
+      "loss": 0.8346,
+      "step": 2972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6909598930832385e-06,
+      "loss": 0.8171,
+      "step": 2973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.690614866073285e-06,
+      "loss": 0.7539,
+      "step": 2974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.690269659271776e-06,
+      "loss": 0.8028,
+      "step": 2975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.689924272707045e-06,
+      "loss": 0.7895,
+      "step": 2976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.689578706407437e-06,
+      "loss": 0.7374,
+      "step": 2977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.689232960401315e-06,
+      "loss": 0.6977,
+      "step": 2978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.688887034717056e-06,
+      "loss": 0.7,
+      "step": 2979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.68854092938305e-06,
+      "loss": 0.7866,
+      "step": 2980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.688194644427704e-06,
+      "loss": 0.7038,
+      "step": 2981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6878481798794385e-06,
+      "loss": 0.7105,
+      "step": 2982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.687501535766688e-06,
+      "loss": 0.7944,
+      "step": 2983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.687154712117906e-06,
+      "loss": 0.6743,
+      "step": 2984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6868077089615545e-06,
+      "loss": 0.8014,
+      "step": 2985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.686460526326115e-06,
+      "loss": 0.7455,
+      "step": 2986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.686113164240081e-06,
+      "loss": 0.7929,
+      "step": 2987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.685765622731963e-06,
+      "loss": 0.7426,
+      "step": 2988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.685417901830283e-06,
+      "loss": 0.6979,
+      "step": 2989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.685070001563582e-06,
+      "loss": 0.7604,
+      "step": 2990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.684721921960411e-06,
+      "loss": 0.7179,
+      "step": 2991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.68437366304934e-06,
+      "loss": 0.7648,
+      "step": 2992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.684025224858951e-06,
+      "loss": 0.7545,
+      "step": 2993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.683676607417842e-06,
+      "loss": 0.7919,
+      "step": 2994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6833278107546255e-06,
+      "loss": 0.9468,
+      "step": 2995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.682978834897927e-06,
+      "loss": 0.7752,
+      "step": 2996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.68262967987639e-06,
+      "loss": 0.7242,
+      "step": 2997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6822803457186696e-06,
+      "loss": 0.8483,
+      "step": 2998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.681930832453437e-06,
+      "loss": 0.873,
+      "step": 2999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.681581140109379e-06,
+      "loss": 0.6599,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.681231268715195e-06,
+      "loss": 0.7866,
+      "step": 3001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.680881218299601e-06,
+      "loss": 0.7255,
+      "step": 3002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.680530988891327e-06,
+      "loss": 0.763,
+      "step": 3003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.680180580519117e-06,
+      "loss": 0.6991,
+      "step": 3004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.679829993211731e-06,
+      "loss": 0.8507,
+      "step": 3005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.679479226997941e-06,
+      "loss": 0.743,
+      "step": 3006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.679128281906537e-06,
+      "loss": 0.6982,
+      "step": 3007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.678777157966323e-06,
+      "loss": 0.8638,
+      "step": 3008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.678425855206116e-06,
+      "loss": 0.8612,
+      "step": 3009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.67807437365475e-06,
+      "loss": 0.7236,
+      "step": 3010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6777227133410695e-06,
+      "loss": 0.7259,
+      "step": 3011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.677370874293939e-06,
+      "loss": 0.8354,
+      "step": 3012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.677018856542235e-06,
+      "loss": 0.7659,
+      "step": 3013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.676666660114848e-06,
+      "loss": 0.7983,
+      "step": 3014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.676314285040684e-06,
+      "loss": 0.6427,
+      "step": 3015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.675961731348665e-06,
+      "loss": 0.814,
+      "step": 3016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.675608999067724e-06,
+      "loss": 0.7694,
+      "step": 3017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6752560882268124e-06,
+      "loss": 0.6593,
+      "step": 3018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.674902998854897e-06,
+      "loss": 0.9585,
+      "step": 3019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.674549730980953e-06,
+      "loss": 0.6826,
+      "step": 3020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.674196284633976e-06,
+      "loss": 0.7445,
+      "step": 3021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6738426598429755e-06,
+      "loss": 0.6829,
+      "step": 3022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.673488856636974e-06,
+      "loss": 0.7828,
+      "step": 3023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.67313487504501e-06,
+      "loss": 0.8615,
+      "step": 3024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.672780715096135e-06,
+      "loss": 0.7305,
+      "step": 3025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.672426376819417e-06,
+      "loss": 0.6768,
+      "step": 3026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.672071860243937e-06,
+      "loss": 0.7024,
+      "step": 3027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6717171653987924e-06,
+      "loss": 0.7498,
+      "step": 3028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.671362292313093e-06,
+      "loss": 0.8467,
+      "step": 3029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.671007241015966e-06,
+      "loss": 0.7411,
+      "step": 3030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.67065201153655e-06,
+      "loss": 0.7367,
+      "step": 3031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.670296603904002e-06,
+      "loss": 0.8471,
+      "step": 3032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.669941018147492e-06,
+      "loss": 0.7381,
+      "step": 3033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6695852542962e-06,
+      "loss": 0.7372,
+      "step": 3034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.669229312379329e-06,
+      "loss": 0.6524,
+      "step": 3035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.66887319242609e-06,
+      "loss": 0.8086,
+      "step": 3036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.668516894465712e-06,
+      "loss": 0.6981,
+      "step": 3037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.668160418527439e-06,
+      "loss": 0.6496,
+      "step": 3038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.667803764640526e-06,
+      "loss": 0.734,
+      "step": 3039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.667446932834245e-06,
+      "loss": 0.7106,
+      "step": 3040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.667089923137884e-06,
+      "loss": 0.7783,
+      "step": 3041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.666732735580743e-06,
+      "loss": 0.712,
+      "step": 3042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.666375370192138e-06,
+      "loss": 0.8681,
+      "step": 3043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.666017827001399e-06,
+      "loss": 0.819,
+      "step": 3044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.665660106037872e-06,
+      "loss": 0.6739,
+      "step": 3045
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.8394511938095093,
+      "eval_runtime": 436.9183,
+      "eval_samples_per_second": 6.864,
+      "eval_steps_per_second": 0.858,
+      "step": 3045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.665302207330914e-06,
+      "loss": 0.7001,
+      "step": 3046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6649441309099e-06,
+      "loss": 0.6503,
+      "step": 3047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.66458587680422e-06,
+      "loss": 0.6742,
+      "step": 3048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.664227445043276e-06,
+      "loss": 0.7077,
+      "step": 3049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.663868835656486e-06,
+      "loss": 0.6973,
+      "step": 3050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.663510048673282e-06,
+      "loss": 0.7119,
+      "step": 3051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.663151084123112e-06,
+      "loss": 0.6916,
+      "step": 3052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.662791942035437e-06,
+      "loss": 0.6187,
+      "step": 3053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.662432622439732e-06,
+      "loss": 0.6696,
+      "step": 3054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.662073125365488e-06,
+      "loss": 0.8191,
+      "step": 3055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.661713450842212e-06,
+      "loss": 0.7124,
+      "step": 3056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6613535988994215e-06,
+      "loss": 0.839,
+      "step": 3057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.660993569566652e-06,
+      "loss": 0.6693,
+      "step": 3058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.660633362873452e-06,
+      "loss": 0.8211,
+      "step": 3059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.660272978849386e-06,
+      "loss": 0.8612,
+      "step": 3060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.65991241752403e-06,
+      "loss": 0.7562,
+      "step": 3061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.659551678926978e-06,
+      "loss": 0.8082,
+      "step": 3062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.659190763087837e-06,
+      "loss": 0.6598,
+      "step": 3063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.658829670036228e-06,
+      "loss": 0.6986,
+      "step": 3064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.658468399801788e-06,
+      "loss": 0.709,
+      "step": 3065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.658106952414167e-06,
+      "loss": 0.6941,
+      "step": 3066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.65774532790303e-06,
+      "loss": 0.7313,
+      "step": 3067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6573835262980584e-06,
+      "loss": 0.7072,
+      "step": 3068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.657021547628946e-06,
+      "loss": 0.715,
+      "step": 3069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6566593919254e-06,
+      "loss": 0.7731,
+      "step": 3070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.656297059217146e-06,
+      "loss": 0.695,
+      "step": 3071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.65593454953392e-06,
+      "loss": 0.8461,
+      "step": 3072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.655571862905475e-06,
+      "loss": 0.6873,
+      "step": 3073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.65520899936158e-06,
+      "loss": 0.6986,
+      "step": 3074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.654845958932014e-06,
+      "loss": 0.685,
+      "step": 3075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.654482741646574e-06,
+      "loss": 0.8261,
+      "step": 3076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.654119347535071e-06,
+      "loss": 0.8286,
+      "step": 3077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.653755776627329e-06,
+      "loss": 0.7168,
+      "step": 3078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.653392028953188e-06,
+      "loss": 0.6896,
+      "step": 3079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.653028104542502e-06,
+      "loss": 0.6423,
+      "step": 3080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6526640034251406e-06,
+      "loss": 0.6369,
+      "step": 3081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6522997256309846e-06,
+      "loss": 0.7074,
+      "step": 3082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6519352711899345e-06,
+      "loss": 0.8062,
+      "step": 3083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.651570640131899e-06,
+      "loss": 0.9353,
+      "step": 3084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.651205832486808e-06,
+      "loss": 0.7202,
+      "step": 3085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6508408482846e-06,
+      "loss": 0.6294,
+      "step": 3086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.650475687555233e-06,
+      "loss": 0.7469,
+      "step": 3087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6501103503286744e-06,
+      "loss": 0.6944,
+      "step": 3088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.649744836634911e-06,
+      "loss": 0.6868,
+      "step": 3089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.649379146503939e-06,
+      "loss": 0.8029,
+      "step": 3090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.649013279965775e-06,
+      "loss": 0.7154,
+      "step": 3091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.648647237050446e-06,
+      "loss": 0.6761,
+      "step": 3092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.648281017787993e-06,
+      "loss": 0.8761,
+      "step": 3093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.647914622208475e-06,
+      "loss": 0.7448,
+      "step": 3094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.647548050341962e-06,
+      "loss": 0.646,
+      "step": 3095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.64718130221854e-06,
+      "loss": 0.7893,
+      "step": 3096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.64681437786831e-06,
+      "loss": 0.6894,
+      "step": 3097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.646447277321386e-06,
+      "loss": 0.7434,
+      "step": 3098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6460800006078975e-06,
+      "loss": 0.7842,
+      "step": 3099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.645712547757989e-06,
+      "loss": 0.7462,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.645344918801817e-06,
+      "loss": 0.8397,
+      "step": 3101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6449771137695556e-06,
+      "loss": 0.8192,
+      "step": 3102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6446091326913915e-06,
+      "loss": 0.8935,
+      "step": 3103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.644240975597526e-06,
+      "loss": 0.7799,
+      "step": 3104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.643872642518175e-06,
+      "loss": 0.8127,
+      "step": 3105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6435041334835696e-06,
+      "loss": 0.6735,
+      "step": 3106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.643135448523953e-06,
+      "loss": 0.7034,
+      "step": 3107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.642766587669586e-06,
+      "loss": 0.7271,
+      "step": 3108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.642397550950742e-06,
+      "loss": 0.7074,
+      "step": 3109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.64202833839771e-06,
+      "loss": 0.893,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.64165895004079e-06,
+      "loss": 0.7474,
+      "step": 3111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.641289385910302e-06,
+      "loss": 0.6945,
+      "step": 3112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.640919646036576e-06,
+      "loss": 0.8126,
+      "step": 3113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.640549730449957e-06,
+      "loss": 0.6233,
+      "step": 3114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.640179639180807e-06,
+      "loss": 0.8071,
+      "step": 3115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.639809372259499e-06,
+      "loss": 0.7491,
+      "step": 3116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.639438929716425e-06,
+      "loss": 0.7939,
+      "step": 3117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.639068311581985e-06,
+      "loss": 0.8518,
+      "step": 3118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.638697517886599e-06,
+      "loss": 0.7097,
+      "step": 3119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.638326548660699e-06,
+      "loss": 0.7845,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.637955403934732e-06,
+      "loss": 0.7809,
+      "step": 3121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6375840837391585e-06,
+      "loss": 0.7542,
+      "step": 3122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.637212588104454e-06,
+      "loss": 0.6169,
+      "step": 3123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6368409170611095e-06,
+      "loss": 0.7042,
+      "step": 3124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.636469070639629e-06,
+      "loss": 0.6182,
+      "step": 3125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.63609704887053e-06,
+      "loss": 0.6926,
+      "step": 3126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.635724851784348e-06,
+      "loss": 0.7839,
+      "step": 3127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.635352479411628e-06,
+      "loss": 0.7408,
+      "step": 3128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.634979931782934e-06,
+      "loss": 0.7076,
+      "step": 3129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6346072089288415e-06,
+      "loss": 0.8569,
+      "step": 3130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.63423431087994e-06,
+      "loss": 0.7017,
+      "step": 3131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6338612376668365e-06,
+      "loss": 0.8717,
+      "step": 3132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.63348798932015e-06,
+      "loss": 0.7339,
+      "step": 3133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.633114565870514e-06,
+      "loss": 0.6783,
+      "step": 3134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.632740967348577e-06,
+      "loss": 0.7764,
+      "step": 3135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.632367193785e-06,
+      "loss": 0.9343,
+      "step": 3136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.631993245210461e-06,
+      "loss": 0.802,
+      "step": 3137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.631619121655653e-06,
+      "loss": 0.77,
+      "step": 3138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.631244823151279e-06,
+      "loss": 0.8214,
+      "step": 3139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.63087034972806e-06,
+      "loss": 0.6301,
+      "step": 3140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.630495701416731e-06,
+      "loss": 0.686,
+      "step": 3141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.63012087824804e-06,
+      "loss": 0.7245,
+      "step": 3142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.629745880252751e-06,
+      "loss": 0.7726,
+      "step": 3143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6293707074616396e-06,
+      "loss": 0.8414,
+      "step": 3144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.628995359905499e-06,
+      "loss": 0.7072,
+      "step": 3145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.628619837615135e-06,
+      "loss": 0.7724,
+      "step": 3146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.628244140621368e-06,
+      "loss": 0.7032,
+      "step": 3147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.627868268955033e-06,
+      "loss": 0.7625,
+      "step": 3148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.627492222646979e-06,
+      "loss": 0.7647,
+      "step": 3149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.627116001728068e-06,
+      "loss": 0.8112,
+      "step": 3150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.62673960622918e-06,
+      "loss": 0.7276,
+      "step": 3151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.626363036181207e-06,
+      "loss": 0.7589,
+      "step": 3152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.625986291615054e-06,
+      "loss": 0.8548,
+      "step": 3153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.625609372561644e-06,
+      "loss": 0.791,
+      "step": 3154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.625232279051908e-06,
+      "loss": 0.7419,
+      "step": 3155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6248550111168e-06,
+      "loss": 0.671,
+      "step": 3156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.624477568787281e-06,
+      "loss": 0.6495,
+      "step": 3157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.62409995209433e-06,
+      "loss": 0.9279,
+      "step": 3158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.623722161068939e-06,
+      "loss": 0.6472,
+      "step": 3159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.623344195742114e-06,
+      "loss": 0.7548,
+      "step": 3160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6229660561448775e-06,
+      "loss": 0.7506,
+      "step": 3161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6225877423082635e-06,
+      "loss": 0.7037,
+      "step": 3162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6222092542633225e-06,
+      "loss": 0.7519,
+      "step": 3163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.621830592041118e-06,
+      "loss": 0.5942,
+      "step": 3164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.621451755672727e-06,
+      "loss": 0.6246,
+      "step": 3165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.621072745189244e-06,
+      "loss": 0.7938,
+      "step": 3166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.620693560621775e-06,
+      "loss": 0.6829,
+      "step": 3167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.62031420200144e-06,
+      "loss": 0.7173,
+      "step": 3168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.619934669359375e-06,
+      "loss": 0.7931,
+      "step": 3169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.61955496272673e-06,
+      "loss": 0.6792,
+      "step": 3170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.619175082134668e-06,
+      "loss": 0.811,
+      "step": 3171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.618795027614369e-06,
+      "loss": 0.7005,
+      "step": 3172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.618414799197023e-06,
+      "loss": 0.6723,
+      "step": 3173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.618034396913838e-06,
+      "loss": 0.7623,
+      "step": 3174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.617653820796035e-06,
+      "loss": 0.6931,
+      "step": 3175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.617273070874849e-06,
+      "loss": 0.7271,
+      "step": 3176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6168921471815295e-06,
+      "loss": 0.6967,
+      "step": 3177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.61651104974734e-06,
+      "loss": 0.6699,
+      "step": 3178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.616129778603559e-06,
+      "loss": 0.7359,
+      "step": 3179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.615748333781479e-06,
+      "loss": 0.7691,
+      "step": 3180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.615366715312406e-06,
+      "loss": 0.7606,
+      "step": 3181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.61498492322766e-06,
+      "loss": 0.7464,
+      "step": 3182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.614602957558578e-06,
+      "loss": 0.6909,
+      "step": 3183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.614220818336508e-06,
+      "loss": 0.6415,
+      "step": 3184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.613838505592813e-06,
+      "loss": 0.6621,
+      "step": 3185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.613456019358872e-06,
+      "loss": 0.8088,
+      "step": 3186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6130733596660774e-06,
+      "loss": 0.7309,
+      "step": 3187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.612690526545834e-06,
+      "loss": 0.7841,
+      "step": 3188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.612307520029562e-06,
+      "loss": 0.6838,
+      "step": 3189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.611924340148697e-06,
+      "loss": 0.65,
+      "step": 3190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.611540986934689e-06,
+      "loss": 0.7158,
+      "step": 3191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.611157460418999e-06,
+      "loss": 0.6471,
+      "step": 3192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.610773760633106e-06,
+      "loss": 0.8616,
+      "step": 3193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.610389887608501e-06,
+      "loss": 0.6555,
+      "step": 3194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.610005841376689e-06,
+      "loss": 0.7924,
+      "step": 3195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.609621621969192e-06,
+      "loss": 0.7456,
+      "step": 3196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.609237229417543e-06,
+      "loss": 0.7865,
+      "step": 3197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.60885266375329e-06,
+      "loss": 0.8234,
+      "step": 3198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.608467925007998e-06,
+      "loss": 0.9047,
+      "step": 3199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.608083013213241e-06,
+      "loss": 0.646,
+      "step": 3200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.607697928400611e-06,
+      "loss": 0.7591,
+      "step": 3201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.607312670601714e-06,
+      "loss": 0.6996,
+      "step": 3202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.606927239848168e-06,
+      "loss": 0.8234,
+      "step": 3203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6065416361716086e-06,
+      "loss": 0.8229,
+      "step": 3204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.606155859603682e-06,
+      "loss": 0.7306,
+      "step": 3205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.605769910176052e-06,
+      "loss": 0.782,
+      "step": 3206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.605383787920393e-06,
+      "loss": 0.6691,
+      "step": 3207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6049974928683966e-06,
+      "loss": 0.7565,
+      "step": 3208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.604611025051768e-06,
+      "loss": 0.774,
+      "step": 3209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.604224384502223e-06,
+      "loss": 0.8376,
+      "step": 3210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.603837571251498e-06,
+      "loss": 0.7213,
+      "step": 3211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.603450585331339e-06,
+      "loss": 0.6772,
+      "step": 3212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.603063426773506e-06,
+      "loss": 0.7308,
+      "step": 3213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.602676095609777e-06,
+      "loss": 0.7077,
+      "step": 3214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.60228859187194e-06,
+      "loss": 0.8243,
+      "step": 3215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.601900915591797e-06,
+      "loss": 0.759,
+      "step": 3216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.60151306680117e-06,
+      "loss": 0.7979,
+      "step": 3217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.601125045531889e-06,
+      "loss": 0.7071,
+      "step": 3218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.600736851815799e-06,
+      "loss": 0.7387,
+      "step": 3219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.600348485684764e-06,
+      "loss": 0.7911,
+      "step": 3220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.599959947170655e-06,
+      "loss": 0.625,
+      "step": 3221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.599571236305362e-06,
+      "loss": 0.7369,
+      "step": 3222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.599182353120789e-06,
+      "loss": 0.7323,
+      "step": 3223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5987932976488514e-06,
+      "loss": 0.716,
+      "step": 3224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.598404069921481e-06,
+      "loss": 0.771,
+      "step": 3225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.598014669970624e-06,
+      "loss": 0.6588,
+      "step": 3226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.597625097828239e-06,
+      "loss": 0.6921,
+      "step": 3227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5972353535262984e-06,
+      "loss": 0.6955,
+      "step": 3228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5968454370967915e-06,
+      "loss": 0.754,
+      "step": 3229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.59645534857172e-06,
+      "loss": 0.7458,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5960650879830995e-06,
+      "loss": 0.7704,
+      "step": 3231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.59567465536296e-06,
+      "loss": 0.6928,
+      "step": 3232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.595284050743346e-06,
+      "loss": 0.7001,
+      "step": 3233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5948932741563155e-06,
+      "loss": 0.792,
+      "step": 3234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.59450232563394e-06,
+      "loss": 0.6716,
+      "step": 3235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.59411120520831e-06,
+      "loss": 0.6979,
+      "step": 3236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.593719912911521e-06,
+      "loss": 0.6493,
+      "step": 3237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.59332844877569e-06,
+      "loss": 0.6586,
+      "step": 3238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.592936812832946e-06,
+      "loss": 0.7564,
+      "step": 3239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.592545005115432e-06,
+      "loss": 0.782,
+      "step": 3240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5921530256553045e-06,
+      "loss": 0.6864,
+      "step": 3241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.591760874484736e-06,
+      "loss": 1.0071,
+      "step": 3242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5913685516359106e-06,
+      "loss": 0.6267,
+      "step": 3243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.590976057141026e-06,
+      "loss": 0.755,
+      "step": 3244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.590583391032299e-06,
+      "loss": 0.7455,
+      "step": 3245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.590190553341955e-06,
+      "loss": 0.7632,
+      "step": 3246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.589797544102237e-06,
+      "loss": 0.8986,
+      "step": 3247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.589404363345399e-06,
+      "loss": 0.6558,
+      "step": 3248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.589011011103711e-06,
+      "loss": 0.6623,
+      "step": 3249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.588617487409458e-06,
+      "loss": 0.7496,
+      "step": 3250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.588223792294937e-06,
+      "loss": 0.6557,
+      "step": 3251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.587829925792461e-06,
+      "loss": 0.6833,
+      "step": 3252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.587435887934354e-06,
+      "loss": 0.7256,
+      "step": 3253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.587041678752957e-06,
+      "loss": 0.8359,
+      "step": 3254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.586647298280625e-06,
+      "loss": 0.6974,
+      "step": 3255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.586252746549726e-06,
+      "loss": 0.6734,
+      "step": 3256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.585858023592642e-06,
+      "loss": 0.6983,
+      "step": 3257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.585463129441769e-06,
+      "loss": 0.6457,
+      "step": 3258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.585068064129518e-06,
+      "loss": 0.8662,
+      "step": 3259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.584672827688312e-06,
+      "loss": 0.7392,
+      "step": 3260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.584277420150591e-06,
+      "loss": 0.6608,
+      "step": 3261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.583881841548807e-06,
+      "loss": 0.7648,
+      "step": 3262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.583486091915426e-06,
+      "loss": 0.6968,
+      "step": 3263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.583090171282929e-06,
+      "loss": 0.8223,
+      "step": 3264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.582694079683812e-06,
+      "loss": 0.5903,
+      "step": 3265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5822978171505815e-06,
+      "loss": 0.7027,
+      "step": 3266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.58190138371576e-06,
+      "loss": 0.797,
+      "step": 3267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.581504779411887e-06,
+      "loss": 0.6876,
+      "step": 3268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.58110800427151e-06,
+      "loss": 0.7217,
+      "step": 3269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.580711058327196e-06,
+      "loss": 0.7501,
+      "step": 3270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5803139416115236e-06,
+      "loss": 0.6154,
+      "step": 3271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5799166541570836e-06,
+      "loss": 0.7792,
+      "step": 3272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5795191959964845e-06,
+      "loss": 0.695,
+      "step": 3273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.579121567162348e-06,
+      "loss": 0.7836,
+      "step": 3274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5787237676873065e-06,
+      "loss": 0.6859,
+      "step": 3275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.57832579760401e-06,
+      "loss": 0.7127,
+      "step": 3276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.577927656945121e-06,
+      "loss": 0.6386,
+      "step": 3277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.577529345743317e-06,
+      "loss": 0.7957,
+      "step": 3278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.577130864031289e-06,
+      "loss": 0.7041,
+      "step": 3279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5767322118417405e-06,
+      "loss": 0.6904,
+      "step": 3280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.576333389207391e-06,
+      "loss": 0.7636,
+      "step": 3281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.575934396160974e-06,
+      "loss": 0.5889,
+      "step": 3282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.575535232735235e-06,
+      "loss": 0.6956,
+      "step": 3283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.575135898962936e-06,
+      "loss": 0.8346,
+      "step": 3284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.574736394876851e-06,
+      "loss": 0.7817,
+      "step": 3285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.574336720509769e-06,
+      "loss": 0.8803,
+      "step": 3286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.573936875894492e-06,
+      "loss": 0.6622,
+      "step": 3287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.573536861063838e-06,
+      "loss": 0.9255,
+      "step": 3288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5731366760506356e-06,
+      "loss": 0.7247,
+      "step": 3289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.572736320887731e-06,
+      "loss": 0.7226,
+      "step": 3290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.572335795607983e-06,
+      "loss": 0.7909,
+      "step": 3291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.571935100244263e-06,
+      "loss": 0.6643,
+      "step": 3292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.571534234829459e-06,
+      "loss": 0.7082,
+      "step": 3293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.57113319939647e-06,
+      "loss": 0.7556,
+      "step": 3294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.57073199397821e-06,
+      "loss": 0.834,
+      "step": 3295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.57033061860761e-06,
+      "loss": 0.7407,
+      "step": 3296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.569929073317609e-06,
+      "loss": 0.7919,
+      "step": 3297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.569527358141166e-06,
+      "loss": 0.7049,
+      "step": 3298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.56912547311125e-06,
+      "loss": 0.7745,
+      "step": 3299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.568723418260844e-06,
+      "loss": 0.5894,
+      "step": 3300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.568321193622948e-06,
+      "loss": 0.6986,
+      "step": 3301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.567918799230573e-06,
+      "loss": 0.6908,
+      "step": 3302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.567516235116745e-06,
+      "loss": 0.7716,
+      "step": 3303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5671135013145055e-06,
+      "loss": 0.6471,
+      "step": 3304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.566710597856906e-06,
+      "loss": 0.8003,
+      "step": 3305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.566307524777015e-06,
+      "loss": 0.6783,
+      "step": 3306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5659042821079135e-06,
+      "loss": 0.8753,
+      "step": 3307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.565500869882698e-06,
+      "loss": 0.7044,
+      "step": 3308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.565097288134479e-06,
+      "loss": 0.6963,
+      "step": 3309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.564693536896378e-06,
+      "loss": 0.9323,
+      "step": 3310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.564289616201532e-06,
+      "loss": 0.8504,
+      "step": 3311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563885526083094e-06,
+      "loss": 0.6491,
+      "step": 3312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563481266574228e-06,
+      "loss": 0.7063,
+      "step": 3313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563076837708114e-06,
+      "loss": 0.7246,
+      "step": 3314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.562672239517943e-06,
+      "loss": 0.6941,
+      "step": 3315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.562267472036923e-06,
+      "loss": 0.7684,
+      "step": 3316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.561862535298276e-06,
+      "loss": 0.7665,
+      "step": 3317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.561457429335234e-06,
+      "loss": 0.6475,
+      "step": 3318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5610521541810475e-06,
+      "loss": 0.8125,
+      "step": 3319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560646709868978e-06,
+      "loss": 0.7465,
+      "step": 3320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560241096432302e-06,
+      "loss": 0.7007,
+      "step": 3321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55983531390431e-06,
+      "loss": 0.6483,
+      "step": 3322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559429362318304e-06,
+      "loss": 0.7549,
+      "step": 3323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559023241707605e-06,
+      "loss": 0.7712,
+      "step": 3324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558616952105543e-06,
+      "loss": 0.7411,
+      "step": 3325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5582104935454635e-06,
+      "loss": 0.69,
+      "step": 3326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557803866060726e-06,
+      "loss": 0.7248,
+      "step": 3327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557397069684705e-06,
+      "loss": 0.6548,
+      "step": 3328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556990104450787e-06,
+      "loss": 0.763,
+      "step": 3329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556582970392372e-06,
+      "loss": 0.8144,
+      "step": 3330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556175667542877e-06,
+      "loss": 0.7513,
+      "step": 3331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5557681959357304e-06,
+      "loss": 0.7387,
+      "step": 3332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555360555604373e-06,
+      "loss": 0.6816,
+      "step": 3333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554952746582263e-06,
+      "loss": 0.6897,
+      "step": 3334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55454476890287e-06,
+      "loss": 0.746,
+      "step": 3335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554136622599678e-06,
+      "loss": 0.7995,
+      "step": 3336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553728307706187e-06,
+      "loss": 0.6688,
+      "step": 3337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553319824255906e-06,
+      "loss": 0.6827,
+      "step": 3338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552911172282361e-06,
+      "loss": 0.6788,
+      "step": 3339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552502351819095e-06,
+      "loss": 0.6932,
+      "step": 3340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552093362899658e-06,
+      "loss": 0.7129,
+      "step": 3341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551684205557617e-06,
+      "loss": 0.7325,
+      "step": 3342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551274879826555e-06,
+      "loss": 0.7109,
+      "step": 3343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550865385740065e-06,
+      "loss": 0.7326,
+      "step": 3344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550455723331758e-06,
+      "loss": 0.7504,
+      "step": 3345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550045892635254e-06,
+      "loss": 0.8379,
+      "step": 3346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.54963589368419e-06,
+      "loss": 0.6786,
+      "step": 3347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549225726512216e-06,
+      "loss": 0.8606,
+      "step": 3348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.548815391152998e-06,
+      "loss": 0.7701,
+      "step": 3349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5484048876402095e-06,
+      "loss": 0.7568,
+      "step": 3350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547994216007546e-06,
+      "loss": 0.631,
+      "step": 3351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5475833762887105e-06,
+      "loss": 0.7335,
+      "step": 3352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.547172368517423e-06,
+      "loss": 0.7903,
+      "step": 3353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546761192727415e-06,
+      "loss": 0.6971,
+      "step": 3354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546349848952435e-06,
+      "loss": 0.7936,
+      "step": 3355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.545938337226243e-06,
+      "loss": 0.7588,
+      "step": 3356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5455266575826125e-06,
+      "loss": 0.723,
+      "step": 3357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.545114810055332e-06,
+      "loss": 0.8147,
+      "step": 3358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544702794678203e-06,
+      "loss": 0.635,
+      "step": 3359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544290611485041e-06,
+      "loss": 0.7702,
+      "step": 3360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543878260509677e-06,
+      "loss": 0.733,
+      "step": 3361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543465741785951e-06,
+      "loss": 0.8044,
+      "step": 3362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5430530553477225e-06,
+      "loss": 0.8515,
+      "step": 3363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542640201228861e-06,
+      "loss": 0.7395,
+      "step": 3364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542227179463251e-06,
+      "loss": 0.6111,
+      "step": 3365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541813990084791e-06,
+      "loss": 0.88,
+      "step": 3366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541400633127393e-06,
+      "loss": 0.6919,
+      "step": 3367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540987108624982e-06,
+      "loss": 0.6578,
+      "step": 3368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540573416611498e-06,
+      "loss": 0.8258,
+      "step": 3369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540159557120894e-06,
+      "loss": 0.6975,
+      "step": 3370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539745530187137e-06,
+      "loss": 0.688,
+      "step": 3371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539331335844208e-06,
+      "loss": 0.6493,
+      "step": 3372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538916974126101e-06,
+      "loss": 0.7147,
+      "step": 3373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538502445066823e-06,
+      "loss": 0.7289,
+      "step": 3374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538087748700399e-06,
+      "loss": 0.738,
+      "step": 3375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5376728850608606e-06,
+      "loss": 0.6526,
+      "step": 3376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53725785418226e-06,
+      "loss": 0.6483,
+      "step": 3377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536842656098659e-06,
+      "loss": 0.6664,
+      "step": 3378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536427290844135e-06,
+      "loss": 0.6531,
+      "step": 3379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536011758452779e-06,
+      "loss": 0.6702,
+      "step": 3380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535596058958694e-06,
+      "loss": 0.7591,
+      "step": 3381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5351801923959984e-06,
+      "loss": 0.8587,
+      "step": 3382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534764158798825e-06,
+      "loss": 0.8184,
+      "step": 3383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5343479582013165e-06,
+      "loss": 0.5994,
+      "step": 3384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533931590637633e-06,
+      "loss": 0.6622,
+      "step": 3385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5335150561419495e-06,
+      "loss": 0.961,
+      "step": 3386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533098354748451e-06,
+      "loss": 0.6807,
+      "step": 3387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532681486491336e-06,
+      "loss": 0.669,
+      "step": 3388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5322644514048205e-06,
+      "loss": 0.6063,
+      "step": 3389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531847249523131e-06,
+      "loss": 0.8993,
+      "step": 3390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5314298808805106e-06,
+      "loss": 0.7971,
+      "step": 3391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531012345511211e-06,
+      "loss": 0.6397,
+      "step": 3392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530594643449503e-06,
+      "loss": 0.7886,
+      "step": 3393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530176774729669e-06,
+      "loss": 0.6779,
+      "step": 3394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529758739386003e-06,
+      "loss": 0.7693,
+      "step": 3395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529340537452818e-06,
+      "loss": 0.6957,
+      "step": 3396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528922168964433e-06,
+      "loss": 0.7859,
+      "step": 3397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528503633955189e-06,
+      "loss": 0.6315,
+      "step": 3398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528084932459435e-06,
+      "loss": 0.713,
+      "step": 3399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5276660645115355e-06,
+      "loss": 0.6326,
+      "step": 3400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.527247030145867e-06,
+      "loss": 0.673,
+      "step": 3401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5268278293968236e-06,
+      "loss": 0.947,
+      "step": 3402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5264084622988095e-06,
+      "loss": 0.7722,
+      "step": 3403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.525988928886244e-06,
+      "loss": 0.6822,
+      "step": 3404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5255692291935585e-06,
+      "loss": 0.7155,
+      "step": 3405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5251493632552e-06,
+      "loss": 0.7324,
+      "step": 3406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.524729331105628e-06,
+      "loss": 0.6223,
+      "step": 3407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5243091327793174e-06,
+      "loss": 0.7755,
+      "step": 3408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.523888768310754e-06,
+      "loss": 0.6447,
+      "step": 3409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.523468237734439e-06,
+      "loss": 0.8099,
+      "step": 3410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.523047541084887e-06,
+      "loss": 0.6944,
+      "step": 3411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.522626678396627e-06,
+      "loss": 0.759,
+      "step": 3412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.522205649704199e-06,
+      "loss": 0.8009,
+      "step": 3413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.521784455042158e-06,
+      "loss": 0.6528,
+      "step": 3414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.521363094445075e-06,
+      "loss": 0.6955,
+      "step": 3415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.520941567947531e-06,
+      "loss": 0.6799,
+      "step": 3416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.520519875584124e-06,
+      "loss": 0.8759,
+      "step": 3417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.520098017389461e-06,
+      "loss": 0.7678,
+      "step": 3418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.519675993398167e-06,
+      "loss": 0.703,
+      "step": 3419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.519253803644879e-06,
+      "loss": 0.7449,
+      "step": 3420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.518831448164248e-06,
+      "loss": 0.73,
+      "step": 3421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.518408926990937e-06,
+      "loss": 0.6782,
+      "step": 3422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.517986240159625e-06,
+      "loss": 0.6621,
+      "step": 3423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.517563387705003e-06,
+      "loss": 0.8002,
+      "step": 3424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5171403696617755e-06,
+      "loss": 0.7708,
+      "step": 3425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.516717186064662e-06,
+      "loss": 0.6802,
+      "step": 3426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.516293836948394e-06,
+      "loss": 0.6595,
+      "step": 3427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.515870322347718e-06,
+      "loss": 0.7506,
+      "step": 3428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.515446642297393e-06,
+      "loss": 0.7758,
+      "step": 3429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5150227968321905e-06,
+      "loss": 0.8654,
+      "step": 3430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.514598785986899e-06,
+      "loss": 0.5876,
+      "step": 3431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.514174609796318e-06,
+      "loss": 0.8315,
+      "step": 3432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.513750268295261e-06,
+      "loss": 0.6709,
+      "step": 3433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.513325761518555e-06,
+      "loss": 0.8488,
+      "step": 3434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.512901089501041e-06,
+      "loss": 0.6053,
+      "step": 3435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.512476252277573e-06,
+      "loss": 0.7195,
+      "step": 3436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.51205124988302e-06,
+      "loss": 0.7236,
+      "step": 3437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.511626082352263e-06,
+      "loss": 0.8924,
+      "step": 3438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.511200749720196e-06,
+      "loss": 0.8398,
+      "step": 3439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.510775252021729e-06,
+      "loss": 0.771,
+      "step": 3440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5103495892917826e-06,
+      "loss": 0.6891,
+      "step": 3441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.509923761565294e-06,
+      "loss": 0.9017,
+      "step": 3442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.50949776887721e-06,
+      "loss": 0.6653,
+      "step": 3443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5090716112624966e-06,
+      "loss": 0.6598,
+      "step": 3444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.508645288756128e-06,
+      "loss": 0.711,
+      "step": 3445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.508218801393095e-06,
+      "loss": 0.5969,
+      "step": 3446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.507792149208399e-06,
+      "loss": 0.69,
+      "step": 3447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.507365332237059e-06,
+      "loss": 0.8006,
+      "step": 3448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.506938350514105e-06,
+      "loss": 0.709,
+      "step": 3449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.506511204074581e-06,
+      "loss": 0.6694,
+      "step": 3450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.506083892953543e-06,
+      "loss": 0.9032,
+      "step": 3451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.505656417186063e-06,
+      "loss": 0.727,
+      "step": 3452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.505228776807226e-06,
+      "loss": 0.7787,
+      "step": 3453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5048009718521286e-06,
+      "loss": 0.6803,
+      "step": 3454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.504373002355883e-06,
+      "loss": 0.6692,
+      "step": 3455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.503944868353615e-06,
+      "loss": 0.6482,
+      "step": 3456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5035165698804615e-06,
+      "loss": 0.7374,
+      "step": 3457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5030881069715756e-06,
+      "loss": 0.7493,
+      "step": 3458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.502659479662123e-06,
+      "loss": 0.8215,
+      "step": 3459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.502230687987281e-06,
+      "loss": 0.7118,
+      "step": 3460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.501801731982244e-06,
+      "loss": 0.7718,
+      "step": 3461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.501372611682217e-06,
+      "loss": 0.7939,
+      "step": 3462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.50094332712242e-06,
+      "loss": 0.8271,
+      "step": 3463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.500513878338084e-06,
+      "loss": 0.756,
+      "step": 3464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.500084265364458e-06,
+      "loss": 0.6239,
+      "step": 3465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.499654488236801e-06,
+      "loss": 0.7876,
+      "step": 3466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.499224546990385e-06,
+      "loss": 0.7986,
+      "step": 3467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.498794441660498e-06,
+      "loss": 0.7868,
+      "step": 3468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.4983641722824415e-06,
+      "loss": 0.6752,
+      "step": 3469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.497933738891527e-06,
+      "loss": 0.668,
+      "step": 3470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.497503141523082e-06,
+      "loss": 0.7325,
+      "step": 3471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.497072380212448e-06,
+      "loss": 0.6958,
+      "step": 3472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.496641454994979e-06,
+      "loss": 0.841,
+      "step": 3473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.496210365906042e-06,
+      "loss": 0.8242,
+      "step": 3474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.495779112981019e-06,
+      "loss": 0.7574,
+      "step": 3475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4953476962553035e-06,
+      "loss": 0.6227,
+      "step": 3476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.494916115764304e-06,
+      "loss": 0.7597,
+      "step": 3477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.494484371543442e-06,
+      "loss": 0.7835,
+      "step": 3478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.494052463628151e-06,
+      "loss": 0.7164,
+      "step": 3479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4936203920538805e-06,
+      "loss": 0.7322,
+      "step": 3480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.493188156856092e-06,
+      "loss": 0.7221,
+      "step": 3481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.49275575807026e-06,
+      "loss": 0.7624,
+      "step": 3482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.492323195731874e-06,
+      "loss": 0.8745,
+      "step": 3483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.491890469876435e-06,
+      "loss": 0.924,
+      "step": 3484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.491457580539459e-06,
+      "loss": 0.654,
+      "step": 3485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.491024527756474e-06,
+      "loss": 0.7153,
+      "step": 3486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.490591311563022e-06,
+      "loss": 0.6519,
+      "step": 3487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4901579319946595e-06,
+      "loss": 0.6732,
+      "step": 3488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.489724389086956e-06,
+      "loss": 0.6245,
+      "step": 3489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.489290682875493e-06,
+      "loss": 0.6851,
+      "step": 3490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.488856813395866e-06,
+      "loss": 0.7216,
+      "step": 3491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4884227806836845e-06,
+      "loss": 0.73,
+      "step": 3492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.487988584774572e-06,
+      "loss": 0.7899,
+      "step": 3493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4875542257041625e-06,
+      "loss": 0.7079,
+      "step": 3494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.487119703508107e-06,
+      "loss": 0.6379,
+      "step": 3495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4866850182220686e-06,
+      "loss": 0.7809,
+      "step": 3496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.486250169881723e-06,
+      "loss": 0.7702,
+      "step": 3497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.485815158522759e-06,
+      "loss": 0.8109,
+      "step": 3498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.48537998418088e-06,
+      "loss": 0.8098,
+      "step": 3499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4849446468918024e-06,
+      "loss": 0.7632,
+      "step": 3500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.484509146691256e-06,
+      "loss": 0.6261,
+      "step": 3501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.484073483614984e-06,
+      "loss": 0.7805,
+      "step": 3502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4836376576987415e-06,
+      "loss": 0.7631,
+      "step": 3503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4832016689783e-06,
+      "loss": 0.7319,
+      "step": 3504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.482765517489442e-06,
+      "loss": 0.7245,
+      "step": 3505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4823292032679635e-06,
+      "loss": 0.7094,
+      "step": 3506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.481892726349675e-06,
+      "loss": 0.7034,
+      "step": 3507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4814560867704005e-06,
+      "loss": 0.686,
+      "step": 3508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.481019284565975e-06,
+      "loss": 0.7468,
+      "step": 3509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.480582319772248e-06,
+      "loss": 0.826,
+      "step": 3510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.480145192425085e-06,
+      "loss": 0.7247,
+      "step": 3511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.47970790256036e-06,
+      "loss": 0.9102,
+      "step": 3512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.479270450213966e-06,
+      "loss": 0.8284,
+      "step": 3513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.478832835421804e-06,
+      "loss": 0.6613,
+      "step": 3514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.478395058219791e-06,
+      "loss": 0.8195,
+      "step": 3515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.477957118643857e-06,
+      "loss": 0.7933,
+      "step": 3516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.477519016729945e-06,
+      "loss": 0.7489,
+      "step": 3517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.477080752514012e-06,
+      "loss": 0.6446,
+      "step": 3518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.476642326032029e-06,
+      "loss": 0.6708,
+      "step": 3519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.476203737319978e-06,
+      "loss": 0.6357,
+      "step": 3520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.475764986413854e-06,
+      "loss": 0.8184,
+      "step": 3521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.47532607334967e-06,
+      "loss": 0.6853,
+      "step": 3522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.474886998163447e-06,
+      "loss": 0.6964,
+      "step": 3523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.474447760891222e-06,
+      "loss": 0.7532,
+      "step": 3524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.474008361569044e-06,
+      "loss": 0.7801,
+      "step": 3525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.473568800232978e-06,
+      "loss": 0.7799,
+      "step": 3526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.473129076919099e-06,
+      "loss": 0.7874,
+      "step": 3527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.472689191663498e-06,
+      "loss": 0.8235,
+      "step": 3528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.472249144502275e-06,
+      "loss": 0.8297,
+      "step": 3529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.471808935471549e-06,
+      "loss": 0.6409,
+      "step": 3530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.471368564607449e-06,
+      "loss": 0.639,
+      "step": 3531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.470928031946117e-06,
+      "loss": 0.6375,
+      "step": 3532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.470487337523709e-06,
+      "loss": 0.6785,
+      "step": 3533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.470046481376395e-06,
+      "loss": 0.7902,
+      "step": 3534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.469605463540358e-06,
+      "loss": 0.6728,
+      "step": 3535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4691642840517935e-06,
+      "loss": 0.876,
+      "step": 3536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4687229429469105e-06,
+      "loss": 0.6995,
+      "step": 3537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.468281440261931e-06,
+      "loss": 0.9229,
+      "step": 3538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.467839776033092e-06,
+      "loss": 0.685,
+      "step": 3539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.467397950296641e-06,
+      "loss": 0.7747,
+      "step": 3540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.466955963088841e-06,
+      "loss": 0.7464,
+      "step": 3541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.466513814445968e-06,
+      "loss": 0.7106,
+      "step": 3542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4660715044043094e-06,
+      "loss": 0.7756,
+      "step": 3543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4656290330001675e-06,
+      "loss": 0.7554,
+      "step": 3544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.465186400269859e-06,
+      "loss": 0.815,
+      "step": 3545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.464743606249709e-06,
+      "loss": 0.6603,
+      "step": 3546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.464300650976063e-06,
+      "loss": 0.816,
+      "step": 3547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4638575344852746e-06,
+      "loss": 0.7576,
+      "step": 3548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.463414256813711e-06,
+      "loss": 0.6662,
+      "step": 3549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.462970817997754e-06,
+      "loss": 0.7267,
+      "step": 3550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.462527218073799e-06,
+      "loss": 0.8038,
+      "step": 3551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.462083457078252e-06,
+      "loss": 0.7899,
+      "step": 3552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.461639535047536e-06,
+      "loss": 0.7479,
+      "step": 3553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.461195452018085e-06,
+      "loss": 0.7434,
+      "step": 3554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.460751208026345e-06,
+      "loss": 0.685,
+      "step": 3555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.460306803108778e-06,
+      "loss": 0.7921,
+      "step": 3556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4598622373018575e-06,
+      "loss": 0.9008,
+      "step": 3557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.459417510642071e-06,
+      "loss": 0.7779,
+      "step": 3558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.458972623165918e-06,
+      "loss": 0.7886,
+      "step": 3559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4585275749099125e-06,
+      "loss": 0.6139,
+      "step": 3560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.458082365910581e-06,
+      "loss": 0.7561,
+      "step": 3561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.457636996204464e-06,
+      "loss": 0.6872,
+      "step": 3562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.457191465828114e-06,
+      "loss": 0.7316,
+      "step": 3563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.456745774818098e-06,
+      "loss": 0.7466,
+      "step": 3564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.456299923210994e-06,
+      "loss": 0.6791,
+      "step": 3565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.455853911043395e-06,
+      "loss": 0.7265,
+      "step": 3566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.455407738351908e-06,
+      "loss": 0.7114,
+      "step": 3567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.454961405173151e-06,
+      "loss": 0.711,
+      "step": 3568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.454514911543756e-06,
+      "loss": 0.7633,
+      "step": 3569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.454068257500369e-06,
+      "loss": 0.6747,
+      "step": 3570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.453621443079648e-06,
+      "loss": 0.7072,
+      "step": 3571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.453174468318265e-06,
+      "loss": 0.7447,
+      "step": 3572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.452727333252905e-06,
+      "loss": 0.7074,
+      "step": 3573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.452280037920266e-06,
+      "loss": 0.6395,
+      "step": 3574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4518325823570575e-06,
+      "loss": 0.785,
+      "step": 3575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.451384966600005e-06,
+      "loss": 0.8329,
+      "step": 3576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4509371906858464e-06,
+      "loss": 0.7583,
+      "step": 3577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.450489254651332e-06,
+      "loss": 0.6246,
+      "step": 3578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.450041158533226e-06,
+      "loss": 0.7235,
+      "step": 3579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.449592902368303e-06,
+      "loss": 0.6861,
+      "step": 3580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.449144486193355e-06,
+      "loss": 0.646,
+      "step": 3581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.448695910045185e-06,
+      "loss": 0.77,
+      "step": 3582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.448247173960609e-06,
+      "loss": 0.6455,
+      "step": 3583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.447798277976456e-06,
+      "loss": 0.7458,
+      "step": 3584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.447349222129568e-06,
+      "loss": 0.8942,
+      "step": 3585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.446900006456802e-06,
+      "loss": 0.6862,
+      "step": 3586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.446450630995025e-06,
+      "loss": 0.8725,
+      "step": 3587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.446001095781122e-06,
+      "loss": 0.7975,
+      "step": 3588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.445551400851984e-06,
+      "loss": 0.6819,
+      "step": 3589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.445101546244521e-06,
+      "loss": 0.7027,
+      "step": 3590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4446515319956555e-06,
+      "loss": 0.777,
+      "step": 3591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.444201358142319e-06,
+      "loss": 0.7486,
+      "step": 3592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.443751024721461e-06,
+      "loss": 0.8345,
+      "step": 3593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.443300531770041e-06,
+      "loss": 0.699,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.442849879325032e-06,
+      "loss": 0.7365,
+      "step": 3595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.442399067423423e-06,
+      "loss": 0.7512,
+      "step": 3596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.44194809610221e-06,
+      "loss": 0.6627,
+      "step": 3597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.44149696539841e-06,
+      "loss": 0.6791,
+      "step": 3598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4410456753490455e-06,
+      "loss": 0.7436,
+      "step": 3599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.440594225991157e-06,
+      "loss": 0.778,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4401426173617965e-06,
+      "loss": 0.7383,
+      "step": 3601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.439690849498029e-06,
+      "loss": 0.7617,
+      "step": 3602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.439238922436934e-06,
+      "loss": 0.7176,
+      "step": 3603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.438786836215599e-06,
+      "loss": 0.7267,
+      "step": 3604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4383345908711326e-06,
+      "loss": 0.6602,
+      "step": 3605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.437882186440649e-06,
+      "loss": 0.6836,
+      "step": 3606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.437429622961283e-06,
+      "loss": 0.7166,
+      "step": 3607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.436976900470172e-06,
+      "loss": 0.704,
+      "step": 3608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.436524019004476e-06,
+      "loss": 0.7997,
+      "step": 3609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.436070978601366e-06,
+      "loss": 0.682,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4356177792980215e-06,
+      "loss": 0.7503,
+      "step": 3611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.43516442113164e-06,
+      "loss": 0.74,
+      "step": 3612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.43471090413943e-06,
+      "loss": 0.7937,
+      "step": 3613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.434257228358612e-06,
+      "loss": 0.7771,
+      "step": 3614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.433803393826423e-06,
+      "loss": 0.7833,
+      "step": 3615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.433349400580109e-06,
+      "loss": 0.7865,
+      "step": 3616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.432895248656932e-06,
+      "loss": 0.6629,
+      "step": 3617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.432440938094165e-06,
+      "loss": 0.6883,
+      "step": 3618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.431986468929095e-06,
+      "loss": 0.7233,
+      "step": 3619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.431531841199023e-06,
+      "loss": 0.6968,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.431077054941259e-06,
+      "loss": 0.7496,
+      "step": 3621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.430622110193133e-06,
+      "loss": 0.728,
+      "step": 3622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.430167006991982e-06,
+      "loss": 0.6555,
+      "step": 3623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.429711745375157e-06,
+      "loss": 0.7821,
+      "step": 3624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.429256325380024e-06,
+      "loss": 0.8562,
+      "step": 3625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.42880074704396e-06,
+      "loss": 0.715,
+      "step": 3626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.428345010404358e-06,
+      "loss": 0.679,
+      "step": 3627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4278891154986195e-06,
+      "loss": 0.8007,
+      "step": 3628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.427433062364162e-06,
+      "loss": 0.6311,
+      "step": 3629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.426976851038418e-06,
+      "loss": 0.7203,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4265204815588266e-06,
+      "loss": 0.7364,
+      "step": 3631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.426063953962846e-06,
+      "loss": 0.7741,
+      "step": 3632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.425607268287945e-06,
+      "loss": 0.8292,
+      "step": 3633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.425150424571603e-06,
+      "loss": 0.8136,
+      "step": 3634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.424693422851318e-06,
+      "loss": 0.9422,
+      "step": 3635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.424236263164596e-06,
+      "loss": 0.7098,
+      "step": 3636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.423778945548959e-06,
+      "loss": 0.6853,
+      "step": 3637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.423321470041938e-06,
+      "loss": 0.6858,
+      "step": 3638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.422863836681082e-06,
+      "loss": 0.718,
+      "step": 3639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.422406045503952e-06,
+      "loss": 0.7985,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.421948096548117e-06,
+      "loss": 0.6847,
+      "step": 3641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.421489989851163e-06,
+      "loss": 0.7145,
+      "step": 3642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4210317254506904e-06,
+      "loss": 0.6828,
+      "step": 3643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.42057330338431e-06,
+      "loss": 0.8048,
+      "step": 3644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.420114723689646e-06,
+      "loss": 0.6891,
+      "step": 3645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.419655986404335e-06,
+      "loss": 0.7267,
+      "step": 3646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.419197091566028e-06,
+      "loss": 0.7436,
+      "step": 3647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.418738039212388e-06,
+      "loss": 0.6983,
+      "step": 3648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.41827882938109e-06,
+      "loss": 0.8127,
+      "step": 3649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4178194621098245e-06,
+      "loss": 0.7996,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.417359937436293e-06,
+      "loss": 0.8571,
+      "step": 3651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.41690025539821e-06,
+      "loss": 0.6623,
+      "step": 3652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.416440416033303e-06,
+      "loss": 0.7187,
+      "step": 3653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.415980419379313e-06,
+      "loss": 0.7604,
+      "step": 3654
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.8331807851791382,
+      "eval_runtime": 437.3553,
+      "eval_samples_per_second": 6.857,
+      "eval_steps_per_second": 0.857,
+      "step": 3654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.415520265473993e-06,
+      "loss": 0.7829,
+      "step": 3655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4150599543551115e-06,
+      "loss": 0.706,
+      "step": 3656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.414599486060446e-06,
+      "loss": 0.8432,
+      "step": 3657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.414138860627788e-06,
+      "loss": 0.7274,
+      "step": 3658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.413678078094945e-06,
+      "loss": 0.7566,
+      "step": 3659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.413217138499734e-06,
+      "loss": 0.6742,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4127560418799855e-06,
+      "loss": 0.6656,
+      "step": 3661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.412294788273544e-06,
+      "loss": 0.6498,
+      "step": 3662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4118333777182646e-06,
+      "loss": 0.6956,
+      "step": 3663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.41137181025202e-06,
+      "loss": 0.6799,
+      "step": 3664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4109100859126895e-06,
+      "loss": 0.7705,
+      "step": 3665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.41044820473817e-06,
+      "loss": 0.7667,
+      "step": 3666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.409986166766371e-06,
+      "loss": 0.6482,
+      "step": 3667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.409523972035211e-06,
+      "loss": 0.8693,
+      "step": 3668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.409061620582626e-06,
+      "loss": 0.6838,
+      "step": 3669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.408599112446561e-06,
+      "loss": 0.7199,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.408136447664977e-06,
+      "loss": 0.7297,
+      "step": 3671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4076736262758464e-06,
+      "loss": 0.8441,
+      "step": 3672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.407210648317154e-06,
+      "loss": 0.7105,
+      "step": 3673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4067475138268985e-06,
+      "loss": 0.631,
+      "step": 3674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.406284222843091e-06,
+      "loss": 0.6654,
+      "step": 3675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.405820775403756e-06,
+      "loss": 0.6297,
+      "step": 3676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.40535717154693e-06,
+      "loss": 0.68,
+      "step": 3677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.404893411310661e-06,
+      "loss": 0.7242,
+      "step": 3678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.404429494733014e-06,
+      "loss": 0.7449,
+      "step": 3679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.403965421852062e-06,
+      "loss": 0.6112,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.403501192705895e-06,
+      "loss": 0.809,
+      "step": 3681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.403036807332612e-06,
+      "loss": 0.8534,
+      "step": 3682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4025722657703295e-06,
+      "loss": 0.7772,
+      "step": 3683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4021075680571714e-06,
+      "loss": 0.7757,
+      "step": 3684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.401642714231278e-06,
+      "loss": 0.6809,
+      "step": 3685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.401177704330803e-06,
+      "loss": 0.6523,
+      "step": 3686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.400712538393908e-06,
+      "loss": 0.6582,
+      "step": 3687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.400247216458774e-06,
+      "loss": 0.6868,
+      "step": 3688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.39978173856359e-06,
+      "loss": 0.7399,
+      "step": 3689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.399316104746561e-06,
+      "loss": 0.761,
+      "step": 3690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.3988503150459005e-06,
+      "loss": 0.6961,
+      "step": 3691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.3983843694998405e-06,
+      "loss": 0.7911,
+      "step": 3692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.397918268146621e-06,
+      "loss": 0.6064,
+      "step": 3693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.3974520110244965e-06,
+      "loss": 0.64,
+      "step": 3694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.396985598171736e-06,
+      "loss": 0.6736,
+      "step": 3695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.396519029626618e-06,
+      "loss": 0.7531,
+      "step": 3696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.396052305427436e-06,
+      "loss": 0.8879,
+      "step": 3697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.395585425612496e-06,
+      "loss": 0.6307,
+      "step": 3698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.395118390220116e-06,
+      "loss": 0.7367,
+      "step": 3699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.394651199288628e-06,
+      "loss": 0.8053,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.394183852856375e-06,
+      "loss": 0.7684,
+      "step": 3701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.3937163509617155e-06,
+      "loss": 0.7275,
+      "step": 3702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.393248693643017e-06,
+      "loss": 0.661,
+      "step": 3703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.392780880938662e-06,
+      "loss": 0.6846,
+      "step": 3704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.3923129128870465e-06,
+      "loss": 0.6666,
+      "step": 3705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.391844789526579e-06,
+      "loss": 0.7609,
+      "step": 3706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.391376510895679e-06,
+      "loss": 0.7245,
+      "step": 3707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.390908077032779e-06,
+      "loss": 0.6148,
+      "step": 3708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.390439487976326e-06,
+      "loss": 0.7941,
+      "step": 3709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.389970743764779e-06,
+      "loss": 0.7676,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.389501844436609e-06,
+      "loss": 0.7031,
+      "step": 3711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.3890327900303005e-06,
+      "loss": 0.6373,
+      "step": 3712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.38856358058435e-06,
+      "loss": 0.6798,
+      "step": 3713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.388094216137267e-06,
+      "loss": 0.6938,
+      "step": 3714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.387624696727576e-06,
+      "loss": 0.7981,
+      "step": 3715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.38715502239381e-06,
+      "loss": 0.6836,
+      "step": 3716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.386685193174517e-06,
+      "loss": 0.6674,
+      "step": 3717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3862152091082585e-06,
+      "loss": 0.7131,
+      "step": 3718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.385745070233608e-06,
+      "loss": 0.7516,
+      "step": 3719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.38527477658915e-06,
+      "loss": 0.6584,
+      "step": 3720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.384804328213483e-06,
+      "loss": 0.6799,
+      "step": 3721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.38433372514522e-06,
+      "loss": 0.7043,
+      "step": 3722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.383862967422985e-06,
+      "loss": 0.6878,
+      "step": 3723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.383392055085414e-06,
+      "loss": 0.7541,
+      "step": 3724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.382920988171155e-06,
+      "loss": 0.8286,
+      "step": 3725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3824497667188735e-06,
+      "loss": 0.7738,
+      "step": 3726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.381978390767241e-06,
+      "loss": 0.6857,
+      "step": 3727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3815068603549485e-06,
+      "loss": 0.7613,
+      "step": 3728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.381035175520693e-06,
+      "loss": 0.6193,
+      "step": 3729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.380563336303188e-06,
+      "loss": 0.6242,
+      "step": 3730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.38009134274116e-06,
+      "loss": 0.7186,
+      "step": 3731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.379619194873347e-06,
+      "loss": 0.6063,
+      "step": 3732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3791468927385e-06,
+      "loss": 0.6917,
+      "step": 3733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3786744363753806e-06,
+      "loss": 0.7035,
+      "step": 3734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.378201825822768e-06,
+      "loss": 0.6613,
+      "step": 3735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.377729061119449e-06,
+      "loss": 0.6647,
+      "step": 3736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.377256142304226e-06,
+      "loss": 0.8426,
+      "step": 3737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.376783069415912e-06,
+      "loss": 0.6995,
+      "step": 3738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3763098424933355e-06,
+      "loss": 0.7147,
+      "step": 3739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.375836461575335e-06,
+      "loss": 0.6636,
+      "step": 3740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.375362926700761e-06,
+      "loss": 0.7091,
+      "step": 3741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.374889237908481e-06,
+      "loss": 0.7957,
+      "step": 3742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.37441539523737e-06,
+      "loss": 0.7468,
+      "step": 3743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.37394139872632e-06,
+      "loss": 0.7326,
+      "step": 3744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3734672484142325e-06,
+      "loss": 0.7642,
+      "step": 3745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.372992944340022e-06,
+      "loss": 0.7321,
+      "step": 3746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.372518486542617e-06,
+      "loss": 0.8876,
+      "step": 3747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.372043875060959e-06,
+      "loss": 0.6417,
+      "step": 3748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.371569109934e-06,
+      "loss": 0.7633,
+      "step": 3749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.371094191200705e-06,
+      "loss": 0.7263,
+      "step": 3750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.370619118900053e-06,
+      "loss": 0.7147,
+      "step": 3751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.370143893071035e-06,
+      "loss": 0.6375,
+      "step": 3752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.369668513752655e-06,
+      "loss": 0.7397,
+      "step": 3753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.369192980983928e-06,
+      "loss": 0.8027,
+      "step": 3754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.368717294803882e-06,
+      "loss": 0.6974,
+      "step": 3755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.368241455251561e-06,
+      "loss": 0.7376,
+      "step": 3756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.367765462366017e-06,
+      "loss": 0.7493,
+      "step": 3757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.367289316186316e-06,
+      "loss": 0.8068,
+      "step": 3758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.366813016751538e-06,
+      "loss": 0.6526,
+      "step": 3759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.366336564100774e-06,
+      "loss": 0.6855,
+      "step": 3760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.365859958273127e-06,
+      "loss": 0.7202,
+      "step": 3761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.365383199307718e-06,
+      "loss": 0.7398,
+      "step": 3762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.364906287243672e-06,
+      "loss": 0.6817,
+      "step": 3763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.364429222120131e-06,
+      "loss": 0.7995,
+      "step": 3764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.363952003976253e-06,
+      "loss": 0.7009,
+      "step": 3765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.363474632851201e-06,
+      "loss": 0.7176,
+      "step": 3766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.362997108784157e-06,
+      "loss": 0.8597,
+      "step": 3767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.362519431814312e-06,
+      "loss": 0.6475,
+      "step": 3768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.362041601980871e-06,
+      "loss": 0.7448,
+      "step": 3769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.361563619323052e-06,
+      "loss": 0.6953,
+      "step": 3770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.361085483880083e-06,
+      "loss": 0.6891,
+      "step": 3771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.360607195691208e-06,
+      "loss": 0.7803,
+      "step": 3772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.360128754795681e-06,
+      "loss": 0.8628,
+      "step": 3773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.359650161232769e-06,
+      "loss": 0.8274,
+      "step": 3774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.359171415041752e-06,
+      "loss": 0.7943,
+      "step": 3775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.358692516261923e-06,
+      "loss": 0.7162,
+      "step": 3776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3582134649325864e-06,
+      "loss": 0.7366,
+      "step": 3777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.35773426109306e-06,
+      "loss": 0.6666,
+      "step": 3778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3572549047826725e-06,
+      "loss": 0.6528,
+      "step": 3779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.356775396040769e-06,
+      "loss": 0.7346,
+      "step": 3780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.356295734906702e-06,
+      "loss": 0.6795,
+      "step": 3781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3558159214198415e-06,
+      "loss": 0.7569,
+      "step": 3782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3553359556195645e-06,
+      "loss": 0.7545,
+      "step": 3783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.354855837545266e-06,
+      "loss": 0.6462,
+      "step": 3784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.354375567236349e-06,
+      "loss": 0.6346,
+      "step": 3785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.353895144732232e-06,
+      "loss": 0.6631,
+      "step": 3786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.353414570072345e-06,
+      "loss": 0.6921,
+      "step": 3787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.352933843296131e-06,
+      "loss": 0.7396,
+      "step": 3788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3524529644430434e-06,
+      "loss": 0.7846,
+      "step": 3789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3519719335525515e-06,
+      "loss": 0.7773,
+      "step": 3790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.351490750664133e-06,
+      "loss": 0.7946,
+      "step": 3791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.351009415817283e-06,
+      "loss": 0.677,
+      "step": 3792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.350527929051505e-06,
+      "loss": 0.7469,
+      "step": 3793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.350046290406316e-06,
+      "loss": 0.7556,
+      "step": 3794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.349564499921246e-06,
+      "loss": 0.729,
+      "step": 3795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.349082557635836e-06,
+      "loss": 0.7607,
+      "step": 3796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.348600463589644e-06,
+      "loss": 0.6949,
+      "step": 3797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.348118217822235e-06,
+      "loss": 0.7097,
+      "step": 3798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.347635820373188e-06,
+      "loss": 0.7078,
+      "step": 3799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.347153271282096e-06,
+      "loss": 0.8465,
+      "step": 3800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.346670570588565e-06,
+      "loss": 0.6728,
+      "step": 3801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3461877183322086e-06,
+      "loss": 0.9116,
+      "step": 3802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.345704714552659e-06,
+      "loss": 0.7309,
+      "step": 3803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.345221559289556e-06,
+      "loss": 0.7052,
+      "step": 3804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.344738252582557e-06,
+      "loss": 0.5784,
+      "step": 3805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.344254794471325e-06,
+      "loss": 0.6306,
+      "step": 3806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.343771184995541e-06,
+      "loss": 0.6637,
+      "step": 3807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.343287424194896e-06,
+      "loss": 0.7807,
+      "step": 3808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.342803512109095e-06,
+      "loss": 0.731,
+      "step": 3809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.342319448777853e-06,
+      "loss": 0.7294,
+      "step": 3810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.341835234240901e-06,
+      "loss": 0.872,
+      "step": 3811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.341350868537977e-06,
+      "loss": 0.8606,
+      "step": 3812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.340866351708838e-06,
+      "loss": 0.7844,
+      "step": 3813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.340381683793247e-06,
+      "loss": 0.674,
+      "step": 3814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.339896864830984e-06,
+      "loss": 0.8269,
+      "step": 3815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.339411894861839e-06,
+      "loss": 0.8412,
+      "step": 3816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.338926773925616e-06,
+      "loss": 0.7311,
+      "step": 3817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.338441502062131e-06,
+      "loss": 0.7435,
+      "step": 3818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.337956079311211e-06,
+      "loss": 0.771,
+      "step": 3819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.337470505712697e-06,
+      "loss": 0.6107,
+      "step": 3820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.336984781306441e-06,
+      "loss": 0.7163,
+      "step": 3821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.336498906132309e-06,
+      "loss": 0.7482,
+      "step": 3822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3360128802301784e-06,
+      "loss": 0.7626,
+      "step": 3823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.335526703639939e-06,
+      "loss": 0.6851,
+      "step": 3824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.335040376401492e-06,
+      "loss": 0.746,
+      "step": 3825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3345538985547535e-06,
+      "loss": 0.6555,
+      "step": 3826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.334067270139649e-06,
+      "loss": 0.6195,
+      "step": 3827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.333580491196119e-06,
+      "loss": 0.765,
+      "step": 3828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.333093561764115e-06,
+      "loss": 0.7891,
+      "step": 3829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.332606481883601e-06,
+      "loss": 0.7459,
+      "step": 3830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.332119251594553e-06,
+      "loss": 0.6798,
+      "step": 3831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.33163187093696e-06,
+      "loss": 0.8384,
+      "step": 3832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.331144339950822e-06,
+      "loss": 0.8222,
+      "step": 3833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.330656658676155e-06,
+      "loss": 0.9111,
+      "step": 3834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.330168827152983e-06,
+      "loss": 0.8472,
+      "step": 3835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.329680845421343e-06,
+      "loss": 0.7561,
+      "step": 3836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.329192713521287e-06,
+      "loss": 0.6566,
+      "step": 3837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3287044314928786e-06,
+      "loss": 0.7422,
+      "step": 3838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.328215999376191e-06,
+      "loss": 0.7672,
+      "step": 3839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.327727417211312e-06,
+      "loss": 0.7807,
+      "step": 3840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.32723868503834e-06,
+      "loss": 0.6658,
+      "step": 3841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.326749802897391e-06,
+      "loss": 0.693,
+      "step": 3842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.326260770828586e-06,
+      "loss": 0.7093,
+      "step": 3843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.325771588872061e-06,
+      "loss": 0.7078,
+      "step": 3844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.325282257067966e-06,
+      "loss": 0.8712,
+      "step": 3845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.324792775456464e-06,
+      "loss": 0.7378,
+      "step": 3846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.324303144077726e-06,
+      "loss": 0.7523,
+      "step": 3847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.323813362971939e-06,
+      "loss": 0.7509,
+      "step": 3848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.323323432179301e-06,
+      "loss": 0.8268,
+      "step": 3849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.322833351740022e-06,
+      "loss": 0.8278,
+      "step": 3850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.322343121694325e-06,
+      "loss": 0.6772,
+      "step": 3851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.321852742082444e-06,
+      "loss": 0.6677,
+      "step": 3852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.321362212944628e-06,
+      "loss": 0.6278,
+      "step": 3853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.320871534321134e-06,
+      "loss": 0.6725,
+      "step": 3854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.320380706252236e-06,
+      "loss": 0.6955,
+      "step": 3855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3198897287782165e-06,
+      "loss": 0.6437,
+      "step": 3856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.319398601939373e-06,
+      "loss": 0.6662,
+      "step": 3857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.318907325776013e-06,
+      "loss": 0.7271,
+      "step": 3858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.318415900328458e-06,
+      "loss": 0.8007,
+      "step": 3859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.317924325637041e-06,
+      "loss": 0.7431,
+      "step": 3860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.317432601742106e-06,
+      "loss": 0.6573,
+      "step": 3861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3169407286840135e-06,
+      "loss": 0.7558,
+      "step": 3862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.316448706503129e-06,
+      "loss": 0.6704,
+      "step": 3863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3159565352398385e-06,
+      "loss": 0.7011,
+      "step": 3864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.315464214934534e-06,
+      "loss": 0.7277,
+      "step": 3865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.314971745627623e-06,
+      "loss": 0.7685,
+      "step": 3866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.314479127359523e-06,
+      "loss": 0.6421,
+      "step": 3867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.313986360170666e-06,
+      "loss": 0.6577,
+      "step": 3868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3134934441014955e-06,
+      "loss": 0.7396,
+      "step": 3869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.313000379192466e-06,
+      "loss": 0.7764,
+      "step": 3870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.312507165484045e-06,
+      "loss": 0.5739,
+      "step": 3871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.312013803016712e-06,
+      "loss": 0.7302,
+      "step": 3872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.311520291830961e-06,
+      "loss": 0.7299,
+      "step": 3873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3110266319672935e-06,
+      "loss": 0.7453,
+      "step": 3874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.310532823466228e-06,
+      "loss": 0.9753,
+      "step": 3875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.310038866368292e-06,
+      "loss": 0.8206,
+      "step": 3876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.309544760714027e-06,
+      "loss": 0.7729,
+      "step": 3877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3090505065439855e-06,
+      "loss": 0.754,
+      "step": 3878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3085561038987335e-06,
+      "loss": 0.6394,
+      "step": 3879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.308061552818848e-06,
+      "loss": 0.7387,
+      "step": 3880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.307566853344918e-06,
+      "loss": 0.6347,
+      "step": 3881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.307072005517545e-06,
+      "loss": 0.8172,
+      "step": 3882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.306577009377344e-06,
+      "loss": 0.6374,
+      "step": 3883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.306081864964941e-06,
+      "loss": 0.7582,
+      "step": 3884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.305586572320974e-06,
+      "loss": 0.915,
+      "step": 3885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3050911314860925e-06,
+      "loss": 0.7125,
+      "step": 3886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.304595542500961e-06,
+      "loss": 0.6782,
+      "step": 3887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.304099805406252e-06,
+      "loss": 0.6841,
+      "step": 3888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.303603920242654e-06,
+      "loss": 0.7274,
+      "step": 3889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.303107887050867e-06,
+      "loss": 0.8553,
+      "step": 3890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.302611705871599e-06,
+      "loss": 0.771,
+      "step": 3891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.302115376745576e-06,
+      "loss": 0.7219,
+      "step": 3892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.301618899713533e-06,
+      "loss": 0.7777,
+      "step": 3893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3011222748162165e-06,
+      "loss": 0.703,
+      "step": 3894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.300625502094388e-06,
+      "loss": 0.6356,
+      "step": 3895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3001285815888174e-06,
+      "loss": 0.6629,
+      "step": 3896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.29963151334029e-06,
+      "loss": 0.6964,
+      "step": 3897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.299134297389602e-06,
+      "loss": 0.742,
+      "step": 3898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2986369337775625e-06,
+      "loss": 0.6728,
+      "step": 3899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.29813942254499e-06,
+      "loss": 0.6874,
+      "step": 3900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.297641763732718e-06,
+      "loss": 0.8578,
+      "step": 3901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.297143957381592e-06,
+      "loss": 0.7753,
+      "step": 3902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.296646003532467e-06,
+      "loss": 0.7534,
+      "step": 3903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.296147902226212e-06,
+      "loss": 0.771,
+      "step": 3904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.295649653503709e-06,
+      "loss": 0.7661,
+      "step": 3905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.29515125740585e-06,
+      "loss": 0.8145,
+      "step": 3906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.294652713973542e-06,
+      "loss": 0.708,
+      "step": 3907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2941540232477005e-06,
+      "loss": 0.8261,
+      "step": 3908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.293655185269255e-06,
+      "loss": 0.7146,
+      "step": 3909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.293156200079149e-06,
+      "loss": 0.8409,
+      "step": 3910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.292657067718331e-06,
+      "loss": 0.681,
+      "step": 3911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.292157788227772e-06,
+      "loss": 0.6761,
+      "step": 3912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.291658361648447e-06,
+      "loss": 0.8432,
+      "step": 3913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.291158788021347e-06,
+      "loss": 0.6831,
+      "step": 3914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.290659067387471e-06,
+      "loss": 0.7199,
+      "step": 3915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2901591997878354e-06,
+      "loss": 0.7721,
+      "step": 3916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2896591852634665e-06,
+      "loss": 0.7257,
+      "step": 3917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2891590238554004e-06,
+      "loss": 0.7612,
+      "step": 3918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.288658715604688e-06,
+      "loss": 0.7264,
+      "step": 3919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.288158260552392e-06,
+      "loss": 0.8424,
+      "step": 3920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.287657658739584e-06,
+      "loss": 0.7585,
+      "step": 3921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.287156910207354e-06,
+      "loss": 0.8332,
+      "step": 3922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2866560149967976e-06,
+      "loss": 0.8016,
+      "step": 3923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.286154973149026e-06,
+      "loss": 0.8051,
+      "step": 3924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.285653784705161e-06,
+      "loss": 0.6859,
+      "step": 3925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.285152449706337e-06,
+      "loss": 0.7886,
+      "step": 3926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.284650968193701e-06,
+      "loss": 0.6084,
+      "step": 3927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.284149340208411e-06,
+      "loss": 0.6037,
+      "step": 3928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.283647565791637e-06,
+      "loss": 0.7996,
+      "step": 3929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.283145644984561e-06,
+      "loss": 0.7869,
+      "step": 3930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.282643577828378e-06,
+      "loss": 0.6945,
+      "step": 3931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.282141364364295e-06,
+      "loss": 0.7323,
+      "step": 3932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.28163900463353e-06,
+      "loss": 0.7201,
+      "step": 3933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.281136498677313e-06,
+      "loss": 0.7036,
+      "step": 3934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.280633846536888e-06,
+      "loss": 0.6151,
+      "step": 3935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.280131048253507e-06,
+      "loss": 0.6833,
+      "step": 3936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.279628103868439e-06,
+      "loss": 0.7047,
+      "step": 3937
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.27912501342296e-06,
+      "loss": 0.7591,
+      "step": 3938
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.278621776958362e-06,
+      "loss": 0.7248,
+      "step": 3939
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2781183945159474e-06,
+      "loss": 0.6404,
+      "step": 3940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2776148661370295e-06,
+      "loss": 0.67,
+      "step": 3941
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.277111191862936e-06,
+      "loss": 0.6535,
+      "step": 3942
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.276607371735004e-06,
+      "loss": 0.7246,
+      "step": 3943
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.276103405794585e-06,
+      "loss": 0.7045,
+      "step": 3944
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.27559929408304e-06,
+      "loss": 0.7596,
+      "step": 3945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.275095036641745e-06,
+      "loss": 0.8033,
+      "step": 3946
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.274590633512085e-06,
+      "loss": 0.6684,
+      "step": 3947
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.274086084735458e-06,
+      "loss": 0.6765,
+      "step": 3948
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.273581390353274e-06,
+      "loss": 0.6603,
+      "step": 3949
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.273076550406956e-06,
+      "loss": 0.7247,
+      "step": 3950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2725715649379375e-06,
+      "loss": 0.7674,
+      "step": 3951
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.272066433987664e-06,
+      "loss": 0.8596,
+      "step": 3952
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.271561157597595e-06,
+      "loss": 0.7588,
+      "step": 3953
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.271055735809197e-06,
+      "loss": 0.7321,
+      "step": 3954
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.270550168663955e-06,
+      "loss": 0.7573,
+      "step": 3955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2700444562033625e-06,
+      "loss": 0.692,
+      "step": 3956
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.269538598468923e-06,
+      "loss": 0.7581,
+      "step": 3957
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.269032595502157e-06,
+      "loss": 0.7581,
+      "step": 3958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2685264473445906e-06,
+      "loss": 0.6909,
+      "step": 3959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.268020154037767e-06,
+      "loss": 0.7595,
+      "step": 3960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.26751371562324e-06,
+      "loss": 0.7272,
+      "step": 3961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.267007132142574e-06,
+      "loss": 0.6847,
+      "step": 3962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.266500403637346e-06,
+      "loss": 0.7067,
+      "step": 3963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.265993530149145e-06,
+      "loss": 0.735,
+      "step": 3964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.265486511719573e-06,
+      "loss": 0.7406,
+      "step": 3965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2649793483902415e-06,
+      "loss": 0.798,
+      "step": 3966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.264472040202775e-06,
+      "loss": 0.7522,
+      "step": 3967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.263964587198812e-06,
+      "loss": 0.8309,
+      "step": 3968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.263456989419999e-06,
+      "loss": 0.671,
+      "step": 3969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.262949246907998e-06,
+      "loss": 0.7892,
+      "step": 3970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.262441359704479e-06,
+      "loss": 0.7835,
+      "step": 3971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.261933327851128e-06,
+      "loss": 0.7055,
+      "step": 3972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.26142515138964e-06,
+      "loss": 0.651,
+      "step": 3973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.260916830361724e-06,
+      "loss": 0.6944,
+      "step": 3974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.260408364809099e-06,
+      "loss": 0.7885,
+      "step": 3975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2598997547734955e-06,
+      "loss": 0.6993,
+      "step": 3976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.259391000296659e-06,
+      "loss": 0.8847,
+      "step": 3977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.258882101420343e-06,
+      "loss": 0.816,
+      "step": 3978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.258373058186316e-06,
+      "loss": 0.7548,
+      "step": 3979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.257863870636357e-06,
+      "loss": 0.7288,
+      "step": 3980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.257354538812255e-06,
+      "loss": 0.6062,
+      "step": 3981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.256845062755814e-06,
+      "loss": 0.8382,
+      "step": 3982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.256335442508849e-06,
+      "loss": 0.7563,
+      "step": 3983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.255825678113186e-06,
+      "loss": 0.7436,
+      "step": 3984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.255315769610663e-06,
+      "loss": 0.6359,
+      "step": 3985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.254805717043129e-06,
+      "loss": 0.6625,
+      "step": 3986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.254295520452447e-06,
+      "loss": 0.7268,
+      "step": 3987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.253785179880491e-06,
+      "loss": 0.7361,
+      "step": 3988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.253274695369146e-06,
+      "loss": 0.7521,
+      "step": 3989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.252764066960308e-06,
+      "loss": 0.7734,
+      "step": 3990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2522532946958885e-06,
+      "loss": 0.7888,
+      "step": 3991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.251742378617808e-06,
+      "loss": 0.7401,
+      "step": 3992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.251231318767997e-06,
+      "loss": 0.7609,
+      "step": 3993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.250720115188401e-06,
+      "loss": 0.7417,
+      "step": 3994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.250208767920978e-06,
+      "loss": 0.6853,
+      "step": 3995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.249697277007693e-06,
+      "loss": 0.7468,
+      "step": 3996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.249185642490529e-06,
+      "loss": 0.6122,
+      "step": 3997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.248673864411476e-06,
+      "loss": 0.7335,
+      "step": 3998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.248161942812537e-06,
+      "loss": 0.6498,
+      "step": 3999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.247649877735729e-06,
+      "loss": 0.7604,
+      "step": 4000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2471376692230785e-06,
+      "loss": 0.7888,
+      "step": 4001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.246625317316622e-06,
+      "loss": 0.6212,
+      "step": 4002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.246112822058412e-06,
+      "loss": 0.7194,
+      "step": 4003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2456001834905125e-06,
+      "loss": 0.8883,
+      "step": 4004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.245087401654993e-06,
+      "loss": 0.8486,
+      "step": 4005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.244574476593945e-06,
+      "loss": 0.8367,
+      "step": 4006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.244061408349462e-06,
+      "loss": 0.7052,
+      "step": 4007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.243548196963653e-06,
+      "loss": 0.6942,
+      "step": 4008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.243034842478641e-06,
+      "loss": 0.8096,
+      "step": 4009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.242521344936559e-06,
+      "loss": 0.8215,
+      "step": 4010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2420077043795514e-06,
+      "loss": 0.6658,
+      "step": 4011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.241493920849773e-06,
+      "loss": 0.7281,
+      "step": 4012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2409799943893935e-06,
+      "loss": 0.7,
+      "step": 4013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.240465925040592e-06,
+      "loss": 0.8331,
+      "step": 4014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.239951712845559e-06,
+      "loss": 0.798,
+      "step": 4015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2394373578465e-06,
+      "loss": 0.7526,
+      "step": 4016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.238922860085628e-06,
+      "loss": 0.7388,
+      "step": 4017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.238408219605171e-06,
+      "loss": 0.6653,
+      "step": 4018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.237893436447366e-06,
+      "loss": 0.8029,
+      "step": 4019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.237378510654464e-06,
+      "loss": 0.8891,
+      "step": 4020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.236863442268727e-06,
+      "loss": 0.8792,
+      "step": 4021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.236348231332428e-06,
+      "loss": 0.7439,
+      "step": 4022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.235832877887852e-06,
+      "loss": 0.635,
+      "step": 4023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.235317381977297e-06,
+      "loss": 0.7688,
+      "step": 4024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.23480174364307e-06,
+      "loss": 0.7778,
+      "step": 4025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2342859629274926e-06,
+      "loss": 0.8263,
+      "step": 4026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.233770039872896e-06,
+      "loss": 0.8028,
+      "step": 4027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.233253974521623e-06,
+      "loss": 0.7022,
+      "step": 4028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.232737766916031e-06,
+      "loss": 0.6198,
+      "step": 4029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.232221417098486e-06,
+      "loss": 0.7315,
+      "step": 4030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.231704925111366e-06,
+      "loss": 0.6354,
+      "step": 4031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.231188290997063e-06,
+      "loss": 0.6491,
+      "step": 4032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.230671514797977e-06,
+      "loss": 0.7105,
+      "step": 4033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.230154596556523e-06,
+      "loss": 1.0078,
+      "step": 4034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.229637536315127e-06,
+      "loss": 0.7634,
+      "step": 4035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.229120334116223e-06,
+      "loss": 0.7166,
+      "step": 4036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.228602990002263e-06,
+      "loss": 0.7786,
+      "step": 4037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.228085504015705e-06,
+      "loss": 0.6971,
+      "step": 4038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.227567876199022e-06,
+      "loss": 0.7833,
+      "step": 4039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.227050106594697e-06,
+      "loss": 0.7944,
+      "step": 4040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.226532195245226e-06,
+      "loss": 0.6933,
+      "step": 4041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.226014142193116e-06,
+      "loss": 0.6438,
+      "step": 4042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.225495947480883e-06,
+      "loss": 0.7596,
+      "step": 4043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.22497761115106e-06,
+      "loss": 0.8001,
+      "step": 4044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.224459133246187e-06,
+      "loss": 0.765,
+      "step": 4045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2239405138088184e-06,
+      "loss": 0.7824,
+      "step": 4046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.223421752881518e-06,
+      "loss": 0.7603,
+      "step": 4047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2229028505068635e-06,
+      "loss": 0.7853,
+      "step": 4048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.222383806727442e-06,
+      "loss": 0.8452,
+      "step": 4049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.221864621585854e-06,
+      "loss": 0.7372,
+      "step": 4050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2213452951247114e-06,
+      "loss": 0.7029,
+      "step": 4051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.220825827386636e-06,
+      "loss": 0.8158,
+      "step": 4052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.220306218414263e-06,
+      "loss": 0.7029,
+      "step": 4053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2197864682502375e-06,
+      "loss": 0.8516,
+      "step": 4054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2192665769372185e-06,
+      "loss": 0.7502,
+      "step": 4055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.218746544517876e-06,
+      "loss": 0.7266,
+      "step": 4056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.218226371034889e-06,
+      "loss": 0.7438,
+      "step": 4057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.217706056530951e-06,
+      "loss": 0.8154,
+      "step": 4058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.217185601048765e-06,
+      "loss": 0.6651,
+      "step": 4059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.216665004631048e-06,
+      "loss": 0.6791,
+      "step": 4060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.216144267320528e-06,
+      "loss": 0.6604,
+      "step": 4061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.215623389159942e-06,
+      "loss": 0.7751,
+      "step": 4062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.21510237019204e-06,
+      "loss": 0.7555,
+      "step": 4063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.214581210459585e-06,
+      "loss": 0.7221,
+      "step": 4064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.214059910005351e-06,
+      "loss": 0.8292,
+      "step": 4065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.213538468872122e-06,
+      "loss": 0.7112,
+      "step": 4066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2130168871026935e-06,
+      "loss": 0.656,
+      "step": 4067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.212495164739875e-06,
+      "loss": 0.6465,
+      "step": 4068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.2119733018264866e-06,
+      "loss": 0.737,
+      "step": 4069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.211451298405358e-06,
+      "loss": 0.809,
+      "step": 4070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.210929154519333e-06,
+      "loss": 0.6678,
+      "step": 4071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.210406870211265e-06,
+      "loss": 0.8214,
+      "step": 4072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.209884445524021e-06,
+      "loss": 0.7347,
+      "step": 4073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.209361880500477e-06,
+      "loss": 0.7064,
+      "step": 4074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.208839175183521e-06,
+      "loss": 0.7347,
+      "step": 4075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.208316329616056e-06,
+      "loss": 0.7262,
+      "step": 4076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.207793343840991e-06,
+      "loss": 0.7431,
+      "step": 4077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.207270217901252e-06,
+      "loss": 0.8673,
+      "step": 4078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.20674695183977e-06,
+      "loss": 0.7336,
+      "step": 4079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.206223545699496e-06,
+      "loss": 0.5833,
+      "step": 4080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.205699999523384e-06,
+      "loss": 0.7397,
+      "step": 4081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.205176313354405e-06,
+      "loss": 0.6882,
+      "step": 4082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.204652487235539e-06,
+      "loss": 0.7492,
+      "step": 4083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.20412852120978e-06,
+      "loss": 0.7134,
+      "step": 4084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.20360441532013e-06,
+      "loss": 0.699,
+      "step": 4085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2030801696096045e-06,
+      "loss": 0.8578,
+      "step": 4086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.202555784121231e-06,
+      "loss": 0.7284,
+      "step": 4087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.202031258898046e-06,
+      "loss": 0.748,
+      "step": 4088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.201506593983102e-06,
+      "loss": 0.8467,
+      "step": 4089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2009817894194575e-06,
+      "loss": 0.7992,
+      "step": 4090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.200456845250186e-06,
+      "loss": 0.6933,
+      "step": 4091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.199931761518371e-06,
+      "loss": 0.725,
+      "step": 4092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.19940653826711e-06,
+      "loss": 0.647,
+      "step": 4093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.198881175539507e-06,
+      "loss": 0.6829,
+      "step": 4094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.198355673378684e-06,
+      "loss": 0.6608,
+      "step": 4095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.197830031827766e-06,
+      "loss": 0.761,
+      "step": 4096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.197304250929899e-06,
+      "loss": 0.8282,
+      "step": 4097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1967783307282325e-06,
+      "loss": 0.7939,
+      "step": 4098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.196252271265933e-06,
+      "loss": 0.6769,
+      "step": 4099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.195726072586174e-06,
+      "loss": 0.7298,
+      "step": 4100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1951997347321424e-06,
+      "loss": 0.7532,
+      "step": 4101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1946732577470396e-06,
+      "loss": 0.6227,
+      "step": 4102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1941466416740725e-06,
+      "loss": 0.7212,
+      "step": 4103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.193619886556464e-06,
+      "loss": 0.8498,
+      "step": 4104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.193092992437445e-06,
+      "loss": 0.7163,
+      "step": 4105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.192565959360261e-06,
+      "loss": 0.7224,
+      "step": 4106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.192038787368166e-06,
+      "loss": 0.7714,
+      "step": 4107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.19151147650443e-06,
+      "loss": 0.6814,
+      "step": 4108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.190984026812329e-06,
+      "loss": 0.7481,
+      "step": 4109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.190456438335153e-06,
+      "loss": 0.6261,
+      "step": 4110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.189928711116203e-06,
+      "loss": 0.6805,
+      "step": 4111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.18940084519879e-06,
+      "loss": 0.697,
+      "step": 4112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.188872840626241e-06,
+      "loss": 0.7808,
+      "step": 4113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.188344697441889e-06,
+      "loss": 0.6817,
+      "step": 4114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.187816415689081e-06,
+      "loss": 0.6838,
+      "step": 4115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.187287995411175e-06,
+      "loss": 0.6913,
+      "step": 4116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1867594366515396e-06,
+      "loss": 0.6397,
+      "step": 4117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.186230739453557e-06,
+      "loss": 0.745,
+      "step": 4118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.18570190386062e-06,
+      "loss": 0.8153,
+      "step": 4119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1851729299161285e-06,
+      "loss": 0.6904,
+      "step": 4120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1846438176635e-06,
+      "loss": 0.7414,
+      "step": 4121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.184114567146161e-06,
+      "loss": 0.7676,
+      "step": 4122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.183585178407547e-06,
+      "loss": 0.8196,
+      "step": 4123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.183055651491107e-06,
+      "loss": 0.636,
+      "step": 4124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.182525986440303e-06,
+      "loss": 0.7299,
+      "step": 4125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1819961832986055e-06,
+      "loss": 0.9058,
+      "step": 4126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1814662421094975e-06,
+      "loss": 0.8057,
+      "step": 4127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1809361629164725e-06,
+      "loss": 0.6877,
+      "step": 4128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.180405945763036e-06,
+      "loss": 0.8015,
+      "step": 4129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.179875590692705e-06,
+      "loss": 0.7324,
+      "step": 4130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.179345097749009e-06,
+      "loss": 0.8253,
+      "step": 4131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.178814466975486e-06,
+      "loss": 0.6684,
+      "step": 4132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.178283698415686e-06,
+      "loss": 0.6972,
+      "step": 4133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.177752792113174e-06,
+      "loss": 0.6406,
+      "step": 4134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.177221748111519e-06,
+      "loss": 0.7346,
+      "step": 4135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.176690566454309e-06,
+      "loss": 0.768,
+      "step": 4136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.176159247185139e-06,
+      "loss": 0.6329,
+      "step": 4137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.175627790347618e-06,
+      "loss": 0.7975,
+      "step": 4138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.175096195985362e-06,
+      "loss": 0.7429,
+      "step": 4139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.174564464142001e-06,
+      "loss": 0.7675,
+      "step": 4140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.174032594861178e-06,
+      "loss": 0.8542,
+      "step": 4141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.173500588186543e-06,
+      "loss": 0.7583,
+      "step": 4142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.172968444161762e-06,
+      "loss": 0.9033,
+      "step": 4143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.172436162830509e-06,
+      "loss": 0.6852,
+      "step": 4144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.171903744236469e-06,
+      "loss": 0.7716,
+      "step": 4145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1713711884233425e-06,
+      "loss": 0.7062,
+      "step": 4146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.170838495434835e-06,
+      "loss": 0.9228,
+      "step": 4147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.170305665314667e-06,
+      "loss": 0.8722,
+      "step": 4148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.169772698106572e-06,
+      "loss": 0.756,
+      "step": 4149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.169239593854291e-06,
+      "loss": 0.698,
+      "step": 4150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.168706352601578e-06,
+      "loss": 0.6887,
+      "step": 4151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.168172974392198e-06,
+      "loss": 0.6144,
+      "step": 4152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.167639459269927e-06,
+      "loss": 0.8321,
+      "step": 4153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1671058072785534e-06,
+      "loss": 0.7314,
+      "step": 4154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.166572018461874e-06,
+      "loss": 0.7573,
+      "step": 4155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1660380928637e-06,
+      "loss": 0.6849,
+      "step": 4156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.165504030527853e-06,
+      "loss": 0.7798,
+      "step": 4157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.164969831498165e-06,
+      "loss": 0.6801,
+      "step": 4158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1644354958184795e-06,
+      "loss": 0.8147,
+      "step": 4159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.163901023532652e-06,
+      "loss": 0.7722,
+      "step": 4160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.163366414684547e-06,
+      "loss": 0.7084,
+      "step": 4161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.162831669318042e-06,
+      "loss": 0.6516,
+      "step": 4162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.162296787477027e-06,
+      "loss": 0.7154,
+      "step": 4163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1617617692054e-06,
+      "loss": 0.7513,
+      "step": 4164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1612266145470734e-06,
+      "loss": 0.7249,
+      "step": 4165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.160691323545969e-06,
+      "loss": 0.6274,
+      "step": 4166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.160155896246019e-06,
+      "loss": 0.6737,
+      "step": 4167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1596203326911675e-06,
+      "loss": 0.7327,
+      "step": 4168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1590846329253715e-06,
+      "loss": 0.8106,
+      "step": 4169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.158548796992598e-06,
+      "loss": 0.8588,
+      "step": 4170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1580128249368224e-06,
+      "loss": 0.7081,
+      "step": 4171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.157476716802037e-06,
+      "loss": 0.6248,
+      "step": 4172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.15694047263224e-06,
+      "loss": 0.7153,
+      "step": 4173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.156404092471444e-06,
+      "loss": 0.7155,
+      "step": 4174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1558675763636715e-06,
+      "loss": 0.7444,
+      "step": 4175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.155330924352955e-06,
+      "loss": 0.7603,
+      "step": 4176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.154794136483341e-06,
+      "loss": 0.6784,
+      "step": 4177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.154257212798885e-06,
+      "loss": 0.6816,
+      "step": 4178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.153720153343654e-06,
+      "loss": 0.6596,
+      "step": 4179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.153182958161727e-06,
+      "loss": 0.6583,
+      "step": 4180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.152645627297193e-06,
+      "loss": 0.7963,
+      "step": 4181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.152108160794153e-06,
+      "loss": 0.6717,
+      "step": 4182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.151570558696718e-06,
+      "loss": 0.6777,
+      "step": 4183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.151032821049013e-06,
+      "loss": 0.6608,
+      "step": 4184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.150494947895168e-06,
+      "loss": 0.734,
+      "step": 4185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.149956939279333e-06,
+      "loss": 0.8336,
+      "step": 4186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.149418795245661e-06,
+      "loss": 0.6948,
+      "step": 4187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.14888051583832e-06,
+      "loss": 0.7069,
+      "step": 4188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1483421011014905e-06,
+      "loss": 0.8279,
+      "step": 4189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.147803551079359e-06,
+      "loss": 0.9009,
+      "step": 4190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.147264865816128e-06,
+      "loss": 0.6753,
+      "step": 4191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.146726045356009e-06,
+      "loss": 0.7383,
+      "step": 4192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.146187089743225e-06,
+      "loss": 0.7254,
+      "step": 4193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.145647999022011e-06,
+      "loss": 0.7749,
+      "step": 4194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.145108773236609e-06,
+      "loss": 0.749,
+      "step": 4195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.144569412431279e-06,
+      "loss": 0.6717,
+      "step": 4196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.144029916650286e-06,
+      "loss": 0.6394,
+      "step": 4197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.143490285937909e-06,
+      "loss": 0.6447,
+      "step": 4198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.142950520338437e-06,
+      "loss": 0.7935,
+      "step": 4199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.14241061989617e-06,
+      "loss": 0.687,
+      "step": 4200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1418705846554215e-06,
+      "loss": 0.8116,
+      "step": 4201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.141330414660512e-06,
+      "loss": 0.8003,
+      "step": 4202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1407901099557766e-06,
+      "loss": 0.8253,
+      "step": 4203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1402496705855595e-06,
+      "loss": 0.6615,
+      "step": 4204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.139709096594216e-06,
+      "loss": 0.8033,
+      "step": 4205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.139168388026113e-06,
+      "loss": 0.7445,
+      "step": 4206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1386275449256304e-06,
+      "loss": 0.773,
+      "step": 4207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.138086567337155e-06,
+      "loss": 0.7196,
+      "step": 4208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.137545455305086e-06,
+      "loss": 0.746,
+      "step": 4209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.137004208873837e-06,
+      "loss": 0.6846,
+      "step": 4210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.136462828087828e-06,
+      "loss": 0.6378,
+      "step": 4211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.135921312991493e-06,
+      "loss": 0.7645,
+      "step": 4212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.135379663629275e-06,
+      "loss": 0.7608,
+      "step": 4213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1348378800456305e-06,
+      "loss": 0.7447,
+      "step": 4214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.134295962285026e-06,
+      "loss": 0.7132,
+      "step": 4215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.133753910391936e-06,
+      "loss": 0.7936,
+      "step": 4216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.133211724410851e-06,
+      "loss": 0.7142,
+      "step": 4217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.132669404386269e-06,
+      "loss": 0.7287,
+      "step": 4218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.132126950362701e-06,
+      "loss": 0.6636,
+      "step": 4219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.131584362384667e-06,
+      "loss": 0.6826,
+      "step": 4220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1310416404967e-06,
+      "loss": 0.753,
+      "step": 4221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.130498784743343e-06,
+      "loss": 0.9058,
+      "step": 4222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.129955795169149e-06,
+      "loss": 0.7957,
+      "step": 4223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.129412671818686e-06,
+      "loss": 0.8293,
+      "step": 4224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.128869414736526e-06,
+      "loss": 0.7307,
+      "step": 4225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.12832602396726e-06,
+      "loss": 0.7698,
+      "step": 4226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.127782499555482e-06,
+      "loss": 0.6924,
+      "step": 4227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.127238841545805e-06,
+      "loss": 0.9881,
+      "step": 4228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.126695049982847e-06,
+      "loss": 0.7501,
+      "step": 4229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.126151124911239e-06,
+      "loss": 0.7469,
+      "step": 4230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.125607066375622e-06,
+      "loss": 0.82,
+      "step": 4231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.12506287442065e-06,
+      "loss": 0.7003,
+      "step": 4232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.124518549090987e-06,
+      "loss": 0.7496,
+      "step": 4233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.123974090431306e-06,
+      "loss": 0.7276,
+      "step": 4234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1234294984862954e-06,
+      "loss": 0.653,
+      "step": 4235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.12288477330065e-06,
+      "loss": 0.8156,
+      "step": 4236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1223399149190765e-06,
+      "loss": 0.6889,
+      "step": 4237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.121794923386295e-06,
+      "loss": 0.6424,
+      "step": 4238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.121249798747035e-06,
+      "loss": 0.7634,
+      "step": 4239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.120704541046034e-06,
+      "loss": 0.8211,
+      "step": 4240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.120159150328048e-06,
+      "loss": 0.7128,
+      "step": 4241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.119613626637835e-06,
+      "loss": 0.7166,
+      "step": 4242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.119067970020169e-06,
+      "loss": 0.8904,
+      "step": 4243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.118522180519837e-06,
+      "loss": 0.6852,
+      "step": 4244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1179762581816296e-06,
+      "loss": 0.9316,
+      "step": 4245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1174302030503546e-06,
+      "loss": 0.6604,
+      "step": 4246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.116884015170829e-06,
+      "loss": 0.8041,
+      "step": 4247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.11633769458788e-06,
+      "loss": 0.711,
+      "step": 4248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.115791241346346e-06,
+      "loss": 0.7184,
+      "step": 4249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1152446554910766e-06,
+      "loss": 0.8524,
+      "step": 4250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.114697937066932e-06,
+      "loss": 0.6697,
+      "step": 4251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.114151086118781e-06,
+      "loss": 0.6951,
+      "step": 4252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.11360410269151e-06,
+      "loss": 0.7386,
+      "step": 4253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.113056986830009e-06,
+      "loss": 0.7842,
+      "step": 4254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.112509738579182e-06,
+      "loss": 0.7146,
+      "step": 4255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.111962357983945e-06,
+      "loss": 0.7139,
+      "step": 4256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.111414845089222e-06,
+      "loss": 0.8353,
+      "step": 4257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.11086719993995e-06,
+      "loss": 0.8017,
+      "step": 4258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1103194225810746e-06,
+      "loss": 0.8177,
+      "step": 4259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.109771513057556e-06,
+      "loss": 0.7794,
+      "step": 4260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.109223471414363e-06,
+      "loss": 0.8566,
+      "step": 4261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1086752976964735e-06,
+      "loss": 0.7696,
+      "step": 4262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.10812699194888e-06,
+      "loss": 0.8024,
+      "step": 4263
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.8261173367500305,
+      "eval_runtime": 437.534,
+      "eval_samples_per_second": 6.854,
+      "eval_steps_per_second": 0.857,
+      "step": 4263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.107578554216583e-06,
+      "loss": 0.8735,
+      "step": 4264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.107029984544595e-06,
+      "loss": 0.777,
+      "step": 4265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1064812829779386e-06,
+      "loss": 0.612,
+      "step": 4266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.105932449561648e-06,
+      "loss": 0.7081,
+      "step": 4267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1053834843407685e-06,
+      "loss": 0.637,
+      "step": 4268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.104834387360354e-06,
+      "loss": 0.5945,
+      "step": 4269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.104285158665472e-06,
+      "loss": 0.7913,
+      "step": 4270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1037357983012e-06,
+      "loss": 0.7759,
+      "step": 4271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.1031863063126255e-06,
+      "loss": 0.7499,
+      "step": 4272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.102636682744846e-06,
+      "loss": 0.692,
+      "step": 4273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.102086927642974e-06,
+      "loss": 0.886,
+      "step": 4274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.101537041052127e-06,
+      "loss": 0.7177,
+      "step": 4275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.100987023017437e-06,
+      "loss": 0.8157,
+      "step": 4276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.100436873584048e-06,
+      "loss": 0.6407,
+      "step": 4277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.09988659279711e-06,
+      "loss": 0.7231,
+      "step": 4278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.099336180701787e-06,
+      "loss": 0.7465,
+      "step": 4279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.098785637343254e-06,
+      "loss": 0.7699,
+      "step": 4280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0982349627666965e-06,
+      "loss": 0.7231,
+      "step": 4281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.097684157017309e-06,
+      "loss": 0.6235,
+      "step": 4282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.097133220140298e-06,
+      "loss": 0.674,
+      "step": 4283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0965821521808825e-06,
+      "loss": 0.8344,
+      "step": 4284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.09603095318429e-06,
+      "loss": 0.6751,
+      "step": 4285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.095479623195758e-06,
+      "loss": 0.8025,
+      "step": 4286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.094928162260538e-06,
+      "loss": 0.7105,
+      "step": 4287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.094376570423889e-06,
+      "loss": 0.7291,
+      "step": 4288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.093824847731082e-06,
+      "loss": 0.6482,
+      "step": 4289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.093272994227401e-06,
+      "loss": 0.7528,
+      "step": 4290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.092721009958137e-06,
+      "loss": 0.6911,
+      "step": 4291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.092168894968592e-06,
+      "loss": 0.7288,
+      "step": 4292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.091616649304082e-06,
+      "loss": 0.6335,
+      "step": 4293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.091064273009931e-06,
+      "loss": 0.745,
+      "step": 4294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0905117661314745e-06,
+      "loss": 0.7148,
+      "step": 4295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0899591287140596e-06,
+      "loss": 0.6468,
+      "step": 4296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.089406360803042e-06,
+      "loss": 0.7006,
+      "step": 4297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.088853462443789e-06,
+      "loss": 0.7982,
+      "step": 4298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.088300433681681e-06,
+      "loss": 0.6665,
+      "step": 4299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.087747274562105e-06,
+      "loss": 0.7117,
+      "step": 4300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.087193985130462e-06,
+      "loss": 0.6697,
+      "step": 4301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.086640565432162e-06,
+      "loss": 0.8401,
+      "step": 4302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.086087015512625e-06,
+      "loss": 0.7133,
+      "step": 4303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.085533335417285e-06,
+      "loss": 0.6916,
+      "step": 4304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.084979525191583e-06,
+      "loss": 0.7073,
+      "step": 4305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0844255848809726e-06,
+      "loss": 0.7492,
+      "step": 4306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0838715145309175e-06,
+      "loss": 0.857,
+      "step": 4307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.083317314186892e-06,
+      "loss": 0.6616,
+      "step": 4308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0827629838943814e-06,
+      "loss": 0.7254,
+      "step": 4309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.082208523698883e-06,
+      "loss": 0.7036,
+      "step": 4310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0816539336459015e-06,
+      "loss": 0.6813,
+      "step": 4311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.081099213780954e-06,
+      "loss": 0.7846,
+      "step": 4312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0805443641495705e-06,
+      "loss": 0.6356,
+      "step": 4313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.079989384797286e-06,
+      "loss": 0.767,
+      "step": 4314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.079434275769653e-06,
+      "loss": 0.6985,
+      "step": 4315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0788790371122294e-06,
+      "loss": 0.7295,
+      "step": 4316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.078323668870587e-06,
+      "loss": 0.6904,
+      "step": 4317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.077768171090305e-06,
+      "loss": 0.7266,
+      "step": 4318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.077212543816975e-06,
+      "loss": 0.7458,
+      "step": 4319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.076656787096203e-06,
+      "loss": 0.7815,
+      "step": 4320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.076100900973597e-06,
+      "loss": 0.7295,
+      "step": 4321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0755448854947836e-06,
+      "loss": 0.647,
+      "step": 4322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0749887407053965e-06,
+      "loss": 0.7156,
+      "step": 4323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.07443246665108e-06,
+      "loss": 0.6725,
+      "step": 4324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.073876063377489e-06,
+      "loss": 0.7123,
+      "step": 4325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.073319530930291e-06,
+      "loss": 0.6844,
+      "step": 4326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.072762869355161e-06,
+      "loss": 0.7327,
+      "step": 4327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.072206078697787e-06,
+      "loss": 0.7137,
+      "step": 4328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.071649159003868e-06,
+      "loss": 0.6947,
+      "step": 4329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.07109211031911e-06,
+      "loss": 0.7066,
+      "step": 4330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.070534932689233e-06,
+      "loss": 0.6966,
+      "step": 4331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.069977626159966e-06,
+      "loss": 0.6286,
+      "step": 4332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.069420190777051e-06,
+      "loss": 0.6667,
+      "step": 4333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.068862626586238e-06,
+      "loss": 0.687,
+      "step": 4334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.068304933633287e-06,
+      "loss": 0.6572,
+      "step": 4335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.06774711196397e-06,
+      "loss": 0.8682,
+      "step": 4336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.067189161624071e-06,
+      "loss": 0.8914,
+      "step": 4337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.066631082659381e-06,
+      "loss": 0.7598,
+      "step": 4338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0660728751157054e-06,
+      "loss": 0.6543,
+      "step": 4339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0655145390388566e-06,
+      "loss": 0.8301,
+      "step": 4340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.064956074474661e-06,
+      "loss": 0.6449,
+      "step": 4341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.064397481468951e-06,
+      "loss": 0.6682,
+      "step": 4342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.063838760067575e-06,
+      "loss": 0.7038,
+      "step": 4343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.063279910316389e-06,
+      "loss": 0.7536,
+      "step": 4344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.062720932261259e-06,
+      "loss": 0.8294,
+      "step": 4345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0621618259480625e-06,
+      "loss": 0.7592,
+      "step": 4346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.061602591422686e-06,
+      "loss": 0.8625,
+      "step": 4347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.061043228731031e-06,
+      "loss": 0.6031,
+      "step": 4348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.060483737919003e-06,
+      "loss": 0.6531,
+      "step": 4349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.059924119032524e-06,
+      "loss": 0.6158,
+      "step": 4350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.059364372117523e-06,
+      "loss": 0.6539,
+      "step": 4351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0588044972199386e-06,
+      "loss": 0.842,
+      "step": 4352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.058244494385725e-06,
+      "loss": 0.86,
+      "step": 4353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.057684363660841e-06,
+      "loss": 0.7135,
+      "step": 4354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.05712410509126e-06,
+      "loss": 0.627,
+      "step": 4355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.056563718722965e-06,
+      "loss": 0.7238,
+      "step": 4356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.056003204601946e-06,
+      "loss": 0.773,
+      "step": 4357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.05544256277421e-06,
+      "loss": 0.7097,
+      "step": 4358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.054881793285768e-06,
+      "loss": 0.6911,
+      "step": 4359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.054320896182646e-06,
+      "loss": 0.8972,
+      "step": 4360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.053759871510878e-06,
+      "loss": 0.7133,
+      "step": 4361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.05319871931651e-06,
+      "loss": 0.7,
+      "step": 4362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.052637439645597e-06,
+      "loss": 0.8128,
+      "step": 4363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.052076032544205e-06,
+      "loss": 0.7043,
+      "step": 4364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.051514498058412e-06,
+      "loss": 0.7085,
+      "step": 4365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.050952836234304e-06,
+      "loss": 0.6472,
+      "step": 4366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0503910471179794e-06,
+      "loss": 0.6618,
+      "step": 4367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.049829130755545e-06,
+      "loss": 0.7814,
+      "step": 4368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.049267087193121e-06,
+      "loss": 0.6992,
+      "step": 4369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.048704916476835e-06,
+      "loss": 0.7119,
+      "step": 4370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.048142618652826e-06,
+      "loss": 0.7486,
+      "step": 4371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.047580193767246e-06,
+      "loss": 0.6826,
+      "step": 4372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0470176418662524e-06,
+      "loss": 0.7556,
+      "step": 4373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0464549629960184e-06,
+      "loss": 0.7896,
+      "step": 4374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.045892157202723e-06,
+      "loss": 0.7916,
+      "step": 4375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.045329224532558e-06,
+      "loss": 0.7769,
+      "step": 4376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.044766165031727e-06,
+      "loss": 0.7279,
+      "step": 4377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.04420297874644e-06,
+      "loss": 0.6022,
+      "step": 4378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.043639665722921e-06,
+      "loss": 0.6923,
+      "step": 4379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.043076226007403e-06,
+      "loss": 0.72,
+      "step": 4380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.04251265964613e-06,
+      "loss": 0.8081,
+      "step": 4381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.041948966685355e-06,
+      "loss": 0.8186,
+      "step": 4382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.041385147171342e-06,
+      "loss": 0.8565,
+      "step": 4383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.040821201150367e-06,
+      "loss": 0.7917,
+      "step": 4384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0402571286687135e-06,
+      "loss": 0.6941,
+      "step": 4385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.039692929772678e-06,
+      "loss": 0.6804,
+      "step": 4386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.039128604508566e-06,
+      "loss": 0.644,
+      "step": 4387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0385641529226935e-06,
+      "loss": 0.6432,
+      "step": 4388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.037999575061388e-06,
+      "loss": 0.7597,
+      "step": 4389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.037434870970985e-06,
+      "loss": 0.6213,
+      "step": 4390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.036870040697833e-06,
+      "loss": 0.7593,
+      "step": 4391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0363050842882885e-06,
+      "loss": 0.7421,
+      "step": 4392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.035740001788721e-06,
+      "loss": 0.6549,
+      "step": 4393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.035174793245508e-06,
+      "loss": 0.5813,
+      "step": 4394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.034609458705037e-06,
+      "loss": 0.5785,
+      "step": 4395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.034043998213708e-06,
+      "loss": 0.6904,
+      "step": 4396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.033478411817931e-06,
+      "loss": 0.7589,
+      "step": 4397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.032912699564125e-06,
+      "loss": 0.7575,
+      "step": 4398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.03234686149872e-06,
+      "loss": 0.6923,
+      "step": 4399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.031780897668157e-06,
+      "loss": 0.7059,
+      "step": 4400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.031214808118886e-06,
+      "loss": 0.7603,
+      "step": 4401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.030648592897368e-06,
+      "loss": 0.6169,
+      "step": 4402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.030082252050075e-06,
+      "loss": 0.6435,
+      "step": 4403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.029515785623487e-06,
+      "loss": 0.7174,
+      "step": 4404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.028949193664098e-06,
+      "loss": 0.7526,
+      "step": 4405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.028382476218408e-06,
+      "loss": 0.9044,
+      "step": 4406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.027815633332932e-06,
+      "loss": 0.6424,
+      "step": 4407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.027248665054191e-06,
+      "loss": 0.5992,
+      "step": 4408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.026681571428717e-06,
+      "loss": 0.6777,
+      "step": 4409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.026114352503058e-06,
+      "loss": 0.6721,
+      "step": 4410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.025547008323762e-06,
+      "loss": 0.6839,
+      "step": 4411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.024979538937398e-06,
+      "loss": 0.7214,
+      "step": 4412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0244119443905365e-06,
+      "loss": 0.6939,
+      "step": 4413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.023844224729764e-06,
+      "loss": 0.6674,
+      "step": 4414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.023276380001674e-06,
+      "loss": 0.6163,
+      "step": 4415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.022708410252873e-06,
+      "loss": 0.8262,
+      "step": 4416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.022140315529974e-06,
+      "loss": 0.6398,
+      "step": 4417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.021572095879605e-06,
+      "loss": 0.7595,
+      "step": 4418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0210037513484015e-06,
+      "loss": 0.8388,
+      "step": 4419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.020435281983007e-06,
+      "loss": 0.8447,
+      "step": 4420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0198666878300815e-06,
+      "loss": 0.7163,
+      "step": 4421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.019297968936289e-06,
+      "loss": 0.5659,
+      "step": 4422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0187291253483066e-06,
+      "loss": 0.9046,
+      "step": 4423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.018160157112822e-06,
+      "loss": 0.6586,
+      "step": 4424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0175910642765335e-06,
+      "loss": 0.6513,
+      "step": 4425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.017021846886146e-06,
+      "loss": 0.6508,
+      "step": 4426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.016452504988378e-06,
+      "loss": 0.5916,
+      "step": 4427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.01588303862996e-06,
+      "loss": 0.8389,
+      "step": 4428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0153134478576265e-06,
+      "loss": 0.5981,
+      "step": 4429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.014743732718127e-06,
+      "loss": 0.6417,
+      "step": 4430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.014173893258221e-06,
+      "loss": 0.7492,
+      "step": 4431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.013603929524676e-06,
+      "loss": 0.627,
+      "step": 4432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.013033841564272e-06,
+      "loss": 0.5933,
+      "step": 4433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.012463629423799e-06,
+      "loss": 0.7069,
+      "step": 4434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.011893293150053e-06,
+      "loss": 0.704,
+      "step": 4435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.011322832789847e-06,
+      "loss": 0.7664,
+      "step": 4436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.010752248389999e-06,
+      "loss": 0.6724,
+      "step": 4437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.01018153999734e-06,
+      "loss": 0.709,
+      "step": 4438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0096107076587076e-06,
+      "loss": 0.6659,
+      "step": 4439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.009039751420954e-06,
+      "loss": 0.5941,
+      "step": 4440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.008468671330941e-06,
+      "loss": 0.8392,
+      "step": 4441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.007897467435536e-06,
+      "loss": 0.7139,
+      "step": 4442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.007326139781622e-06,
+      "loss": 0.7817,
+      "step": 4443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.006754688416089e-06,
+      "loss": 0.6553,
+      "step": 4444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.006183113385838e-06,
+      "loss": 0.6785,
+      "step": 4445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0056114147377804e-06,
+      "loss": 0.6993,
+      "step": 4446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0050395925188376e-06,
+      "loss": 0.8366,
+      "step": 4447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.004467646775942e-06,
+      "loss": 0.6243,
+      "step": 4448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.003895577556032e-06,
+      "loss": 0.6762,
+      "step": 4449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.003323384906062e-06,
+      "loss": 0.8513,
+      "step": 4450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.002751068872995e-06,
+      "loss": 0.6474,
+      "step": 4451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.002178629503801e-06,
+      "loss": 0.7336,
+      "step": 4452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.001606066845462e-06,
+      "loss": 0.7307,
+      "step": 4453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.001033380944971e-06,
+      "loss": 0.9034,
+      "step": 4454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.000460571849331e-06,
+      "loss": 0.711,
+      "step": 4455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999887639605553e-06,
+      "loss": 0.8759,
+      "step": 4456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.99931458426066e-06,
+      "loss": 0.648,
+      "step": 4457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.998741405861687e-06,
+      "loss": 0.7423,
+      "step": 4458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.998168104455673e-06,
+      "loss": 0.7035,
+      "step": 4459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.997594680089673e-06,
+      "loss": 0.7978,
+      "step": 4460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9970211328107495e-06,
+      "loss": 0.6773,
+      "step": 4461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.996447462665976e-06,
+      "loss": 0.697,
+      "step": 4462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.995873669702435e-06,
+      "loss": 0.7194,
+      "step": 4463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.995299753967221e-06,
+      "loss": 0.691,
+      "step": 4464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.994725715507436e-06,
+      "loss": 0.7653,
+      "step": 4465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.994151554370193e-06,
+      "loss": 0.7852,
+      "step": 4466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9935772706026175e-06,
+      "loss": 0.7155,
+      "step": 4467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993002864251841e-06,
+      "loss": 0.6257,
+      "step": 4468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.992428335365008e-06,
+      "loss": 0.804,
+      "step": 4469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9918536839892715e-06,
+      "loss": 0.6823,
+      "step": 4470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.991278910171795e-06,
+      "loss": 0.8549,
+      "step": 4471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.990704013959755e-06,
+      "loss": 0.6015,
+      "step": 4472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.990128995400332e-06,
+      "loss": 0.6795,
+      "step": 4473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.989553854540721e-06,
+      "loss": 0.7843,
+      "step": 4474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.988978591428126e-06,
+      "loss": 0.7468,
+      "step": 4475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.988403206109761e-06,
+      "loss": 0.8742,
+      "step": 4476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.987827698632849e-06,
+      "loss": 0.6013,
+      "step": 4477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.987252069044625e-06,
+      "loss": 0.7165,
+      "step": 4478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.986676317392332e-06,
+      "loss": 0.6306,
+      "step": 4479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9861004437232245e-06,
+      "loss": 0.8921,
+      "step": 4480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.985524448084567e-06,
+      "loss": 0.8121,
+      "step": 4481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.984948330523633e-06,
+      "loss": 0.7604,
+      "step": 4482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.984372091087707e-06,
+      "loss": 0.7894,
+      "step": 4483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.983795729824082e-06,
+      "loss": 0.6407,
+      "step": 4484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.983219246780061e-06,
+      "loss": 0.7331,
+      "step": 4485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.982642642002962e-06,
+      "loss": 0.636,
+      "step": 4486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.982065915540107e-06,
+      "loss": 0.6877,
+      "step": 4487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.981489067438828e-06,
+      "loss": 0.7487,
+      "step": 4488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.980912097746471e-06,
+      "loss": 0.7484,
+      "step": 4489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.98033500651039e-06,
+      "loss": 0.6656,
+      "step": 4490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.979757793777947e-06,
+      "loss": 0.6369,
+      "step": 4491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.979180459596519e-06,
+      "loss": 0.8766,
+      "step": 4492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.978603004013488e-06,
+      "loss": 0.7487,
+      "step": 4493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.978025427076247e-06,
+      "loss": 0.599,
+      "step": 4494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.977447728832203e-06,
+      "loss": 0.7374,
+      "step": 4495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.976869909328766e-06,
+      "loss": 0.788,
+      "step": 4496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.976291968613362e-06,
+      "loss": 0.7786,
+      "step": 4497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.975713906733423e-06,
+      "loss": 0.676,
+      "step": 4498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.975135723736394e-06,
+      "loss": 0.7397,
+      "step": 4499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.974557419669729e-06,
+      "loss": 0.6468,
+      "step": 4500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9739789945808895e-06,
+      "loss": 0.6328,
+      "step": 4501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.973400448517351e-06,
+      "loss": 0.7386,
+      "step": 4502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.972821781526595e-06,
+      "loss": 0.7486,
+      "step": 4503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.972242993656115e-06,
+      "loss": 0.5862,
+      "step": 4504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.971664084953416e-06,
+      "loss": 0.7069,
+      "step": 4505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.971085055466009e-06,
+      "loss": 0.696,
+      "step": 4506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.970505905241419e-06,
+      "loss": 0.8036,
+      "step": 4507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.969926634327177e-06,
+      "loss": 0.7096,
+      "step": 4508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9693472427708266e-06,
+      "loss": 0.863,
+      "step": 4509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.968767730619921e-06,
+      "loss": 0.6645,
+      "step": 4510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.968188097922022e-06,
+      "loss": 0.7114,
+      "step": 4511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9676083447247025e-06,
+      "loss": 0.7486,
+      "step": 4512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.967028471075544e-06,
+      "loss": 0.7188,
+      "step": 4513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.966448477022141e-06,
+      "loss": 0.7255,
+      "step": 4514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.965868362612094e-06,
+      "loss": 0.7103,
+      "step": 4515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.965288127893016e-06,
+      "loss": 0.6242,
+      "step": 4516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.964707772912527e-06,
+      "loss": 0.677,
+      "step": 4517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.964127297718261e-06,
+      "loss": 0.6616,
+      "step": 4518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.963546702357859e-06,
+      "loss": 0.7211,
+      "step": 4519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.962965986878971e-06,
+      "loss": 0.7192,
+      "step": 4520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.962385151329261e-06,
+      "loss": 0.7462,
+      "step": 4521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.961804195756399e-06,
+      "loss": 0.694,
+      "step": 4522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9612231202080655e-06,
+      "loss": 0.8333,
+      "step": 4523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.960641924731952e-06,
+      "loss": 0.6497,
+      "step": 4524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.960060609375759e-06,
+      "loss": 0.8181,
+      "step": 4525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9594791741871985e-06,
+      "loss": 0.6272,
+      "step": 4526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.958897619213988e-06,
+      "loss": 0.8109,
+      "step": 4527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.958315944503861e-06,
+      "loss": 0.6276,
+      "step": 4528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.957734150104557e-06,
+      "loss": 0.7531,
+      "step": 4529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.957152236063824e-06,
+      "loss": 0.8014,
+      "step": 4530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9565702024294225e-06,
+      "loss": 0.6507,
+      "step": 4531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9559880492491235e-06,
+      "loss": 0.7873,
+      "step": 4532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.955405776570705e-06,
+      "loss": 0.7726,
+      "step": 4533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.954823384441956e-06,
+      "loss": 0.7321,
+      "step": 4534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.954240872910677e-06,
+      "loss": 0.6796,
+      "step": 4535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.953658242024675e-06,
+      "loss": 0.8036,
+      "step": 4536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.953075491831769e-06,
+      "loss": 0.8611,
+      "step": 4537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.95249262237979e-06,
+      "loss": 0.7127,
+      "step": 4538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.951909633716572e-06,
+      "loss": 0.7857,
+      "step": 4539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.951326525889966e-06,
+      "loss": 0.6432,
+      "step": 4540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.950743298947828e-06,
+      "loss": 0.7465,
+      "step": 4541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.950159952938026e-06,
+      "loss": 0.7061,
+      "step": 4542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.949576487908438e-06,
+      "loss": 0.6646,
+      "step": 4543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.94899290390695e-06,
+      "loss": 0.6742,
+      "step": 4544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.94840920098146e-06,
+      "loss": 0.6089,
+      "step": 4545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.947825379179874e-06,
+      "loss": 0.6973,
+      "step": 4546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.947241438550108e-06,
+      "loss": 0.6624,
+      "step": 4547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.946657379140089e-06,
+      "loss": 0.6354,
+      "step": 4548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.946073200997751e-06,
+      "loss": 0.6912,
+      "step": 4549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.945488904171041e-06,
+      "loss": 0.6588,
+      "step": 4550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.944904488707915e-06,
+      "loss": 0.7686,
+      "step": 4551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.944319954656336e-06,
+      "loss": 0.9165,
+      "step": 4552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.94373530206428e-06,
+      "loss": 0.638,
+      "step": 4553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.943150530979732e-06,
+      "loss": 0.6299,
+      "step": 4554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.942565641450685e-06,
+      "loss": 0.6968,
+      "step": 4555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.941980633525144e-06,
+      "loss": 0.6864,
+      "step": 4556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.941395507251123e-06,
+      "loss": 0.6768,
+      "step": 4557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.940810262676644e-06,
+      "loss": 0.7319,
+      "step": 4558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.940224899849741e-06,
+      "loss": 0.7355,
+      "step": 4559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.939639418818456e-06,
+      "loss": 0.7793,
+      "step": 4560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9390538196308434e-06,
+      "loss": 0.6547,
+      "step": 4561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.938468102334964e-06,
+      "loss": 0.8414,
+      "step": 4562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.937882266978889e-06,
+      "loss": 0.621,
+      "step": 4563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9372963136107015e-06,
+      "loss": 0.7101,
+      "step": 4564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9367102422784905e-06,
+      "loss": 0.8098,
+      "step": 4565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.93612405303036e-06,
+      "loss": 0.6985,
+      "step": 4566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.935537745914419e-06,
+      "loss": 0.6279,
+      "step": 4567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.934951320978788e-06,
+      "loss": 0.7123,
+      "step": 4568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.934364778271598e-06,
+      "loss": 0.7729,
+      "step": 4569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9337781178409865e-06,
+      "loss": 0.8315,
+      "step": 4570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.933191339735104e-06,
+      "loss": 0.6314,
+      "step": 4571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.932604444002109e-06,
+      "loss": 0.6547,
+      "step": 4572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9320174306901715e-06,
+      "loss": 0.6753,
+      "step": 4573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.931430299847469e-06,
+      "loss": 0.8767,
+      "step": 4574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.930843051522188e-06,
+      "loss": 0.6652,
+      "step": 4575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.930255685762529e-06,
+      "loss": 0.6599,
+      "step": 4576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.929668202616696e-06,
+      "loss": 0.6976,
+      "step": 4577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.929080602132909e-06,
+      "loss": 0.7369,
+      "step": 4578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.928492884359391e-06,
+      "loss": 0.7212,
+      "step": 4579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.927905049344381e-06,
+      "loss": 0.7264,
+      "step": 4580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.927317097136123e-06,
+      "loss": 0.6798,
+      "step": 4581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.926729027782873e-06,
+      "loss": 0.6925,
+      "step": 4582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9261408413328965e-06,
+      "loss": 0.6501,
+      "step": 4583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.925552537834466e-06,
+      "loss": 0.8952,
+      "step": 4584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.924964117335868e-06,
+      "loss": 0.8461,
+      "step": 4585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.924375579885396e-06,
+      "loss": 0.6597,
+      "step": 4586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.923786925531351e-06,
+      "loss": 0.8051,
+      "step": 4587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9231981543220485e-06,
+      "loss": 0.7519,
+      "step": 4588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.92260926630581e-06,
+      "loss": 0.8136,
+      "step": 4589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.922020261530969e-06,
+      "loss": 0.6962,
+      "step": 4590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.921431140045865e-06,
+      "loss": 0.704,
+      "step": 4591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.920841901898851e-06,
+      "loss": 0.6955,
+      "step": 4592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.920252547138287e-06,
+      "loss": 0.7658,
+      "step": 4593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.919663075812543e-06,
+      "loss": 0.8482,
+      "step": 4594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9190734879700005e-06,
+      "loss": 0.8121,
+      "step": 4595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9184837836590475e-06,
+      "loss": 0.7885,
+      "step": 4596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.917893962928084e-06,
+      "loss": 0.8175,
+      "step": 4597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.917304025825519e-06,
+      "loss": 0.9091,
+      "step": 4598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.91671397239977e-06,
+      "loss": 0.6977,
+      "step": 4599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.916123802699266e-06,
+      "loss": 0.7614,
+      "step": 4600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9155335167724424e-06,
+      "loss": 0.6518,
+      "step": 4601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.914943114667747e-06,
+      "loss": 0.7307,
+      "step": 4602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.914352596433637e-06,
+      "loss": 0.8273,
+      "step": 4603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.913761962118577e-06,
+      "loss": 0.7249,
+      "step": 4604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.913171211771044e-06,
+      "loss": 0.6557,
+      "step": 4605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.912580345439521e-06,
+      "loss": 0.6339,
+      "step": 4606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.911989363172504e-06,
+      "loss": 0.8029,
+      "step": 4607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.911398265018497e-06,
+      "loss": 0.6988,
+      "step": 4608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9108070510260125e-06,
+      "loss": 0.7765,
+      "step": 4609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9102157212435745e-06,
+      "loss": 0.5789,
+      "step": 4610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9096242757197145e-06,
+      "loss": 0.7539,
+      "step": 4611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.909032714502976e-06,
+      "loss": 0.7459,
+      "step": 4612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.908441037641909e-06,
+      "loss": 0.6271,
+      "step": 4613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.907849245185077e-06,
+      "loss": 0.6353,
+      "step": 4614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.907257337181046e-06,
+      "loss": 0.6824,
+      "step": 4615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9066653136784015e-06,
+      "loss": 0.7318,
+      "step": 4616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9060731747257286e-06,
+      "loss": 0.6361,
+      "step": 4617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9054809203716285e-06,
+      "loss": 0.7126,
+      "step": 4618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.904888550664709e-06,
+      "loss": 0.5458,
+      "step": 4619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.904296065653588e-06,
+      "loss": 0.6438,
+      "step": 4620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.903703465386892e-06,
+      "loss": 0.821,
+      "step": 4621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.90311074991326e-06,
+      "loss": 0.8618,
+      "step": 4622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.902517919281336e-06,
+      "loss": 0.7169,
+      "step": 4623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.901924973539777e-06,
+      "loss": 0.6777,
+      "step": 4624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.901331912737248e-06,
+      "loss": 0.7742,
+      "step": 4625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.900738736922424e-06,
+      "loss": 0.7316,
+      "step": 4626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.900145446143987e-06,
+      "loss": 0.6376,
+      "step": 4627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.899552040450634e-06,
+      "loss": 0.7433,
+      "step": 4628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.898958519891064e-06,
+      "loss": 0.7632,
+      "step": 4629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.898364884513992e-06,
+      "loss": 0.6675,
+      "step": 4630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.897771134368139e-06,
+      "loss": 0.7267,
+      "step": 4631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8971772695022365e-06,
+      "loss": 0.701,
+      "step": 4632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8965832899650246e-06,
+      "loss": 0.7611,
+      "step": 4633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.895989195805253e-06,
+      "loss": 0.6557,
+      "step": 4634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.895394987071682e-06,
+      "loss": 0.7335,
+      "step": 4635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.894800663813081e-06,
+      "loss": 0.7617,
+      "step": 4636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.894206226078226e-06,
+      "loss": 0.6007,
+      "step": 4637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.893611673915907e-06,
+      "loss": 0.7563,
+      "step": 4638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.893017007374918e-06,
+      "loss": 0.7229,
+      "step": 4639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.892422226504069e-06,
+      "loss": 0.6812,
+      "step": 4640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.891827331352173e-06,
+      "loss": 0.6333,
+      "step": 4641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.891232321968058e-06,
+      "loss": 0.8099,
+      "step": 4642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.890637198400555e-06,
+      "loss": 0.6893,
+      "step": 4643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8900419606985106e-06,
+      "loss": 0.6235,
+      "step": 4644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.889446608910777e-06,
+      "loss": 0.7577,
+      "step": 4645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8888511430862174e-06,
+      "loss": 0.7273,
+      "step": 4646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.888255563273703e-06,
+      "loss": 0.6706,
+      "step": 4647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.887659869522115e-06,
+      "loss": 0.9214,
+      "step": 4648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8870640618803455e-06,
+      "loss": 0.8403,
+      "step": 4649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.886468140397294e-06,
+      "loss": 0.6471,
+      "step": 4650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.885872105121869e-06,
+      "loss": 0.7633,
+      "step": 4651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.88527595610299e-06,
+      "loss": 0.7599,
+      "step": 4652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8846796933895845e-06,
+      "loss": 0.6314,
+      "step": 4653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.884083317030592e-06,
+      "loss": 0.6774,
+      "step": 4654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8834868270749556e-06,
+      "loss": 0.7295,
+      "step": 4655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.882890223571634e-06,
+      "loss": 0.7414,
+      "step": 4656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.882293506569592e-06,
+      "loss": 0.7357,
+      "step": 4657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.881696676117804e-06,
+      "loss": 0.6587,
+      "step": 4658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.881099732265255e-06,
+      "loss": 0.8123,
+      "step": 4659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8805026750609356e-06,
+      "loss": 0.7015,
+      "step": 4660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8799055045538515e-06,
+      "loss": 0.8029,
+      "step": 4661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8793082207930134e-06,
+      "loss": 0.6665,
+      "step": 4662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8787108238274426e-06,
+      "loss": 0.8391,
+      "step": 4663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.87811331370617e-06,
+      "loss": 0.6025,
+      "step": 4664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8775156904782345e-06,
+      "loss": 0.8031,
+      "step": 4665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8769179541926845e-06,
+      "loss": 0.769,
+      "step": 4666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8763201048985805e-06,
+      "loss": 0.8304,
+      "step": 4667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8757221426449886e-06,
+      "loss": 0.8171,
+      "step": 4668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.875124067480986e-06,
+      "loss": 0.6808,
+      "step": 4669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.874525879455659e-06,
+      "loss": 0.6501,
+      "step": 4670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.873927578618103e-06,
+      "loss": 0.7779,
+      "step": 4671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.873329165017423e-06,
+      "loss": 0.7331,
+      "step": 4672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.872730638702732e-06,
+      "loss": 0.7612,
+      "step": 4673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.872131999723155e-06,
+      "loss": 0.7999,
+      "step": 4674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.871533248127821e-06,
+      "loss": 0.6354,
+      "step": 4675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8709343839658754e-06,
+      "loss": 0.7748,
+      "step": 4676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.870335407286468e-06,
+      "loss": 0.8288,
+      "step": 4677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.869736318138757e-06,
+      "loss": 0.6833,
+      "step": 4678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8691371165719144e-06,
+      "loss": 0.6686,
+      "step": 4679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.868537802635117e-06,
+      "loss": 0.839,
+      "step": 4680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.867938376377554e-06,
+      "loss": 0.6989,
+      "step": 4681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.867338837848422e-06,
+      "loss": 0.6965,
+      "step": 4682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8667391870969265e-06,
+      "loss": 0.5928,
+      "step": 4683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.866139424172283e-06,
+      "loss": 0.7483,
+      "step": 4684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.865539549123718e-06,
+      "loss": 0.8426,
+      "step": 4685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8649395620004635e-06,
+      "loss": 0.7113,
+      "step": 4686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8643394628517626e-06,
+      "loss": 0.7218,
+      "step": 4687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.863739251726869e-06,
+      "loss": 0.6134,
+      "step": 4688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.863138928675042e-06,
+      "loss": 0.7077,
+      "step": 4689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.862538493745555e-06,
+      "loss": 0.6955,
+      "step": 4690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.861937946987685e-06,
+      "loss": 0.6103,
+      "step": 4691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8613372884507235e-06,
+      "loss": 0.6389,
+      "step": 4692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.860736518183967e-06,
+      "loss": 0.6566,
+      "step": 4693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8601356362367235e-06,
+      "loss": 0.6481,
+      "step": 4694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.85953464265831e-06,
+      "loss": 0.7654,
+      "step": 4695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.85893353749805e-06,
+      "loss": 0.6982,
+      "step": 4696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.858332320805281e-06,
+      "loss": 0.7033,
+      "step": 4697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8577309926293464e-06,
+      "loss": 0.7632,
+      "step": 4698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8571295530195985e-06,
+      "loss": 0.6494,
+      "step": 4699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8565280020254e-06,
+      "loss": 0.5899,
+      "step": 4700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.855926339696122e-06,
+      "loss": 0.7648,
+      "step": 4701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8553245660811455e-06,
+      "loss": 0.7592,
+      "step": 4702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.85472268122986e-06,
+      "loss": 0.6721,
+      "step": 4703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8541206851916645e-06,
+      "loss": 0.8242,
+      "step": 4704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.853518578015967e-06,
+      "loss": 0.7023,
+      "step": 4705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.852916359752184e-06,
+      "loss": 0.8245,
+      "step": 4706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.852314030449743e-06,
+      "loss": 0.6669,
+      "step": 4707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.851711590158077e-06,
+      "loss": 0.7757,
+      "step": 4708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.851109038926634e-06,
+      "loss": 0.7499,
+      "step": 4709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8505063768048635e-06,
+      "loss": 0.8004,
+      "step": 4710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.849903603842231e-06,
+      "loss": 0.6597,
+      "step": 4711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8493007200882056e-06,
+      "loss": 0.7134,
+      "step": 4712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.848697725592271e-06,
+      "loss": 0.7103,
+      "step": 4713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.848094620403916e-06,
+      "loss": 0.7109,
+      "step": 4714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.84749140457264e-06,
+      "loss": 0.642,
+      "step": 4715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.84688807814795e-06,
+      "loss": 0.6186,
+      "step": 4716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.846284641179363e-06,
+      "loss": 0.661,
+      "step": 4717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.845681093716406e-06,
+      "loss": 0.7384,
+      "step": 4718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.845077435808615e-06,
+      "loss": 0.8408,
+      "step": 4719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.844473667505533e-06,
+      "loss": 0.6847,
+      "step": 4720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8438697888567135e-06,
+      "loss": 0.7811,
+      "step": 4721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.843265799911719e-06,
+      "loss": 0.7114,
+      "step": 4722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.842661700720123e-06,
+      "loss": 0.6709,
+      "step": 4723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.842057491331503e-06,
+      "loss": 0.7102,
+      "step": 4724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.84145317179545e-06,
+      "loss": 0.7529,
+      "step": 4725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8408487421615645e-06,
+      "loss": 0.7553,
+      "step": 4726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.840244202479452e-06,
+      "loss": 0.7004,
+      "step": 4727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.839639552798729e-06,
+      "loss": 0.6766,
+      "step": 4728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.839034793169022e-06,
+      "loss": 0.8011,
+      "step": 4729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.838429923639966e-06,
+      "loss": 0.6886,
+      "step": 4730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8378249442612036e-06,
+      "loss": 0.6948,
+      "step": 4731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.837219855082388e-06,
+      "loss": 0.7119,
+      "step": 4732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.836614656153184e-06,
+      "loss": 0.6563,
+      "step": 4733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8360093475232575e-06,
+      "loss": 0.7695,
+      "step": 4734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.835403929242292e-06,
+      "loss": 0.7086,
+      "step": 4735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.834798401359975e-06,
+      "loss": 0.7073,
+      "step": 4736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.834192763926005e-06,
+      "loss": 0.7108,
+      "step": 4737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.833587016990087e-06,
+      "loss": 0.8034,
+      "step": 4738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.832981160601939e-06,
+      "loss": 0.6109,
+      "step": 4739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.832375194811284e-06,
+      "loss": 0.7036,
+      "step": 4740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8317691196678565e-06,
+      "loss": 0.6273,
+      "step": 4741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.831162935221401e-06,
+      "loss": 0.722,
+      "step": 4742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8305566415216655e-06,
+      "loss": 0.7061,
+      "step": 4743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.829950238618414e-06,
+      "loss": 0.8627,
+      "step": 4744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.829343726561415e-06,
+      "loss": 0.6659,
+      "step": 4745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.828737105400446e-06,
+      "loss": 0.8171,
+      "step": 4746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.828130375185297e-06,
+      "loss": 0.7207,
+      "step": 4747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.827523535965762e-06,
+      "loss": 0.6316,
+      "step": 4748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.826916587791648e-06,
+      "loss": 0.7363,
+      "step": 4749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.826309530712769e-06,
+      "loss": 0.6892,
+      "step": 4750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.825702364778947e-06,
+      "loss": 0.7295,
+      "step": 4751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.825095090040016e-06,
+      "loss": 0.6553,
+      "step": 4752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8244877065458175e-06,
+      "loss": 0.6585,
+      "step": 4753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8238802143462005e-06,
+      "loss": 0.8618,
+      "step": 4754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.823272613491025e-06,
+      "loss": 0.6644,
+      "step": 4755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.822664904030158e-06,
+      "loss": 0.812,
+      "step": 4756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.822057086013476e-06,
+      "loss": 0.7151,
+      "step": 4757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.821449159490866e-06,
+      "loss": 0.7284,
+      "step": 4758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.820841124512222e-06,
+      "loss": 0.7522,
+      "step": 4759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8202329811274475e-06,
+      "loss": 0.642,
+      "step": 4760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.819624729386456e-06,
+      "loss": 0.7165,
+      "step": 4761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.819016369339168e-06,
+      "loss": 0.774,
+      "step": 4762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.818407901035514e-06,
+      "loss": 0.9034,
+      "step": 4763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.817799324525433e-06,
+      "loss": 0.6973,
+      "step": 4764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.817190639858873e-06,
+      "loss": 0.7775,
+      "step": 4765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.816581847085792e-06,
+      "loss": 0.6779,
+      "step": 4766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.815972946256153e-06,
+      "loss": 0.6427,
+      "step": 4767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.815363937419933e-06,
+      "loss": 0.7124,
+      "step": 4768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.814754820627116e-06,
+      "loss": 0.646,
+      "step": 4769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.814145595927693e-06,
+      "loss": 0.7367,
+      "step": 4770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.813536263371666e-06,
+      "loss": 0.7178,
+      "step": 4771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8129268230090446e-06,
+      "loss": 0.7003,
+      "step": 4772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.812317274889847e-06,
+      "loss": 0.7901,
+      "step": 4773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8117076190641035e-06,
+      "loss": 0.8073,
+      "step": 4774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.811097855581848e-06,
+      "loss": 0.7334,
+      "step": 4775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8104879844931276e-06,
+      "loss": 0.7856,
+      "step": 4776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.809878005847995e-06,
+      "loss": 0.6975,
+      "step": 4777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8092679196965142e-06,
+      "loss": 0.7231,
+      "step": 4778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.808657726088758e-06,
+      "loss": 0.6969,
+      "step": 4779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.808047425074807e-06,
+      "loss": 0.7231,
+      "step": 4780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.80743701670475e-06,
+      "loss": 0.7204,
+      "step": 4781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.806826501028685e-06,
+      "loss": 0.6158,
+      "step": 4782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.806215878096721e-06,
+      "loss": 0.6503,
+      "step": 4783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.805605147958971e-06,
+      "loss": 0.6928,
+      "step": 4784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.804994310665563e-06,
+      "loss": 0.6765,
+      "step": 4785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8043833662666286e-06,
+      "loss": 0.677,
+      "step": 4786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8037723148123106e-06,
+      "loss": 0.7169,
+      "step": 4787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8031611563527604e-06,
+      "loss": 0.6573,
+      "step": 4788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.802549890938138e-06,
+      "loss": 0.774,
+      "step": 4789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8019385186186124e-06,
+      "loss": 0.7231,
+      "step": 4790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.80132703944436e-06,
+      "loss": 0.625,
+      "step": 4791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8007154534655686e-06,
+      "loss": 0.6769,
+      "step": 4792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.800103760732432e-06,
+      "loss": 0.9219,
+      "step": 4793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.799491961295155e-06,
+      "loss": 0.6244,
+      "step": 4794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.798880055203949e-06,
+      "loss": 0.8165,
+      "step": 4795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7982680425090355e-06,
+      "loss": 0.8091,
+      "step": 4796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.797655923260646e-06,
+      "loss": 0.6119,
+      "step": 4797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.797043697509017e-06,
+      "loss": 0.7933,
+      "step": 4798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.796431365304398e-06,
+      "loss": 0.6308,
+      "step": 4799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7958189266970446e-06,
+      "loss": 0.7784,
+      "step": 4800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7952063817372213e-06,
+      "loss": 0.7149,
+      "step": 4801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7945937304752032e-06,
+      "loss": 0.6144,
+      "step": 4802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7939809729612715e-06,
+      "loss": 0.7874,
+      "step": 4803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7933681092457177e-06,
+      "loss": 0.6948,
+      "step": 4804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.792755139378842e-06,
+      "loss": 0.8451,
+      "step": 4805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7921420634109523e-06,
+      "loss": 0.7552,
+      "step": 4806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.791528881392367e-06,
+      "loss": 0.6467,
+      "step": 4807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7909155933734108e-06,
+      "loss": 0.6705,
+      "step": 4808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7903021994044205e-06,
+      "loss": 0.6844,
+      "step": 4809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7896886995357374e-06,
+      "loss": 0.6581,
+      "step": 4810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7890750938177146e-06,
+      "loss": 0.6912,
+      "step": 4811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.788461382300712e-06,
+      "loss": 0.6267,
+      "step": 4812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.787847565035101e-06,
+      "loss": 0.7516,
+      "step": 4813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7872336420712585e-06,
+      "loss": 0.7648,
+      "step": 4814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.786619613459571e-06,
+      "loss": 0.6498,
+      "step": 4815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.786005479250434e-06,
+      "loss": 0.7478,
+      "step": 4816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7853912394942525e-06,
+      "loss": 0.7753,
+      "step": 4817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.78477689424144e-06,
+      "loss": 0.6187,
+      "step": 4818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7841624435424157e-06,
+      "loss": 0.7249,
+      "step": 4819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7835478874476116e-06,
+      "loss": 0.6215,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.782933226007466e-06,
+      "loss": 0.7262,
+      "step": 4821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.782318459272426e-06,
+      "loss": 0.756,
+      "step": 4822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7817035872929486e-06,
+      "loss": 0.711,
+      "step": 4823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.781088610119497e-06,
+      "loss": 0.7115,
+      "step": 4824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.780473527802546e-06,
+      "loss": 0.7267,
+      "step": 4825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7798583403925766e-06,
+      "loss": 0.7719,
+      "step": 4826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.77924304794008e-06,
+      "loss": 0.6931,
+      "step": 4827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.778627650495556e-06,
+      "loss": 0.7297,
+      "step": 4828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7780121481095113e-06,
+      "loss": 0.8444,
+      "step": 4829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.777396540832463e-06,
+      "loss": 0.8071,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7767808287149364e-06,
+      "loss": 0.8302,
+      "step": 4831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7761650118074644e-06,
+      "loss": 0.6482,
+      "step": 4832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7755490901605895e-06,
+      "loss": 0.7154,
+      "step": 4833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7749330638248634e-06,
+      "loss": 0.7641,
+      "step": 4834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7743169328508446e-06,
+      "loss": 0.8254,
+      "step": 4835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.773700697289101e-06,
+      "loss": 0.6973,
+      "step": 4836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.773084357190211e-06,
+      "loss": 0.7228,
+      "step": 4837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.772467912604757e-06,
+      "loss": 0.6948,
+      "step": 4838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.771851363583335e-06,
+      "loss": 0.7739,
+      "step": 4839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.771234710176548e-06,
+      "loss": 0.6147,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7706179524350046e-06,
+      "loss": 0.687,
+      "step": 4841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7700010904093255e-06,
+      "loss": 0.7408,
+      "step": 4842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.769384124150139e-06,
+      "loss": 0.6108,
+      "step": 4843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.768767053708081e-06,
+      "loss": 0.6879,
+      "step": 4844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7681498791337966e-06,
+      "loss": 0.7803,
+      "step": 4845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7675326004779405e-06,
+      "loss": 0.8395,
+      "step": 4846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.766915217791175e-06,
+      "loss": 0.7582,
+      "step": 4847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7662977311241693e-06,
+      "loss": 0.7127,
+      "step": 4848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7656801405276043e-06,
+      "loss": 0.7487,
+      "step": 4849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.765062446052168e-06,
+      "loss": 0.6676,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7644446477485554e-06,
+      "loss": 0.6297,
+      "step": 4851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7638267456674715e-06,
+      "loss": 0.6833,
+      "step": 4852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7632087398596307e-06,
+      "loss": 0.6944,
+      "step": 4853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7625906303757543e-06,
+      "loss": 0.6939,
+      "step": 4854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7619724172665735e-06,
+      "loss": 0.8505,
+      "step": 4855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.761354100582827e-06,
+      "loss": 0.683,
+      "step": 4856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7607356803752615e-06,
+      "loss": 0.7698,
+      "step": 4857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7601171566946324e-06,
+      "loss": 0.6943,
+      "step": 4858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7594985295917074e-06,
+      "loss": 0.6363,
+      "step": 4859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7588797991172554e-06,
+      "loss": 0.6702,
+      "step": 4860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.758260965322061e-06,
+      "loss": 0.7115,
+      "step": 4861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7576420282569117e-06,
+      "loss": 0.6365,
+      "step": 4862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.757022987972607e-06,
+      "loss": 0.7728,
+      "step": 4863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7564038445199534e-06,
+      "loss": 0.7788,
+      "step": 4864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7557845979497666e-06,
+      "loss": 0.7831,
+      "step": 4865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.755165248312871e-06,
+      "loss": 0.7163,
+      "step": 4866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7545457956600974e-06,
+      "loss": 0.7924,
+      "step": 4867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.753926240042287e-06,
+      "loss": 0.8194,
+      "step": 4868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.753306581510289e-06,
+      "loss": 0.6525,
+      "step": 4869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.752686820114961e-06,
+      "loss": 0.742,
+      "step": 4870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7520669559071695e-06,
+      "loss": 0.6658,
+      "step": 4871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.751446988937788e-06,
+      "loss": 0.6881,
+      "step": 4872
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.8202974200248718,
+      "eval_runtime": 437.3788,
+      "eval_samples_per_second": 6.857,
+      "eval_steps_per_second": 0.857,
+      "step": 4872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7508269192576997e-06,
+      "loss": 0.7262,
+      "step": 4873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.750206746917796e-06,
+      "loss": 0.7234,
+      "step": 4874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.749586471968978e-06,
+      "loss": 0.6674,
+      "step": 4875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7489660944621514e-06,
+      "loss": 0.7265,
+      "step": 4876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7483456144482344e-06,
+      "loss": 0.9388,
+      "step": 4877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7477250319781524e-06,
+      "loss": 0.7594,
+      "step": 4878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.747104347102837e-06,
+      "loss": 0.6481,
+      "step": 4879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7464835598732315e-06,
+      "loss": 0.6668,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7458626703402852e-06,
+      "loss": 0.69,
+      "step": 4881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7452416785549573e-06,
+      "loss": 0.6504,
+      "step": 4882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.744620584568215e-06,
+      "loss": 0.6764,
+      "step": 4883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7439993884310326e-06,
+      "loss": 0.6937,
+      "step": 4884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.743378090194396e-06,
+      "loss": 0.6403,
+      "step": 4885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.742756689909295e-06,
+      "loss": 0.7157,
+      "step": 4886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.742135187626732e-06,
+      "loss": 0.7039,
+      "step": 4887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.741513583397714e-06,
+      "loss": 0.8438,
+      "step": 4888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7408918772732606e-06,
+      "loss": 0.6783,
+      "step": 4889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7402700693043946e-06,
+      "loss": 0.8087,
+      "step": 4890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7396481595421524e-06,
+      "loss": 0.7518,
+      "step": 4891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7390261480375758e-06,
+      "loss": 0.6898,
+      "step": 4892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7384040348417137e-06,
+      "loss": 0.6527,
+      "step": 4893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7377818200056287e-06,
+      "loss": 0.7624,
+      "step": 4894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.737159503580385e-06,
+      "loss": 0.6954,
+      "step": 4895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.736537085617059e-06,
+      "loss": 0.6467,
+      "step": 4896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.735914566166736e-06,
+      "loss": 0.7196,
+      "step": 4897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7352919452805076e-06,
+      "loss": 0.6497,
+      "step": 4898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.734669223009475e-06,
+      "loss": 0.6697,
+      "step": 4899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7340463994047454e-06,
+      "loss": 0.6985,
+      "step": 4900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.733423474517438e-06,
+      "loss": 0.7005,
+      "step": 4901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.732800448398678e-06,
+      "loss": 0.6809,
+      "step": 4902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7321773210995995e-06,
+      "loss": 0.7925,
+      "step": 4903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7315540926713447e-06,
+      "loss": 0.7301,
+      "step": 4904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.730930763165063e-06,
+      "loss": 0.6976,
+      "step": 4905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7303073326319154e-06,
+      "loss": 0.7061,
+      "step": 4906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7296838011230676e-06,
+      "loss": 0.7054,
+      "step": 4907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.729060168689696e-06,
+      "loss": 0.7603,
+      "step": 4908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.728436435382983e-06,
+      "loss": 0.7952,
+      "step": 4909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7278126012541217e-06,
+      "loss": 0.6097,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7271886663543116e-06,
+      "loss": 0.7295,
+      "step": 4911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.726564630734762e-06,
+      "loss": 0.7451,
+      "step": 4912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.72594049444669e-06,
+      "loss": 0.6527,
+      "step": 4913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7253162575413194e-06,
+      "loss": 0.8062,
+      "step": 4914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7246919200698845e-06,
+      "loss": 0.6535,
+      "step": 4915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.724067482083627e-06,
+      "loss": 0.6301,
+      "step": 4916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.723442943633796e-06,
+      "loss": 0.754,
+      "step": 4917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7228183047716505e-06,
+      "loss": 0.8414,
+      "step": 4918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7221935655484563e-06,
+      "loss": 0.6021,
+      "step": 4919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7215687260154875e-06,
+      "loss": 0.8464,
+      "step": 4920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.720943786224027e-06,
+      "loss": 0.7086,
+      "step": 4921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7203187462253677e-06,
+      "loss": 0.7347,
+      "step": 4922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7196936060708065e-06,
+      "loss": 0.6476,
+      "step": 4923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7190683658116515e-06,
+      "loss": 0.7651,
+      "step": 4924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.71844302549922e-06,
+      "loss": 0.8481,
+      "step": 4925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.717817585184833e-06,
+      "loss": 0.7699,
+      "step": 4926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.717192044919826e-06,
+      "loss": 0.7132,
+      "step": 4927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7165664047555363e-06,
+      "loss": 0.6893,
+      "step": 4928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7159406647433133e-06,
+      "loss": 0.7658,
+      "step": 4929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.715314824934514e-06,
+      "loss": 0.8182,
+      "step": 4930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7146888853805037e-06,
+      "loss": 0.6693,
+      "step": 4931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7140628461326554e-06,
+      "loss": 0.6363,
+      "step": 4932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.71343670724235e-06,
+      "loss": 0.758,
+      "step": 4933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7128104687609766e-06,
+      "loss": 0.7046,
+      "step": 4934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.712184130739934e-06,
+      "loss": 0.6954,
+      "step": 4935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.711557693230627e-06,
+      "loss": 0.7531,
+      "step": 4936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7109311562844695e-06,
+      "loss": 0.7066,
+      "step": 4937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7103045199528844e-06,
+      "loss": 0.8301,
+      "step": 4938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7096777842873006e-06,
+      "loss": 0.8611,
+      "step": 4939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.709050949339158e-06,
+      "loss": 0.7616,
+      "step": 4940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.708424015159902e-06,
+      "loss": 0.7187,
+      "step": 4941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.707796981800989e-06,
+      "loss": 0.6794,
+      "step": 4942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7071698493138795e-06,
+      "loss": 0.9154,
+      "step": 4943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7065426177500476e-06,
+      "loss": 0.5896,
+      "step": 4944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.705915287160969e-06,
+      "loss": 0.6336,
+      "step": 4945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7052878575981334e-06,
+      "loss": 0.7392,
+      "step": 4946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7046603291130347e-06,
+      "loss": 0.8413,
+      "step": 4947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.704032701757177e-06,
+      "loss": 0.7012,
+      "step": 4948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.703404975582072e-06,
+      "loss": 0.6962,
+      "step": 4949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.702777150639239e-06,
+      "loss": 0.6311,
+      "step": 4950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7021492269802074e-06,
+      "loss": 0.6988,
+      "step": 4951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7015212046565106e-06,
+      "loss": 0.7185,
+      "step": 4952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7008930837196943e-06,
+      "loss": 0.7428,
+      "step": 4953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7002648642213106e-06,
+      "loss": 0.7066,
+      "step": 4954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.699636546212919e-06,
+      "loss": 0.6584,
+      "step": 4955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.699008129746089e-06,
+      "loss": 0.7719,
+      "step": 4956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.698379614872395e-06,
+      "loss": 0.5812,
+      "step": 4957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6977510016434228e-06,
+      "loss": 0.8568,
+      "step": 4958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.697122290110765e-06,
+      "loss": 0.6526,
+      "step": 4959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6964934803260215e-06,
+      "loss": 0.7901,
+      "step": 4960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6958645723408015e-06,
+      "loss": 0.714,
+      "step": 4961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.695235566206722e-06,
+      "loss": 0.6857,
+      "step": 4962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6946064619754073e-06,
+      "loss": 0.6478,
+      "step": 4963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6939772596984895e-06,
+      "loss": 0.665,
+      "step": 4964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6933479594276113e-06,
+      "loss": 0.6184,
+      "step": 4965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6927185612144193e-06,
+      "loss": 0.7377,
+      "step": 4966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6920890651105723e-06,
+      "loss": 0.6747,
+      "step": 4967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6914594711677347e-06,
+      "loss": 0.7866,
+      "step": 4968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.690829779437579e-06,
+      "loss": 0.7216,
+      "step": 4969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.690199989971788e-06,
+      "loss": 0.6735,
+      "step": 4970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.689570102822048e-06,
+      "loss": 0.8314,
+      "step": 4971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6889401180400587e-06,
+      "loss": 0.7536,
+      "step": 4972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.688310035677524e-06,
+      "loss": 0.6701,
+      "step": 4973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.687679855786156e-06,
+      "loss": 0.7509,
+      "step": 4974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.687049578417678e-06,
+      "loss": 0.605,
+      "step": 4975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.686419203623817e-06,
+      "loss": 0.8247,
+      "step": 4976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6857887314563117e-06,
+      "loss": 0.8093,
+      "step": 4977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6851581619669064e-06,
+      "loss": 0.7483,
+      "step": 4978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.684527495207354e-06,
+      "loss": 0.7747,
+      "step": 4979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6838967312294164e-06,
+      "loss": 0.698,
+      "step": 4980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.683265870084862e-06,
+      "loss": 0.7853,
+      "step": 4981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.682634911825469e-06,
+      "loss": 0.6218,
+      "step": 4982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.68200385650302e-06,
+      "loss": 0.7664,
+      "step": 4983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6813727041693097e-06,
+      "loss": 0.7625,
+      "step": 4984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6807414548761387e-06,
+      "loss": 0.6824,
+      "step": 4985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6801101086753156e-06,
+      "loss": 0.7158,
+      "step": 4986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6794786656186577e-06,
+      "loss": 0.7735,
+      "step": 4987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6788471257579894e-06,
+      "loss": 0.7502,
+      "step": 4988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6782154891451437e-06,
+      "loss": 0.692,
+      "step": 4989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6775837558319606e-06,
+      "loss": 0.7774,
+      "step": 4990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6769519258702896e-06,
+      "loss": 0.7523,
+      "step": 4991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.676319999311987e-06,
+      "loss": 0.6381,
+      "step": 4992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.675687976208917e-06,
+      "loss": 0.8309,
+      "step": 4993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6750558566129526e-06,
+      "loss": 0.6548,
+      "step": 4994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6744236405759724e-06,
+      "loss": 0.6232,
+      "step": 4995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.673791328149866e-06,
+      "loss": 0.744,
+      "step": 4996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6731589193865303e-06,
+      "loss": 0.6702,
+      "step": 4997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6725264143378673e-06,
+      "loss": 0.6679,
+      "step": 4998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6718938130557914e-06,
+      "loss": 0.7025,
+      "step": 4999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6712611155922196e-06,
+      "loss": 0.6535,
+      "step": 5000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6706283219990823e-06,
+      "loss": 0.7336,
+      "step": 5001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6699954323283125e-06,
+      "loss": 0.657,
+      "step": 5002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.669362446631857e-06,
+      "loss": 0.8308,
+      "step": 5003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.668729364961664e-06,
+      "loss": 0.6468,
+      "step": 5004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6680961873696942e-06,
+      "loss": 0.719,
+      "step": 5005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6674629139079144e-06,
+      "loss": 0.8016,
+      "step": 5006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6668295446282996e-06,
+      "loss": 0.6651,
+      "step": 5007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.666196079582834e-06,
+      "loss": 0.8363,
+      "step": 5008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.665562518823506e-06,
+      "loss": 0.6701,
+      "step": 5009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6649288624023162e-06,
+      "loss": 0.652,
+      "step": 5010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6642951103712703e-06,
+      "loss": 0.7072,
+      "step": 5011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.663661262782382e-06,
+      "loss": 0.6471,
+      "step": 5012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6630273196876747e-06,
+      "loss": 0.7103,
+      "step": 5013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.662393281139177e-06,
+      "loss": 0.6494,
+      "step": 5014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.661759147188927e-06,
+      "loss": 0.6094,
+      "step": 5015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.661124917888971e-06,
+      "loss": 0.6222,
+      "step": 5016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.660490593291362e-06,
+      "loss": 0.766,
+      "step": 5017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.659856173448162e-06,
+      "loss": 0.6868,
+      "step": 5018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6592216584114387e-06,
+      "loss": 0.7112,
+      "step": 5019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6585870482332704e-06,
+      "loss": 0.8145,
+      "step": 5020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6579523429657405e-06,
+      "loss": 0.6693,
+      "step": 5021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.657317542660943e-06,
+      "loss": 0.7534,
+      "step": 5022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.656682647370977e-06,
+      "loss": 0.63,
+      "step": 5023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.656047657147951e-06,
+      "loss": 0.6937,
+      "step": 5024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6554125720439805e-06,
+      "loss": 0.5854,
+      "step": 5025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6547773921111895e-06,
+      "loss": 0.5878,
+      "step": 5026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.654142117401711e-06,
+      "loss": 0.6356,
+      "step": 5027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6535067479676816e-06,
+      "loss": 0.6172,
+      "step": 5028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6528712838612503e-06,
+      "loss": 0.7804,
+      "step": 5029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.652235725134571e-06,
+      "loss": 0.695,
+      "step": 5030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6516000718398063e-06,
+      "loss": 0.6789,
+      "step": 5031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.650964324029127e-06,
+      "loss": 0.771,
+      "step": 5032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.650328481754711e-06,
+      "loss": 0.6294,
+      "step": 5033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.649692545068744e-06,
+      "loss": 0.7529,
+      "step": 5034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6490565140234187e-06,
+      "loss": 0.9312,
+      "step": 5035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.648420388670938e-06,
+      "loss": 0.6031,
+      "step": 5036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6477841690635102e-06,
+      "loss": 0.7628,
+      "step": 5037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6471478552533525e-06,
+      "loss": 0.707,
+      "step": 5038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6465114472926887e-06,
+      "loss": 0.7326,
+      "step": 5039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6458749452337516e-06,
+      "loss": 0.6544,
+      "step": 5040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.645238349128782e-06,
+      "loss": 0.701,
+      "step": 5041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.644601659030026e-06,
+      "loss": 0.7028,
+      "step": 5042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6439648749897394e-06,
+      "loss": 0.6854,
+      "step": 5043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6433279970601852e-06,
+      "loss": 0.7264,
+      "step": 5044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.642691025293636e-06,
+      "loss": 0.7437,
+      "step": 5045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6420539597423687e-06,
+      "loss": 0.8988,
+      "step": 5046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.641416800458669e-06,
+      "loss": 0.7879,
+      "step": 5047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.640779547494833e-06,
+      "loss": 0.7211,
+      "step": 5048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6401422009031606e-06,
+      "loss": 0.7118,
+      "step": 5049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6395047607359615e-06,
+      "loss": 0.6621,
+      "step": 5050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.638867227045552e-06,
+      "loss": 0.6883,
+      "step": 5051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.638229599884258e-06,
+      "loss": 0.7129,
+      "step": 5052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6375918793044113e-06,
+      "loss": 0.8432,
+      "step": 5053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.636954065358352e-06,
+      "loss": 0.7402,
+      "step": 5054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6363161580984275e-06,
+      "loss": 0.828,
+      "step": 5055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6356781575769934e-06,
+      "loss": 0.8102,
+      "step": 5056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.635040063846412e-06,
+      "loss": 0.5993,
+      "step": 5057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6344018769590555e-06,
+      "loss": 0.6399,
+      "step": 5058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6337635969673e-06,
+      "loss": 0.7264,
+      "step": 5059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6331252239235337e-06,
+      "loss": 0.6592,
+      "step": 5060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.632486757880148e-06,
+      "loss": 0.6871,
+      "step": 5061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6318481988895458e-06,
+      "loss": 0.825,
+      "step": 5062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.631209547004134e-06,
+      "loss": 0.7023,
+      "step": 5063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6305708022763316e-06,
+      "loss": 0.6098,
+      "step": 5064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.629931964758561e-06,
+      "loss": 0.6703,
+      "step": 5065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6292930345032533e-06,
+      "loss": 0.7743,
+      "step": 5066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6286540115628494e-06,
+      "loss": 0.6172,
+      "step": 5067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6280148959897954e-06,
+      "loss": 0.6307,
+      "step": 5068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.627375687836545e-06,
+      "loss": 0.7728,
+      "step": 5069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.626736387155562e-06,
+      "loss": 0.6596,
+      "step": 5070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.626096993999315e-06,
+      "loss": 0.6439,
+      "step": 5071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6254575084202807e-06,
+      "loss": 0.748,
+      "step": 5072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6248179304709453e-06,
+      "loss": 0.6707,
+      "step": 5073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6241782602038002e-06,
+      "loss": 0.6353,
+      "step": 5074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6235384976713468e-06,
+      "loss": 0.7719,
+      "step": 5075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6228986429260904e-06,
+      "loss": 0.6798,
+      "step": 5076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6222586960205487e-06,
+      "loss": 0.5733,
+      "step": 5077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6216186570072426e-06,
+      "loss": 0.7813,
+      "step": 5078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.620978525938703e-06,
+      "loss": 0.6843,
+      "step": 5079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6203383028674677e-06,
+      "loss": 0.7772,
+      "step": 5080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.619697987846082e-06,
+      "loss": 0.5779,
+      "step": 5081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.619057580927099e-06,
+      "loss": 0.7279,
+      "step": 5082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6184170821630787e-06,
+      "loss": 0.7191,
+      "step": 5083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6177764916065906e-06,
+      "loss": 0.6513,
+      "step": 5084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6171358093102083e-06,
+      "loss": 0.7555,
+      "step": 5085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6164950353265154e-06,
+      "loss": 0.7178,
+      "step": 5086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.615854169708104e-06,
+      "loss": 0.659,
+      "step": 5087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.61521321250757e-06,
+      "loss": 0.7414,
+      "step": 5088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.614572163777521e-06,
+      "loss": 0.6927,
+      "step": 5089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6139310235705684e-06,
+      "loss": 0.7558,
+      "step": 5090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.613289791939334e-06,
+      "loss": 0.773,
+      "step": 5091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6126484689364453e-06,
+      "loss": 0.7473,
+      "step": 5092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6120070546145388e-06,
+      "loss": 0.6668,
+      "step": 5093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6113655490262573e-06,
+      "loss": 0.7098,
+      "step": 5094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6107239522242508e-06,
+      "loss": 0.6517,
+      "step": 5095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.610082264261179e-06,
+      "loss": 0.8126,
+      "step": 5096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6094404851897052e-06,
+      "loss": 0.7026,
+      "step": 5097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.608798615062505e-06,
+      "loss": 0.665,
+      "step": 5098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6081566539322566e-06,
+      "loss": 0.7276,
+      "step": 5099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6075146018516494e-06,
+      "loss": 0.8115,
+      "step": 5100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.606872458873379e-06,
+      "loss": 0.7817,
+      "step": 5101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6062302250501475e-06,
+      "loss": 0.7282,
+      "step": 5102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.605587900434667e-06,
+      "loss": 0.5648,
+      "step": 5103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.604945485079653e-06,
+      "loss": 0.6577,
+      "step": 5104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.604302979037832e-06,
+      "loss": 0.7429,
+      "step": 5105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6036603823619376e-06,
+      "loss": 0.8533,
+      "step": 5106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6030176951047085e-06,
+      "loss": 0.7236,
+      "step": 5107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.602374917318894e-06,
+      "loss": 0.7735,
+      "step": 5108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.601732049057247e-06,
+      "loss": 0.6932,
+      "step": 5109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6010890903725316e-06,
+      "loss": 0.7031,
+      "step": 5110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.600446041317517e-06,
+      "loss": 0.8057,
+      "step": 5111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5998029019449805e-06,
+      "loss": 0.7914,
+      "step": 5112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5991596723077076e-06,
+      "loss": 0.6746,
+      "step": 5113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.59851635245849e-06,
+      "loss": 0.7045,
+      "step": 5114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5978729424501275e-06,
+      "loss": 0.7512,
+      "step": 5115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.597229442335426e-06,
+      "loss": 0.8996,
+      "step": 5116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5965858521672013e-06,
+      "loss": 0.6891,
+      "step": 5117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.595942171998274e-06,
+      "loss": 0.6588,
+      "step": 5118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5952984018814734e-06,
+      "loss": 0.7766,
+      "step": 5119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5946545418696367e-06,
+      "loss": 0.83,
+      "step": 5120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5940105920156067e-06,
+      "loss": 0.7409,
+      "step": 5121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5933665523722362e-06,
+      "loss": 0.7775,
+      "step": 5122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5927224229923823e-06,
+      "loss": 0.5857,
+      "step": 5123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5920782039289115e-06,
+      "loss": 0.9332,
+      "step": 5124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.591433895234698e-06,
+      "loss": 0.6509,
+      "step": 5125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.590789496962621e-06,
+      "loss": 0.7033,
+      "step": 5126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.59014500916557e-06,
+      "loss": 0.6366,
+      "step": 5127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.589500431896439e-06,
+      "loss": 0.7311,
+      "step": 5128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5888557652081317e-06,
+      "loss": 0.6101,
+      "step": 5129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5882110091535584e-06,
+      "loss": 0.667,
+      "step": 5130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5875661637856353e-06,
+      "loss": 0.6818,
+      "step": 5131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5869212291572887e-06,
+      "loss": 0.7406,
+      "step": 5132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5862762053214497e-06,
+      "loss": 0.7435,
+      "step": 5133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.585631092331058e-06,
+      "loss": 0.641,
+      "step": 5134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.58498589023906e-06,
+      "loss": 0.7835,
+      "step": 5135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5843405990984105e-06,
+      "loss": 0.7319,
+      "step": 5136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5836952189620692e-06,
+      "loss": 0.8724,
+      "step": 5137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.583049749883007e-06,
+      "loss": 0.6902,
+      "step": 5138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.582404191914198e-06,
+      "loss": 0.7504,
+      "step": 5139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.581758545108626e-06,
+      "loss": 0.7447,
+      "step": 5140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5811128095192825e-06,
+      "loss": 0.8303,
+      "step": 5141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5804669851991635e-06,
+      "loss": 0.7626,
+      "step": 5142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5798210722012756e-06,
+      "loss": 0.7302,
+      "step": 5143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.579175070578631e-06,
+      "loss": 0.6666,
+      "step": 5144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5785289803842482e-06,
+      "loss": 0.5903,
+      "step": 5145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5778828016711557e-06,
+      "loss": 0.6319,
+      "step": 5146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5772365344923866e-06,
+      "loss": 0.7777,
+      "step": 5147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5765901789009825e-06,
+      "loss": 0.7662,
+      "step": 5148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5759437349499927e-06,
+      "loss": 0.8785,
+      "step": 5149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.575297202692472e-06,
+      "loss": 0.7563,
+      "step": 5150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5746505821814852e-06,
+      "loss": 0.715,
+      "step": 5151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.574003873470101e-06,
+      "loss": 0.8086,
+      "step": 5152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.573357076611399e-06,
+      "loss": 0.7769,
+      "step": 5153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5727101916584626e-06,
+      "loss": 0.7207,
+      "step": 5154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5720632186643845e-06,
+      "loss": 0.7652,
+      "step": 5155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5714161576822637e-06,
+      "loss": 0.7076,
+      "step": 5156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.570769008765207e-06,
+      "loss": 0.661,
+      "step": 5157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5701217719663284e-06,
+      "loss": 0.7168,
+      "step": 5158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.569474447338749e-06,
+      "loss": 0.6274,
+      "step": 5159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.568827034935597e-06,
+      "loss": 0.6539,
+      "step": 5160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5681795348100075e-06,
+      "loss": 0.6613,
+      "step": 5161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.567531947015123e-06,
+      "loss": 0.6659,
+      "step": 5162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5668842716040945e-06,
+      "loss": 0.6344,
+      "step": 5163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5662365086300775e-06,
+      "loss": 0.7166,
+      "step": 5164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.565588658146238e-06,
+      "loss": 0.6595,
+      "step": 5165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.564940720205745e-06,
+      "loss": 0.6624,
+      "step": 5166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.564292694861779e-06,
+      "loss": 0.7669,
+      "step": 5167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5636445821675247e-06,
+      "loss": 0.6965,
+      "step": 5168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.562996382176176e-06,
+      "loss": 0.8983,
+      "step": 5169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5623480949409324e-06,
+      "loss": 0.7854,
+      "step": 5170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5616997205150007e-06,
+      "loss": 0.729,
+      "step": 5171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5610512589515967e-06,
+      "loss": 0.777,
+      "step": 5172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.56040271030394e-06,
+      "loss": 0.7378,
+      "step": 5173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5597540746252613e-06,
+      "loss": 0.8231,
+      "step": 5174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5591053519687948e-06,
+      "loss": 0.6278,
+      "step": 5175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5584565423877836e-06,
+      "loss": 0.6689,
+      "step": 5176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5578076459354786e-06,
+      "loss": 0.7982,
+      "step": 5177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.557158662665137e-06,
+      "loss": 0.6451,
+      "step": 5178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.556509592630023e-06,
+      "loss": 0.6331,
+      "step": 5179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.555860435883408e-06,
+      "loss": 0.6952,
+      "step": 5180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5552111924785705e-06,
+      "loss": 0.7537,
+      "step": 5181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5545618624687973e-06,
+      "loss": 0.6836,
+      "step": 5182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.553912445907379e-06,
+      "loss": 0.7274,
+      "step": 5183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5532629428476174e-06,
+      "loss": 0.7867,
+      "step": 5184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5526133533428185e-06,
+      "loss": 0.7625,
+      "step": 5185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5519636774462973e-06,
+      "loss": 0.7438,
+      "step": 5186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.551313915211374e-06,
+      "loss": 0.8373,
+      "step": 5187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5506640666913776e-06,
+      "loss": 0.694,
+      "step": 5188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.550014131939644e-06,
+      "loss": 0.7122,
+      "step": 5189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5493641110095145e-06,
+      "loss": 0.8335,
+      "step": 5190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5487140039543394e-06,
+      "loss": 0.7503,
+      "step": 5191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5480638108274747e-06,
+      "loss": 0.6498,
+      "step": 5192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.547413531682285e-06,
+      "loss": 0.7734,
+      "step": 5193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5467631665721395e-06,
+      "loss": 0.6176,
+      "step": 5194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5461127155504167e-06,
+      "loss": 0.7233,
+      "step": 5195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5454621786705024e-06,
+      "loss": 0.742,
+      "step": 5196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5448115559857876e-06,
+      "loss": 0.7069,
+      "step": 5197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5441608475496713e-06,
+      "loss": 0.6439,
+      "step": 5198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.543510053415559e-06,
+      "loss": 0.7777,
+      "step": 5199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5428591736368646e-06,
+      "loss": 0.6831,
+      "step": 5200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.542208208267008e-06,
+      "loss": 0.7384,
+      "step": 5201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5415571573594156e-06,
+      "loss": 0.8031,
+      "step": 5202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5409060209675218e-06,
+      "loss": 0.7542,
+      "step": 5203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5402547991447667e-06,
+      "loss": 0.6725,
+      "step": 5204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5396034919446006e-06,
+      "loss": 0.6647,
+      "step": 5205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.538952099420477e-06,
+      "loss": 0.6635,
+      "step": 5206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5383006216258575e-06,
+      "loss": 0.6659,
+      "step": 5207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5376490586142133e-06,
+      "loss": 0.846,
+      "step": 5208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5369974104390188e-06,
+      "loss": 0.8433,
+      "step": 5209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.536345677153758e-06,
+      "loss": 0.7669,
+      "step": 5210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.53569385881192e-06,
+      "loss": 0.7207,
+      "step": 5211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.535041955467002e-06,
+      "loss": 0.701,
+      "step": 5212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5343899671725085e-06,
+      "loss": 0.7847,
+      "step": 5213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5337378939819508e-06,
+      "loss": 0.7123,
+      "step": 5214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5330857359488464e-06,
+      "loss": 0.6932,
+      "step": 5215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.53243349312672e-06,
+      "loss": 0.6359,
+      "step": 5216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.531781165569104e-06,
+      "loss": 0.6275,
+      "step": 5217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.531128753329537e-06,
+      "loss": 0.6648,
+      "step": 5218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5304762564615645e-06,
+      "loss": 0.7382,
+      "step": 5219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5298236750187403e-06,
+      "loss": 0.615,
+      "step": 5220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.529171009054623e-06,
+      "loss": 0.6401,
+      "step": 5221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5285182586227794e-06,
+      "loss": 0.7517,
+      "step": 5222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5278654237767823e-06,
+      "loss": 0.6747,
+      "step": 5223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5272125045702144e-06,
+      "loss": 0.6392,
+      "step": 5224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.526559501056661e-06,
+      "loss": 0.7401,
+      "step": 5225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5259064132897173e-06,
+      "loss": 0.7343,
+      "step": 5226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.525253241322985e-06,
+      "loss": 0.8899,
+      "step": 5227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.524599985210071e-06,
+      "loss": 0.7498,
+      "step": 5228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5239466450045917e-06,
+      "loss": 0.6437,
+      "step": 5229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5232932207601678e-06,
+      "loss": 0.6607,
+      "step": 5230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5226397125304294e-06,
+      "loss": 0.7155,
+      "step": 5231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5219861203690103e-06,
+      "loss": 0.6858,
+      "step": 5232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5213324443295547e-06,
+      "loss": 0.7235,
+      "step": 5233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5206786844657133e-06,
+      "loss": 0.7674,
+      "step": 5234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5200248408311397e-06,
+      "loss": 0.6901,
+      "step": 5235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5193709134794994e-06,
+      "loss": 0.6491,
+      "step": 5236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5187169024644608e-06,
+      "loss": 0.6243,
+      "step": 5237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5180628078397024e-06,
+      "loss": 0.7278,
+      "step": 5238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.517408629658907e-06,
+      "loss": 0.6612,
+      "step": 5239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.516754367975766e-06,
+      "loss": 0.7879,
+      "step": 5240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5161000228439764e-06,
+      "loss": 0.8178,
+      "step": 5241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.515445594317243e-06,
+      "loss": 0.6494,
+      "step": 5242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5147910824492775e-06,
+      "loss": 0.6593,
+      "step": 5243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.514136487293797e-06,
+      "loss": 0.8257,
+      "step": 5244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5134818089045274e-06,
+      "loss": 0.7024,
+      "step": 5245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5128270473352006e-06,
+      "loss": 0.6874,
+      "step": 5246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5121722026395537e-06,
+      "loss": 0.6958,
+      "step": 5247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5115172748713343e-06,
+      "loss": 0.6945,
+      "step": 5248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5108622640842925e-06,
+      "loss": 0.7982,
+      "step": 5249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5102071703321888e-06,
+      "loss": 0.6789,
+      "step": 5250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5095519936687885e-06,
+      "loss": 0.739,
+      "step": 5251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.508896734147864e-06,
+      "loss": 0.7015,
+      "step": 5252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5082413918231962e-06,
+      "loss": 0.6755,
+      "step": 5253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5075859667485702e-06,
+      "loss": 0.6408,
+      "step": 5254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5069304589777792e-06,
+      "loss": 0.7023,
+      "step": 5255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5062748685646225e-06,
+      "loss": 0.831,
+      "step": 5256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.505619195562908e-06,
+      "loss": 0.6493,
+      "step": 5257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.504963440026448e-06,
+      "loss": 0.7481,
+      "step": 5258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.504307602009063e-06,
+      "loss": 0.6441,
+      "step": 5259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5036516815645805e-06,
+      "loss": 0.7136,
+      "step": 5260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.502995678746834e-06,
+      "loss": 0.7617,
+      "step": 5261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5023395936096637e-06,
+      "loss": 0.7471,
+      "step": 5262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5016834262069167e-06,
+      "loss": 0.6821,
+      "step": 5263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5010271765924474e-06,
+      "loss": 0.6513,
+      "step": 5264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.500370844820117e-06,
+      "loss": 0.748,
+      "step": 5265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4997144309437914e-06,
+      "loss": 0.816,
+      "step": 5266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4990579350173465e-06,
+      "loss": 0.8215,
+      "step": 5267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4984013570946623e-06,
+      "loss": 0.7957,
+      "step": 5268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.497744697229626e-06,
+      "loss": 0.6802,
+      "step": 5269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4970879554761337e-06,
+      "loss": 0.722,
+      "step": 5270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.496431131888085e-06,
+      "loss": 0.7595,
+      "step": 5271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4957742265193895e-06,
+      "loss": 0.6472,
+      "step": 5272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4951172394239596e-06,
+      "loss": 0.7208,
+      "step": 5273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.494460170655718e-06,
+      "loss": 0.8068,
+      "step": 5274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.493803020268592e-06,
+      "loss": 0.666,
+      "step": 5275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.493145788316516e-06,
+      "loss": 0.853,
+      "step": 5276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4924884748534326e-06,
+      "loss": 0.6559,
+      "step": 5277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.491831079933289e-06,
+      "loss": 0.6471,
+      "step": 5278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4911736036100397e-06,
+      "loss": 0.8204,
+      "step": 5279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4905160459376464e-06,
+      "loss": 0.8749,
+      "step": 5280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.489858406970078e-06,
+      "loss": 0.5914,
+      "step": 5281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4892006867613083e-06,
+      "loss": 0.7454,
+      "step": 5282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4885428853653187e-06,
+      "loss": 0.756,
+      "step": 5283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4878850028360983e-06,
+      "loss": 0.8364,
+      "step": 5284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4872270392276403e-06,
+      "loss": 0.6791,
+      "step": 5285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4865689945939473e-06,
+      "loss": 0.5952,
+      "step": 5286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.485910868989027e-06,
+      "loss": 0.7111,
+      "step": 5287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4852526624668942e-06,
+      "loss": 0.6261,
+      "step": 5288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4845943750815698e-06,
+      "loss": 0.6733,
+      "step": 5289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4839360068870825e-06,
+      "loss": 0.7162,
+      "step": 5290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.483277557937467e-06,
+      "loss": 0.6647,
+      "step": 5291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4826190282867633e-06,
+      "loss": 0.768,
+      "step": 5292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.481960417989021e-06,
+      "loss": 0.7506,
+      "step": 5293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4813017270982936e-06,
+      "loss": 0.7148,
+      "step": 5294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4806429556686417e-06,
+      "loss": 0.7358,
+      "step": 5295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4799841037541344e-06,
+      "loss": 0.7441,
+      "step": 5296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4793251714088444e-06,
+      "loss": 0.7669,
+      "step": 5297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4786661586868536e-06,
+      "loss": 0.7777,
+      "step": 5298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4780070656422493e-06,
+      "loss": 0.6813,
+      "step": 5299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4773478923291263e-06,
+      "loss": 0.7473,
+      "step": 5300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4766886388015847e-06,
+      "loss": 0.7429,
+      "step": 5301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4760293051137313e-06,
+      "loss": 0.8519,
+      "step": 5302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4753698913196808e-06,
+      "loss": 0.7525,
+      "step": 5303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.474710397473553e-06,
+      "loss": 0.8453,
+      "step": 5304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.474050823629476e-06,
+      "loss": 0.6976,
+      "step": 5305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.473391169841581e-06,
+      "loss": 0.562,
+      "step": 5306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.472731436164011e-06,
+      "loss": 0.7088,
+      "step": 5307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.472071622650911e-06,
+      "loss": 0.7148,
+      "step": 5308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4714117293564344e-06,
+      "loss": 0.6518,
+      "step": 5309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4707517563347417e-06,
+      "loss": 0.7689,
+      "step": 5310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.470091703639998e-06,
+      "loss": 0.7126,
+      "step": 5311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4694315713263775e-06,
+      "loss": 0.6822,
+      "step": 5312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.468771359448059e-06,
+      "loss": 0.7818,
+      "step": 5313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4681110680592277e-06,
+      "loss": 0.7603,
+      "step": 5314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4674506972140783e-06,
+      "loss": 0.6354,
+      "step": 5315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4667902469668068e-06,
+      "loss": 0.6974,
+      "step": 5316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.466129717371621e-06,
+      "loss": 0.829,
+      "step": 5317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.465469108482732e-06,
+      "loss": 0.915,
+      "step": 5318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4648084203543587e-06,
+      "loss": 0.7509,
+      "step": 5319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4641476530407257e-06,
+      "loss": 0.7033,
+      "step": 5320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.463486806596064e-06,
+      "loss": 0.8075,
+      "step": 5321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4628258810746133e-06,
+      "loss": 0.7523,
+      "step": 5322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.462164876530617e-06,
+      "loss": 0.6513,
+      "step": 5323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4615037930183263e-06,
+      "loss": 0.6661,
+      "step": 5324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4608426305919983e-06,
+      "loss": 0.6747,
+      "step": 5325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.460181389305897e-06,
+      "loss": 0.6646,
+      "step": 5326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4595200692142933e-06,
+      "loss": 0.751,
+      "step": 5327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.458858670371463e-06,
+      "loss": 0.6673,
+      "step": 5328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.458197192831692e-06,
+      "loss": 0.6784,
+      "step": 5329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4575356366492673e-06,
+      "loss": 0.6571,
+      "step": 5330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4568740018784874e-06,
+      "loss": 0.8489,
+      "step": 5331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.456212288573653e-06,
+      "loss": 0.7318,
+      "step": 5332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.455550496789074e-06,
+      "loss": 0.6996,
+      "step": 5333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.454888626579067e-06,
+      "loss": 0.6482,
+      "step": 5334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4542266779979526e-06,
+      "loss": 0.6545,
+      "step": 5335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4535646511000605e-06,
+      "loss": 0.5931,
+      "step": 5336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.452902545939724e-06,
+      "loss": 0.6501,
+      "step": 5337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4522403625712863e-06,
+      "loss": 0.7045,
+      "step": 5338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4515781010490947e-06,
+      "loss": 0.7708,
+      "step": 5339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.450915761427502e-06,
+      "loss": 0.7781,
+      "step": 5340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.45025334376087e-06,
+      "loss": 0.7816,
+      "step": 5341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4495908481035657e-06,
+      "loss": 0.7089,
+      "step": 5342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4489282745099617e-06,
+      "loss": 0.7573,
+      "step": 5343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4482656230344387e-06,
+      "loss": 0.7721,
+      "step": 5344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.447602893731382e-06,
+      "loss": 0.7274,
+      "step": 5345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4469400866551846e-06,
+      "loss": 0.6853,
+      "step": 5346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.446277201860245e-06,
+      "loss": 0.7193,
+      "step": 5347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.44561423940097e-06,
+      "loss": 0.7684,
+      "step": 5348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.44495119933177e-06,
+      "loss": 0.7087,
+      "step": 5349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4442880817070625e-06,
+      "loss": 0.6315,
+      "step": 5350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4436248865812728e-06,
+      "loss": 0.7875,
+      "step": 5351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.442961614008832e-06,
+      "loss": 0.7126,
+      "step": 5352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4422982640441767e-06,
+      "loss": 0.6461,
+      "step": 5353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.44163483674175e-06,
+      "loss": 0.7489,
+      "step": 5354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4409713321560022e-06,
+      "loss": 0.6625,
+      "step": 5355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.44030775034139e-06,
+      "loss": 0.6691,
+      "step": 5356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.439644091352376e-06,
+      "loss": 0.6957,
+      "step": 5357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4389803552434275e-06,
+      "loss": 0.5843,
+      "step": 5358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4383165420690207e-06,
+      "loss": 0.6092,
+      "step": 5359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.437652651883638e-06,
+      "loss": 0.6996,
+      "step": 5360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4369886847417653e-06,
+      "loss": 0.8025,
+      "step": 5361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4363246406978984e-06,
+      "loss": 0.6579,
+      "step": 5362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4356605198065368e-06,
+      "loss": 0.795,
+      "step": 5363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.434996322122187e-06,
+      "loss": 0.719,
+      "step": 5364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4343320476993635e-06,
+      "loss": 0.6907,
+      "step": 5365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4336676965925842e-06,
+      "loss": 0.6917,
+      "step": 5366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4330032688563766e-06,
+      "loss": 0.7101,
+      "step": 5367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.43233876454527e-06,
+      "loss": 0.7737,
+      "step": 5368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4316741837138046e-06,
+      "loss": 0.7484,
+      "step": 5369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4310095264165245e-06,
+      "loss": 0.6164,
+      "step": 5370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4303447927079796e-06,
+      "loss": 0.5748,
+      "step": 5371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4296799826427286e-06,
+      "loss": 0.6008,
+      "step": 5372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4290150962753334e-06,
+      "loss": 0.7198,
+      "step": 5373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.428350133660364e-06,
+      "loss": 0.6943,
+      "step": 5374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4276850948523965e-06,
+      "loss": 0.7716,
+      "step": 5375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4270199799060132e-06,
+      "loss": 0.7326,
+      "step": 5376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.426354788875802e-06,
+      "loss": 0.7609,
+      "step": 5377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4256895218163577e-06,
+      "loss": 0.6782,
+      "step": 5378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.425024178782281e-06,
+      "loss": 0.7219,
+      "step": 5379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.424358759828179e-06,
+      "loss": 0.6569,
+      "step": 5380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.423693265008665e-06,
+      "loss": 0.7819,
+      "step": 5381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4230276943783586e-06,
+      "loss": 0.763,
+      "step": 5382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.422362047991885e-06,
+      "loss": 0.6627,
+      "step": 5383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4216963259038772e-06,
+      "loss": 0.6818,
+      "step": 5384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4210305281689726e-06,
+      "loss": 0.8132,
+      "step": 5385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4203646548418163e-06,
+      "loss": 0.6677,
+      "step": 5386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.419698705977058e-06,
+      "loss": 0.7192,
+      "step": 5387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4190326816293553e-06,
+      "loss": 0.7302,
+      "step": 5388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.41836658185337e-06,
+      "loss": 0.7107,
+      "step": 5389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.417700406703773e-06,
+      "loss": 0.6225,
+      "step": 5390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4170341562352383e-06,
+      "loss": 0.7884,
+      "step": 5391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4163678305024477e-06,
+      "loss": 0.6216,
+      "step": 5392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4157014295600893e-06,
+      "loss": 0.6614,
+      "step": 5393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.415034953462857e-06,
+      "loss": 0.7433,
+      "step": 5394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4143684022654506e-06,
+      "loss": 0.8028,
+      "step": 5395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4137017760225765e-06,
+      "loss": 0.6905,
+      "step": 5396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.413035074788946e-06,
+      "loss": 0.8186,
+      "step": 5397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4123682986192803e-06,
+      "loss": 0.7291,
+      "step": 5398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4117014475683012e-06,
+      "loss": 0.7286,
+      "step": 5399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4110345216907416e-06,
+      "loss": 0.86,
+      "step": 5400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.410367521041337e-06,
+      "loss": 0.6865,
+      "step": 5401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4097004456748315e-06,
+      "loss": 0.7846,
+      "step": 5402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4090332956459737e-06,
+      "loss": 0.6812,
+      "step": 5403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4083660710095195e-06,
+      "loss": 0.6407,
+      "step": 5404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.40769877182023e-06,
+      "loss": 0.6635,
+      "step": 5405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4070313981328736e-06,
+      "loss": 0.6906,
+      "step": 5406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4063639500022235e-06,
+      "loss": 0.6866,
+      "step": 5407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4056964274830584e-06,
+      "loss": 0.7122,
+      "step": 5408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.405028830630166e-06,
+      "loss": 0.6016,
+      "step": 5409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4043611594983376e-06,
+      "loss": 0.6854,
+      "step": 5410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4036934141423716e-06,
+      "loss": 0.7302,
+      "step": 5411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.403025594617072e-06,
+      "loss": 0.819,
+      "step": 5412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4023577009772487e-06,
+      "loss": 0.7534,
+      "step": 5413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4016897332777196e-06,
+      "loss": 0.6544,
+      "step": 5414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.401021691573305e-06,
+      "loss": 0.7152,
+      "step": 5415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4003535759188354e-06,
+      "loss": 0.6205,
+      "step": 5416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.399685386369145e-06,
+      "loss": 0.6516,
+      "step": 5417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.399017122979073e-06,
+      "loss": 0.7971,
+      "step": 5418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.398348785803468e-06,
+      "loss": 0.7225,
+      "step": 5419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.3976803748971822e-06,
+      "loss": 0.6781,
+      "step": 5420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.3970118903150738e-06,
+      "loss": 0.7906,
+      "step": 5421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.3963433321120087e-06,
+      "loss": 0.8123,
+      "step": 5422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3956747003428574e-06,
+      "loss": 0.7715,
+      "step": 5423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3950059950624974e-06,
+      "loss": 0.7421,
+      "step": 5424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.394337216325811e-06,
+      "loss": 0.5545,
+      "step": 5425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.393668364187688e-06,
+      "loss": 0.6214,
+      "step": 5426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.392999438703022e-06,
+      "loss": 0.6585,
+      "step": 5427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3923304399267147e-06,
+      "loss": 0.7608,
+      "step": 5428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.391661367913675e-06,
+      "loss": 0.6582,
+      "step": 5429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3909922227188135e-06,
+      "loss": 0.6832,
+      "step": 5430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.390323004397051e-06,
+      "loss": 0.6468,
+      "step": 5431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3896537130033115e-06,
+      "loss": 0.6128,
+      "step": 5432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.388984348592528e-06,
+      "loss": 0.7091,
+      "step": 5433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3883149112196346e-06,
+      "loss": 0.7028,
+      "step": 5434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.387645400939577e-06,
+      "loss": 0.6882,
+      "step": 5435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3869758178073035e-06,
+      "loss": 0.6239,
+      "step": 5436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3863061618777687e-06,
+      "loss": 0.7618,
+      "step": 5437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.385636433205934e-06,
+      "loss": 0.7975,
+      "step": 5438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3849666318467665e-06,
+      "loss": 0.7532,
+      "step": 5439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3842967578552394e-06,
+      "loss": 0.6705,
+      "step": 5440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3836268112863315e-06,
+      "loss": 0.703,
+      "step": 5441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.382956792195027e-06,
+      "loss": 0.7266,
+      "step": 5442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3822867006363177e-06,
+      "loss": 0.7429,
+      "step": 5443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3816165366651993e-06,
+      "loss": 0.681,
+      "step": 5444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.380946300336676e-06,
+      "loss": 0.6393,
+      "step": 5445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3802759917057554e-06,
+      "loss": 0.899,
+      "step": 5446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.379605610827452e-06,
+      "loss": 0.6734,
+      "step": 5447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.378935157756788e-06,
+      "loss": 0.7395,
+      "step": 5448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.378264632548788e-06,
+      "loss": 0.7591,
+      "step": 5449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3775940352584845e-06,
+      "loss": 0.6537,
+      "step": 5450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.376923365940916e-06,
+      "loss": 0.645,
+      "step": 5451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3762526246511285e-06,
+      "loss": 0.678,
+      "step": 5452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3755818114441702e-06,
+      "loss": 1.0178,
+      "step": 5453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.374910926375097e-06,
+      "loss": 0.6161,
+      "step": 5454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3742399694989726e-06,
+      "loss": 0.7107,
+      "step": 5455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3735689408708623e-06,
+      "loss": 0.7494,
+      "step": 5456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.372897840545842e-06,
+      "loss": 0.6456,
+      "step": 5457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.37222666857899e-06,
+      "loss": 0.6655,
+      "step": 5458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3715554250253924e-06,
+      "loss": 0.7273,
+      "step": 5459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.37088410994014e-06,
+      "loss": 0.7697,
+      "step": 5460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.370212723378331e-06,
+      "loss": 0.8025,
+      "step": 5461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3695412653950678e-06,
+      "loss": 0.7334,
+      "step": 5462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3688697360454594e-06,
+      "loss": 0.6568,
+      "step": 5463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3681981353846206e-06,
+      "loss": 0.7489,
+      "step": 5464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3675264634676718e-06,
+      "loss": 0.7486,
+      "step": 5465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3668547203497395e-06,
+      "loss": 0.5963,
+      "step": 5466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3661829060859574e-06,
+      "loss": 0.6771,
+      "step": 5467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.365511020731461e-06,
+      "loss": 0.7681,
+      "step": 5468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.364839064341397e-06,
+      "loss": 0.6957,
+      "step": 5469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.364167036970914e-06,
+      "loss": 0.7015,
+      "step": 5470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.363494938675168e-06,
+      "loss": 0.8422,
+      "step": 5471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.36282276950932e-06,
+      "loss": 0.743,
+      "step": 5472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3621505295285377e-06,
+      "loss": 0.7319,
+      "step": 5473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.361478218787995e-06,
+      "loss": 0.6436,
+      "step": 5474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.360805837342869e-06,
+      "loss": 0.6643,
+      "step": 5475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3601333852483464e-06,
+      "loss": 0.669,
+      "step": 5476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3594608625596163e-06,
+      "loss": 0.7548,
+      "step": 5477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.358788269331875e-06,
+      "loss": 0.6565,
+      "step": 5478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3581156056203257e-06,
+      "loss": 0.8737,
+      "step": 5479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3574428714801762e-06,
+      "loss": 0.7933,
+      "step": 5480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3567700669666393e-06,
+      "loss": 0.6466,
+      "step": 5481
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.8166519403457642,
+      "eval_runtime": 437.5663,
+      "eval_samples_per_second": 6.854,
+      "eval_steps_per_second": 0.857,
+      "step": 5481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.356097192134935e-06,
+      "loss": 0.6619,
+      "step": 5482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3554242470402888e-06,
+      "loss": 0.625,
+      "step": 5483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3547512317379303e-06,
+      "loss": 0.7667,
+      "step": 5484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.354078146283098e-06,
+      "loss": 0.6645,
+      "step": 5485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.353404990731034e-06,
+      "loss": 0.7537,
+      "step": 5486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.352731765136985e-06,
+      "loss": 0.7025,
+      "step": 5487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3520584695562076e-06,
+      "loss": 0.6636,
+      "step": 5488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3513851040439593e-06,
+      "loss": 0.6168,
+      "step": 5489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3507116686555073e-06,
+      "loss": 0.7182,
+      "step": 5490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3500381634461223e-06,
+      "loss": 0.7038,
+      "step": 5491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3493645884710803e-06,
+      "loss": 0.6536,
+      "step": 5492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.348690943785665e-06,
+      "loss": 0.6898,
+      "step": 5493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3480172294451646e-06,
+      "loss": 0.7459,
+      "step": 5494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3473434455048736e-06,
+      "loss": 0.6405,
+      "step": 5495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3466695920200915e-06,
+      "loss": 0.7175,
+      "step": 5496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.345995669046123e-06,
+      "loss": 0.6778,
+      "step": 5497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3453216766382807e-06,
+      "loss": 0.8607,
+      "step": 5498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3446476148518808e-06,
+      "loss": 0.6936,
+      "step": 5499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3439734837422466e-06,
+      "loss": 0.6316,
+      "step": 5500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3432992833647056e-06,
+      "loss": 0.6476,
+      "step": 5501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.342625013774593e-06,
+      "loss": 0.6332,
+      "step": 5502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3419506750272468e-06,
+      "loss": 0.7359,
+      "step": 5503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3412762671780134e-06,
+      "loss": 0.7851,
+      "step": 5504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3406017902822447e-06,
+      "loss": 0.6756,
+      "step": 5505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3399272443952956e-06,
+      "loss": 0.6305,
+      "step": 5506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.339252629572529e-06,
+      "loss": 0.7199,
+      "step": 5507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.338577945869314e-06,
+      "loss": 0.7578,
+      "step": 5508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.337903193341024e-06,
+      "loss": 0.7518,
+      "step": 5509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.337228372043037e-06,
+      "loss": 0.6756,
+      "step": 5510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3365534820307392e-06,
+      "loss": 0.8337,
+      "step": 5511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3358785233595213e-06,
+      "loss": 0.7851,
+      "step": 5512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.335203496084779e-06,
+      "loss": 0.786,
+      "step": 5513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.334528400261915e-06,
+      "loss": 0.755,
+      "step": 5514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3338532359463355e-06,
+      "loss": 0.7767,
+      "step": 5515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3331780031934548e-06,
+      "loss": 0.6678,
+      "step": 5516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.33250270205869e-06,
+      "loss": 0.8036,
+      "step": 5517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3318273325974683e-06,
+      "loss": 0.744,
+      "step": 5518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3311518948652172e-06,
+      "loss": 0.7316,
+      "step": 5519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.330476388917373e-06,
+      "loss": 0.7018,
+      "step": 5520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.329800814809378e-06,
+      "loss": 0.7325,
+      "step": 5521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3291251725966765e-06,
+      "loss": 0.6096,
+      "step": 5522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3284494623347235e-06,
+      "loss": 0.6536,
+      "step": 5523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.327773684078975e-06,
+      "loss": 0.6476,
+      "step": 5524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.327097837884896e-06,
+      "loss": 0.7168,
+      "step": 5525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3264219238079543e-06,
+      "loss": 0.6664,
+      "step": 5526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.325745941903625e-06,
+      "loss": 0.79,
+      "step": 5527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.32506989222739e-06,
+      "loss": 0.7296,
+      "step": 5528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3243937748347323e-06,
+      "loss": 0.6652,
+      "step": 5529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3237175897811446e-06,
+      "loss": 0.8552,
+      "step": 5530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.323041337122125e-06,
+      "loss": 0.7664,
+      "step": 5531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3223650169131734e-06,
+      "loss": 0.6802,
+      "step": 5532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3216886292098e-06,
+      "loss": 0.8474,
+      "step": 5533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3210121740675165e-06,
+      "loss": 0.6725,
+      "step": 5534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.320335651541844e-06,
+      "loss": 0.6223,
+      "step": 5535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.319659061688305e-06,
+      "loss": 0.7421,
+      "step": 5536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.318982404562431e-06,
+      "loss": 0.6798,
+      "step": 5537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.318305680219759e-06,
+      "loss": 0.6803,
+      "step": 5538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.317628888715827e-06,
+      "loss": 0.8469,
+      "step": 5539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3169520301061836e-06,
+      "loss": 0.6908,
+      "step": 5540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3162751044463802e-06,
+      "loss": 0.6885,
+      "step": 5541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.315598111791976e-06,
+      "loss": 0.6777,
+      "step": 5542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3149210521985326e-06,
+      "loss": 0.6943,
+      "step": 5543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.314243925721619e-06,
+      "loss": 0.6599,
+      "step": 5544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3135667324168096e-06,
+      "loss": 0.7546,
+      "step": 5545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3128894723396844e-06,
+      "loss": 0.7883,
+      "step": 5546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3122121455458277e-06,
+      "loss": 0.6776,
+      "step": 5547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3115347520908314e-06,
+      "loss": 0.746,
+      "step": 5548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.31085729203029e-06,
+      "loss": 0.7546,
+      "step": 5549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.310179765419807e-06,
+      "loss": 0.6928,
+      "step": 5550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.309502172314987e-06,
+      "loss": 0.6648,
+      "step": 5551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3088245127714454e-06,
+      "loss": 0.8048,
+      "step": 5552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.308146786844797e-06,
+      "loss": 0.6904,
+      "step": 5553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3074689945906666e-06,
+      "loss": 0.7379,
+      "step": 5554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3067911360646844e-06,
+      "loss": 0.6381,
+      "step": 5555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.306113211322482e-06,
+      "loss": 0.8044,
+      "step": 5556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3054352204197014e-06,
+      "loss": 0.7461,
+      "step": 5557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.304757163411987e-06,
+      "loss": 0.6642,
+      "step": 5558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.304079040354988e-06,
+      "loss": 0.832,
+      "step": 5559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3034008513043625e-06,
+      "loss": 0.6544,
+      "step": 5560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.302722596315771e-06,
+      "loss": 0.7146,
+      "step": 5561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3020442754448796e-06,
+      "loss": 0.6153,
+      "step": 5562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.301365888747361e-06,
+      "loss": 0.8012,
+      "step": 5563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3006874362788934e-06,
+      "loss": 0.8776,
+      "step": 5564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3000089180951588e-06,
+      "loss": 0.691,
+      "step": 5565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2993303342518466e-06,
+      "loss": 0.66,
+      "step": 5566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2986516848046503e-06,
+      "loss": 0.7786,
+      "step": 5567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2979729698092677e-06,
+      "loss": 0.796,
+      "step": 5568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.297294189321406e-06,
+      "loss": 0.7255,
+      "step": 5569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.296615343396773e-06,
+      "loss": 0.6635,
+      "step": 5570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2959364320910854e-06,
+      "loss": 0.6474,
+      "step": 5571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.295257455460062e-06,
+      "loss": 0.7432,
+      "step": 5572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2945784135594307e-06,
+      "loss": 0.6644,
+      "step": 5573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2938993064449216e-06,
+      "loss": 0.6672,
+      "step": 5574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2932201341722717e-06,
+      "loss": 0.6424,
+      "step": 5575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.292540896797225e-06,
+      "loss": 0.7629,
+      "step": 5576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.291861594375526e-06,
+      "loss": 0.7216,
+      "step": 5577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.291182226962929e-06,
+      "loss": 0.8052,
+      "step": 5578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.290502794615192e-06,
+      "loss": 0.683,
+      "step": 5579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2898232973880783e-06,
+      "loss": 0.6677,
+      "step": 5580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2891437353373566e-06,
+      "loss": 0.6826,
+      "step": 5581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2884641085188007e-06,
+      "loss": 0.6425,
+      "step": 5582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2877844169881905e-06,
+      "loss": 0.6339,
+      "step": 5583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2871046608013108e-06,
+      "loss": 0.691,
+      "step": 5584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2864248400139514e-06,
+      "loss": 0.7763,
+      "step": 5585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2857449546819076e-06,
+      "loss": 0.6516,
+      "step": 5586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2850650048609787e-06,
+      "loss": 0.6975,
+      "step": 5587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2843849906069735e-06,
+      "loss": 0.6134,
+      "step": 5588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2837049119757008e-06,
+      "loss": 0.6815,
+      "step": 5589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2830247690229784e-06,
+      "loss": 0.7438,
+      "step": 5590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2823445618046267e-06,
+      "loss": 0.6952,
+      "step": 5591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.281664290376474e-06,
+      "loss": 0.8005,
+      "step": 5592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2809839547943513e-06,
+      "loss": 0.6397,
+      "step": 5593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.280303555114097e-06,
+      "loss": 0.701,
+      "step": 5594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.279623091391555e-06,
+      "loss": 0.6953,
+      "step": 5595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2789425636825717e-06,
+      "loss": 0.6701,
+      "step": 5596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.278261972043001e-06,
+      "loss": 0.7282,
+      "step": 5597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.277581316528701e-06,
+      "loss": 0.6833,
+      "step": 5598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.276900597195537e-06,
+      "loss": 0.7553,
+      "step": 5599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.276219814099376e-06,
+      "loss": 0.7158,
+      "step": 5600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2755389672960937e-06,
+      "loss": 0.8396,
+      "step": 5601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.274858056841569e-06,
+      "loss": 0.7742,
+      "step": 5602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.274177082791687e-06,
+      "loss": 0.7822,
+      "step": 5603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.273496045202338e-06,
+      "loss": 0.7383,
+      "step": 5604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2728149441294167e-06,
+      "loss": 0.6882,
+      "step": 5605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2721337796288227e-06,
+      "loss": 0.6729,
+      "step": 5606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2714525517564634e-06,
+      "loss": 0.776,
+      "step": 5607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2707712605682477e-06,
+      "loss": 0.847,
+      "step": 5608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.270089906120093e-06,
+      "loss": 0.5772,
+      "step": 5609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.26940848846792e-06,
+      "loss": 0.7279,
+      "step": 5610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2687270076676546e-06,
+      "loss": 0.7619,
+      "step": 5611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.268045463775229e-06,
+      "loss": 0.6552,
+      "step": 5612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2673638568465802e-06,
+      "loss": 0.695,
+      "step": 5613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2666821869376498e-06,
+      "loss": 0.7249,
+      "step": 5614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.266000454104384e-06,
+      "loss": 0.6646,
+      "step": 5615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2653186584027362e-06,
+      "loss": 0.7536,
+      "step": 5616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.264636799888663e-06,
+      "loss": 0.8116,
+      "step": 5617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.263954878618128e-06,
+      "loss": 0.7088,
+      "step": 5618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2632728946470975e-06,
+      "loss": 0.6969,
+      "step": 5619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.262590848031545e-06,
+      "loss": 0.7513,
+      "step": 5620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.261908738827449e-06,
+      "loss": 0.8519,
+      "step": 5621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2612265670907915e-06,
+      "loss": 0.6108,
+      "step": 5622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.260544332877563e-06,
+      "loss": 0.6312,
+      "step": 5623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.259862036243754e-06,
+      "loss": 0.7509,
+      "step": 5624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.259179677245364e-06,
+      "loss": 0.7798,
+      "step": 5625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2584972559383983e-06,
+      "loss": 0.8368,
+      "step": 5626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.257814772378863e-06,
+      "loss": 0.6506,
+      "step": 5627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2571322266227745e-06,
+      "loss": 0.7015,
+      "step": 5628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2564496187261496e-06,
+      "loss": 0.6198,
+      "step": 5629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.255766948745014e-06,
+      "loss": 0.6591,
+      "step": 5630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2550842167353956e-06,
+      "loss": 0.7582,
+      "step": 5631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2544014227533295e-06,
+      "loss": 0.6698,
+      "step": 5632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.253718566854854e-06,
+      "loss": 0.633,
+      "step": 5633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.253035649096015e-06,
+      "loss": 0.8156,
+      "step": 5634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2523526695328618e-06,
+      "loss": 0.72,
+      "step": 5635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2516696282214475e-06,
+      "loss": 0.7263,
+      "step": 5636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.250986525217833e-06,
+      "loss": 0.875,
+      "step": 5637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2503033605780826e-06,
+      "loss": 0.7474,
+      "step": 5638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2496201343582655e-06,
+      "loss": 0.7252,
+      "step": 5639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2489368466144576e-06,
+      "loss": 0.7172,
+      "step": 5640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.248253497402738e-06,
+      "loss": 0.7279,
+      "step": 5641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.247570086779192e-06,
+      "loss": 0.6879,
+      "step": 5642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.246886614799909e-06,
+      "loss": 0.7315,
+      "step": 5643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2462030815209846e-06,
+      "loss": 0.6998,
+      "step": 5644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.245519486998519e-06,
+      "loss": 0.6593,
+      "step": 5645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.244835831288616e-06,
+      "loss": 0.6892,
+      "step": 5646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2441521144473875e-06,
+      "loss": 0.6398,
+      "step": 5647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.243468336530947e-06,
+      "loss": 0.7138,
+      "step": 5648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2427844975954147e-06,
+      "loss": 0.7322,
+      "step": 5649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.242100597696916e-06,
+      "loss": 0.6493,
+      "step": 5650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2414166368915816e-06,
+      "loss": 0.6656,
+      "step": 5651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2407326152355463e-06,
+      "loss": 0.6341,
+      "step": 5652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2400485327849494e-06,
+      "loss": 0.7691,
+      "step": 5653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.239364389595937e-06,
+      "loss": 0.7218,
+      "step": 5654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2386801857246587e-06,
+      "loss": 0.6712,
+      "step": 5655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.23799592122727e-06,
+      "loss": 0.7823,
+      "step": 5656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.23731159615993e-06,
+      "loss": 0.6439,
+      "step": 5657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2366272105788037e-06,
+      "loss": 0.7174,
+      "step": 5658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.235942764540062e-06,
+      "loss": 0.6529,
+      "step": 5659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2352582580998793e-06,
+      "loss": 0.6742,
+      "step": 5660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2345736913144354e-06,
+      "loss": 0.6948,
+      "step": 5661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2338890642399155e-06,
+      "loss": 0.686,
+      "step": 5662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.2332043769325096e-06,
+      "loss": 0.7573,
+      "step": 5663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.232519629448412e-06,
+      "loss": 0.6392,
+      "step": 5664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.231834821843821e-06,
+      "loss": 0.6259,
+      "step": 5665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.231149954174944e-06,
+      "loss": 0.7088,
+      "step": 5666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2304650264979887e-06,
+      "loss": 0.5871,
+      "step": 5667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.229780038869169e-06,
+      "loss": 0.7801,
+      "step": 5668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2290949913447063e-06,
+      "loss": 0.7045,
+      "step": 5669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.228409883980823e-06,
+      "loss": 0.6687,
+      "step": 5670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.22772471683375e-06,
+      "loss": 0.7745,
+      "step": 5671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2270394899597203e-06,
+      "loss": 0.6851,
+      "step": 5672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.226354203414973e-06,
+      "loss": 0.7213,
+      "step": 5673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2256688572557513e-06,
+      "loss": 0.617,
+      "step": 5674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.224983451538306e-06,
+      "loss": 0.6599,
+      "step": 5675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2242979863188887e-06,
+      "loss": 0.732,
+      "step": 5676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.223612461653759e-06,
+      "loss": 0.6254,
+      "step": 5677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.22292687759918e-06,
+      "loss": 0.7059,
+      "step": 5678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.222241234211421e-06,
+      "loss": 0.658,
+      "step": 5679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2215555315467538e-06,
+      "loss": 0.8186,
+      "step": 5680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.220869769661457e-06,
+      "loss": 0.8115,
+      "step": 5681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2201839486118143e-06,
+      "loss": 0.8323,
+      "step": 5682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.219498068454112e-06,
+      "loss": 0.7208,
+      "step": 5683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.218812129244644e-06,
+      "loss": 0.8733,
+      "step": 5684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.218126131039707e-06,
+      "loss": 0.6858,
+      "step": 5685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2174400738956042e-06,
+      "loss": 0.6022,
+      "step": 5686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.216753957868641e-06,
+      "loss": 0.6479,
+      "step": 5687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.216067783015131e-06,
+      "loss": 0.8449,
+      "step": 5688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2153815493913903e-06,
+      "loss": 0.6411,
+      "step": 5689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2146952570537417e-06,
+      "loss": 0.7045,
+      "step": 5690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2140089060585096e-06,
+      "loss": 0.6488,
+      "step": 5691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.213322496462027e-06,
+      "loss": 0.5988,
+      "step": 5692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2126360283206293e-06,
+      "loss": 0.6331,
+      "step": 5693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2119495016906575e-06,
+      "loss": 0.7275,
+      "step": 5694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.211262916628457e-06,
+      "loss": 0.7069,
+      "step": 5695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.210576273190378e-06,
+      "loss": 0.5939,
+      "step": 5696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.209889571432777e-06,
+      "loss": 0.6784,
+      "step": 5697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2092028114120128e-06,
+      "loss": 0.6672,
+      "step": 5698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2085159931844513e-06,
+      "loss": 0.7123,
+      "step": 5699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.207829116806461e-06,
+      "loss": 0.7555,
+      "step": 5700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.207142182334417e-06,
+      "loss": 0.8012,
+      "step": 5701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2064551898246993e-06,
+      "loss": 0.6388,
+      "step": 5702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.20576813933369e-06,
+      "loss": 0.6608,
+      "step": 5703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.205081030917779e-06,
+      "loss": 0.7149,
+      "step": 5704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2043938646333582e-06,
+      "loss": 0.8006,
+      "step": 5705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2037066405368278e-06,
+      "loss": 0.7081,
+      "step": 5706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2030193586845896e-06,
+      "loss": 0.6615,
+      "step": 5707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.202332019133052e-06,
+      "loss": 0.7332,
+      "step": 5708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2016446219386264e-06,
+      "loss": 0.6267,
+      "step": 5709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.200957167157731e-06,
+      "loss": 0.6214,
+      "step": 5710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.2002696548467872e-06,
+      "loss": 0.6648,
+      "step": 5711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1995820850622205e-06,
+      "loss": 0.8312,
+      "step": 5712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.198894457860465e-06,
+      "loss": 0.646,
+      "step": 5713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1982067732979528e-06,
+      "loss": 0.8646,
+      "step": 5714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.197519031431127e-06,
+      "loss": 0.6546,
+      "step": 5715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.196831232316434e-06,
+      "loss": 0.6965,
+      "step": 5716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1961433760103224e-06,
+      "loss": 0.7885,
+      "step": 5717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.195455462569247e-06,
+      "loss": 0.7903,
+      "step": 5718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1947674920496676e-06,
+      "loss": 0.7128,
+      "step": 5719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.194079464508048e-06,
+      "loss": 0.6264,
+      "step": 5720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1933913800008583e-06,
+      "loss": 0.7059,
+      "step": 5721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.19270323858457e-06,
+      "loss": 0.6214,
+      "step": 5722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1920150403156634e-06,
+      "loss": 0.7688,
+      "step": 5723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1913267852506204e-06,
+      "loss": 0.596,
+      "step": 5724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1906384734459282e-06,
+      "loss": 0.6937,
+      "step": 5725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1899501049580796e-06,
+      "loss": 0.7209,
+      "step": 5726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1892616798435717e-06,
+      "loss": 0.7451,
+      "step": 5727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1885731981589053e-06,
+      "loss": 0.7245,
+      "step": 5728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1878846599605865e-06,
+      "loss": 0.7955,
+      "step": 5729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.187196065305127e-06,
+      "loss": 0.6504,
+      "step": 5730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.186507414249041e-06,
+      "loss": 0.6979,
+      "step": 5731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1858187068488498e-06,
+      "loss": 0.7089,
+      "step": 5732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1851299431610765e-06,
+      "loss": 0.7094,
+      "step": 5733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1844411232422516e-06,
+      "loss": 0.6297,
+      "step": 5734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1837522471489092e-06,
+      "loss": 0.6783,
+      "step": 5735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.183063314937587e-06,
+      "loss": 0.5937,
+      "step": 5736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.182374326664829e-06,
+      "loss": 0.6202,
+      "step": 5737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1816852823871814e-06,
+      "loss": 0.7272,
+      "step": 5738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.180996182161198e-06,
+      "loss": 0.7388,
+      "step": 5739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1803070260434356e-06,
+      "loss": 0.7818,
+      "step": 5740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1796178140904543e-06,
+      "loss": 0.6571,
+      "step": 5741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1789285463588215e-06,
+      "loss": 0.7729,
+      "step": 5742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1782392229051073e-06,
+      "loss": 0.6982,
+      "step": 5743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1775498437858875e-06,
+      "loss": 0.6749,
+      "step": 5744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1768604090577417e-06,
+      "loss": 0.7205,
+      "step": 5745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1761709187772535e-06,
+      "loss": 0.7981,
+      "step": 5746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1754813730010126e-06,
+      "loss": 0.7378,
+      "step": 5747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1747917717856125e-06,
+      "loss": 0.6783,
+      "step": 5748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1741021151876512e-06,
+      "loss": 0.7544,
+      "step": 5749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1734124032637302e-06,
+      "loss": 0.5834,
+      "step": 5750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1727226360704584e-06,
+      "loss": 0.6722,
+      "step": 5751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.172032813664446e-06,
+      "loss": 0.5952,
+      "step": 5752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1713429361023096e-06,
+      "loss": 0.6678,
+      "step": 5753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.170653003440671e-06,
+      "loss": 0.8721,
+      "step": 5754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1699630157361533e-06,
+      "loss": 0.6175,
+      "step": 5755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1692729730453882e-06,
+      "loss": 0.7011,
+      "step": 5756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.168582875425009e-06,
+      "loss": 0.7342,
+      "step": 5757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1678927229316543e-06,
+      "loss": 0.6363,
+      "step": 5758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1672025156219677e-06,
+      "loss": 0.8262,
+      "step": 5759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.166512253552597e-06,
+      "loss": 0.7956,
+      "step": 5760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1658219367801946e-06,
+      "loss": 0.7096,
+      "step": 5761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1651315653614167e-06,
+      "loss": 0.766,
+      "step": 5762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1644411393529256e-06,
+      "loss": 0.8461,
+      "step": 5763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.163750658811385e-06,
+      "loss": 0.7076,
+      "step": 5764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1630601237934676e-06,
+      "loss": 0.637,
+      "step": 5765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.162369534355847e-06,
+      "loss": 0.8265,
+      "step": 5766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.161678890555202e-06,
+      "loss": 0.6559,
+      "step": 5767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1609881924482166e-06,
+      "loss": 0.7751,
+      "step": 5768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1602974400915783e-06,
+      "loss": 0.7933,
+      "step": 5769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1596066335419796e-06,
+      "loss": 0.6533,
+      "step": 5770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.158915772856118e-06,
+      "loss": 0.6674,
+      "step": 5771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.158224858090695e-06,
+      "loss": 0.7077,
+      "step": 5772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1575338893024164e-06,
+      "loss": 0.6302,
+      "step": 5773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.156842866547992e-06,
+      "loss": 0.7538,
+      "step": 5774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.156151789884137e-06,
+      "loss": 0.6531,
+      "step": 5775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1554606593675698e-06,
+      "loss": 0.7165,
+      "step": 5776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.154769475055014e-06,
+      "loss": 0.6449,
+      "step": 5777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.154078237003199e-06,
+      "loss": 0.6491,
+      "step": 5778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.153386945268855e-06,
+      "loss": 0.7184,
+      "step": 5779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1526955999087204e-06,
+      "loss": 0.6759,
+      "step": 5780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.152004200979536e-06,
+      "loss": 0.6951,
+      "step": 5781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1513127485380475e-06,
+      "loss": 0.717,
+      "step": 5782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1506212426410038e-06,
+      "loss": 0.7036,
+      "step": 5783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.14992968334516e-06,
+      "loss": 0.6354,
+      "step": 5784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1492380707072756e-06,
+      "loss": 0.7442,
+      "step": 5785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1485464047841126e-06,
+      "loss": 0.7257,
+      "step": 5786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1478546856324386e-06,
+      "loss": 0.7131,
+      "step": 5787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.1471629133090257e-06,
+      "loss": 0.6736,
+      "step": 5788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1464710878706505e-06,
+      "loss": 0.7299,
+      "step": 5789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1457792093740924e-06,
+      "loss": 0.7557,
+      "step": 5790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1450872778761375e-06,
+      "loss": 0.648,
+      "step": 5791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1443952934335755e-06,
+      "loss": 0.7332,
+      "step": 5792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1437032561031987e-06,
+      "loss": 0.6755,
+      "step": 5793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1430111659418057e-06,
+      "loss": 0.7921,
+      "step": 5794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.142319023006198e-06,
+      "loss": 0.7085,
+      "step": 5795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1416268273531837e-06,
+      "loss": 0.7622,
+      "step": 5796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.140934579039573e-06,
+      "loss": 0.66,
+      "step": 5797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.140242278122181e-06,
+      "loss": 0.6716,
+      "step": 5798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.139549924657827e-06,
+      "loss": 0.8183,
+      "step": 5799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1388575187033355e-06,
+      "loss": 0.6175,
+      "step": 5800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1381650603155356e-06,
+      "loss": 0.7535,
+      "step": 5801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1374725495512574e-06,
+      "loss": 0.9269,
+      "step": 5802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1367799864673396e-06,
+      "loss": 0.7641,
+      "step": 5803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1360873711206234e-06,
+      "loss": 0.7029,
+      "step": 5804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1353947035679534e-06,
+      "loss": 0.8526,
+      "step": 5805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1347019838661796e-06,
+      "loss": 0.6937,
+      "step": 5806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1340092120721553e-06,
+      "loss": 0.6975,
+      "step": 5807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1333163882427405e-06,
+      "loss": 0.7563,
+      "step": 5808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.132623512434795e-06,
+      "loss": 0.6708,
+      "step": 5809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1319305847051884e-06,
+      "loss": 0.8114,
+      "step": 5810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1312376051107908e-06,
+      "loss": 0.7203,
+      "step": 5811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1305445737084763e-06,
+      "loss": 0.7127,
+      "step": 5812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1298514905551262e-06,
+      "loss": 0.7454,
+      "step": 5813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1291583557076226e-06,
+      "loss": 0.6259,
+      "step": 5814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.128465169222855e-06,
+      "loss": 0.6925,
+      "step": 5815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.127771931157715e-06,
+      "loss": 0.8112,
+      "step": 5816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.127078641569099e-06,
+      "loss": 0.7206,
+      "step": 5817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.126385300513908e-06,
+      "loss": 0.6166,
+      "step": 5818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1256919080490465e-06,
+      "loss": 0.6538,
+      "step": 5819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.124998464231425e-06,
+      "loss": 0.652,
+      "step": 5820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1243049691179557e-06,
+      "loss": 0.6809,
+      "step": 5821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.123611422765556e-06,
+      "loss": 0.763,
+      "step": 5822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1229178252311493e-06,
+      "loss": 0.6753,
+      "step": 5823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.122224176571659e-06,
+      "loss": 0.7394,
+      "step": 5824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1215304768440185e-06,
+      "loss": 0.6637,
+      "step": 5825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.12083672610516e-06,
+      "loss": 0.6241,
+      "step": 5826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1201429244120217e-06,
+      "loss": 0.7697,
+      "step": 5827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1194490718215482e-06,
+      "loss": 0.7894,
+      "step": 5828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.118755168390685e-06,
+      "loss": 0.8019,
+      "step": 5829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.118061214176385e-06,
+      "loss": 0.7682,
+      "step": 5830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1173672092356017e-06,
+      "loss": 0.7578,
+      "step": 5831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1166731536252955e-06,
+      "loss": 0.7392,
+      "step": 5832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1159790474024285e-06,
+      "loss": 0.7441,
+      "step": 5833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1152848906239703e-06,
+      "loss": 0.7846,
+      "step": 5834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1145906833468926e-06,
+      "loss": 0.6883,
+      "step": 5835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1138964256281707e-06,
+      "loss": 0.6583,
+      "step": 5836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1132021175247844e-06,
+      "loss": 0.7956,
+      "step": 5837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1125077590937193e-06,
+      "loss": 0.692,
+      "step": 5838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.111813350391964e-06,
+      "loss": 0.7246,
+      "step": 5839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1111188914765096e-06,
+      "loss": 0.73,
+      "step": 5840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1104243824043532e-06,
+      "loss": 0.7019,
+      "step": 5841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1097298232324973e-06,
+      "loss": 0.6138,
+      "step": 5842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1090352140179447e-06,
+      "loss": 0.7019,
+      "step": 5843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.108340554817706e-06,
+      "loss": 0.7018,
+      "step": 5844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.107645845688793e-06,
+      "loss": 0.7183,
+      "step": 5845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1069510866882235e-06,
+      "loss": 0.7838,
+      "step": 5846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1062562778730194e-06,
+      "loss": 0.8228,
+      "step": 5847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1055614193002053e-06,
+      "loss": 0.6961,
+      "step": 5848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1048665110268123e-06,
+      "loss": 0.6588,
+      "step": 5849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1041715531098715e-06,
+      "loss": 0.6429,
+      "step": 5850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.103476545606423e-06,
+      "loss": 0.8806,
+      "step": 5851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.102781488573507e-06,
+      "loss": 0.7081,
+      "step": 5852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.102086382068169e-06,
+      "loss": 0.7984,
+      "step": 5853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.101391226147461e-06,
+      "loss": 0.7541,
+      "step": 5854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1006960208684343e-06,
+      "loss": 0.7046,
+      "step": 5855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1000007662881483e-06,
+      "loss": 0.655,
+      "step": 5856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.099305462463665e-06,
+      "loss": 0.7154,
+      "step": 5857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0986101094520505e-06,
+      "loss": 0.7609,
+      "step": 5858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.097914707310374e-06,
+      "loss": 0.7374,
+      "step": 5859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.097219256095711e-06,
+      "loss": 0.7609,
+      "step": 5860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.096523755865139e-06,
+      "loss": 0.7871,
+      "step": 5861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.09582820667574e-06,
+      "loss": 0.6817,
+      "step": 5862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0951326085846e-06,
+      "loss": 0.649,
+      "step": 5863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.09443696164881e-06,
+      "loss": 0.7017,
+      "step": 5864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.093741265925463e-06,
+      "loss": 0.7023,
+      "step": 5865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.093045521471658e-06,
+      "loss": 0.7031,
+      "step": 5866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.092349728344498e-06,
+      "loss": 0.9375,
+      "step": 5867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0916538866010882e-06,
+      "loss": 0.6687,
+      "step": 5868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0909579962985387e-06,
+      "loss": 0.6866,
+      "step": 5869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0902620574939644e-06,
+      "loss": 0.7084,
+      "step": 5870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.089566070244483e-06,
+      "loss": 0.6488,
+      "step": 5871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.088870034607216e-06,
+      "loss": 0.6724,
+      "step": 5872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0881739506392914e-06,
+      "loss": 0.7623,
+      "step": 5873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0874778183978373e-06,
+      "loss": 0.7213,
+      "step": 5874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0867816379399885e-06,
+      "loss": 0.7262,
+      "step": 5875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0860854093228838e-06,
+      "loss": 0.8031,
+      "step": 5876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.085389132603664e-06,
+      "loss": 0.7947,
+      "step": 5877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0846928078394754e-06,
+      "loss": 0.6756,
+      "step": 5878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.083996435087468e-06,
+      "loss": 0.6815,
+      "step": 5879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.083300014404796e-06,
+      "loss": 0.711,
+      "step": 5880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.082603545848616e-06,
+      "loss": 0.915,
+      "step": 5881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.081907029476091e-06,
+      "loss": 0.8787,
+      "step": 5882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0812104653443854e-06,
+      "loss": 0.6778,
+      "step": 5883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0805138535106687e-06,
+      "loss": 0.6866,
+      "step": 5884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0798171940321153e-06,
+      "loss": 0.6794,
+      "step": 5885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.079120486965901e-06,
+      "loss": 0.6951,
+      "step": 5886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.07842373236921e-06,
+      "loss": 0.8205,
+      "step": 5887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.077726930299224e-06,
+      "loss": 0.6841,
+      "step": 5888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0770300808131345e-06,
+      "loss": 0.6644,
+      "step": 5889
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0763331839681323e-06,
+      "loss": 0.7642,
+      "step": 5890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.075636239821416e-06,
+      "loss": 0.6732,
+      "step": 5891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0749392484301852e-06,
+      "loss": 0.6177,
+      "step": 5892
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0742422098516445e-06,
+      "loss": 0.587,
+      "step": 5893
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.073545124143003e-06,
+      "loss": 0.7822,
+      "step": 5894
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0728479913614727e-06,
+      "loss": 0.7681,
+      "step": 5895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.07215081156427e-06,
+      "loss": 0.69,
+      "step": 5896
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.071453584808615e-06,
+      "loss": 0.6831,
+      "step": 5897
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.07075631115173e-06,
+      "loss": 0.5548,
+      "step": 5898
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0700589906508454e-06,
+      "loss": 0.5767,
+      "step": 5899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.069361623363191e-06,
+      "loss": 0.5777,
+      "step": 5900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0686642093460015e-06,
+      "loss": 0.7094,
+      "step": 5901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0679667486565185e-06,
+      "loss": 0.6418,
+      "step": 5902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.067269241351983e-06,
+      "loss": 0.7496,
+      "step": 5903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.066571687489643e-06,
+      "loss": 0.734,
+      "step": 5904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0658740871267486e-06,
+      "loss": 0.8024,
+      "step": 5905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0651764403205554e-06,
+      "loss": 0.7095,
+      "step": 5906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0644787471283206e-06,
+      "loss": 0.7397,
+      "step": 5907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.063781007607307e-06,
+      "loss": 0.6685,
+      "step": 5908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.06308322181478e-06,
+      "loss": 0.7593,
+      "step": 5909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0623853898080097e-06,
+      "loss": 0.7142,
+      "step": 5910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.06168751164427e-06,
+      "loss": 0.7198,
+      "step": 5911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0609895873808383e-06,
+      "loss": 0.7674,
+      "step": 5912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0602916170749943e-06,
+      "loss": 0.6711,
+      "step": 5913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0595936007840243e-06,
+      "loss": 0.6489,
+      "step": 5914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0588955385652176e-06,
+      "loss": 0.7881,
+      "step": 5915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0581974304758644e-06,
+      "loss": 0.6889,
+      "step": 5916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0574992765732624e-06,
+      "loss": 0.8327,
+      "step": 5917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0568010769147123e-06,
+      "loss": 0.7626,
+      "step": 5918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0561028315575163e-06,
+      "loss": 0.8748,
+      "step": 5919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.055404540558983e-06,
+      "loss": 0.7115,
+      "step": 5920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0547062039764218e-06,
+      "loss": 0.7595,
+      "step": 5921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0540078218671498e-06,
+      "loss": 0.6838,
+      "step": 5922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.053309394288485e-06,
+      "loss": 0.6821,
+      "step": 5923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0526109212977494e-06,
+      "loss": 0.7813,
+      "step": 5924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0519124029522703e-06,
+      "loss": 0.7103,
+      "step": 5925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.051213839309376e-06,
+      "loss": 0.7506,
+      "step": 5926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0505152304264013e-06,
+      "loss": 0.6273,
+      "step": 5927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0498165763606823e-06,
+      "loss": 0.7744,
+      "step": 5928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0491178771695613e-06,
+      "loss": 0.646,
+      "step": 5929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.048419132910383e-06,
+      "loss": 0.629,
+      "step": 5930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0477203436404944e-06,
+      "loss": 0.6703,
+      "step": 5931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0470215094172495e-06,
+      "loss": 0.6748,
+      "step": 5932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0463226302980024e-06,
+      "loss": 0.6848,
+      "step": 5933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0456237063401146e-06,
+      "loss": 0.7831,
+      "step": 5934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0449247376009467e-06,
+      "loss": 0.7241,
+      "step": 5935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0442257241378675e-06,
+      "loss": 0.6442,
+      "step": 5936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0435266660082473e-06,
+      "loss": 0.8031,
+      "step": 5937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0428275632694594e-06,
+      "loss": 0.7743,
+      "step": 5938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.042128415978883e-06,
+      "loss": 0.7004,
+      "step": 5939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.041429224193898e-06,
+      "loss": 0.6731,
+      "step": 5940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0407299879718903e-06,
+      "loss": 0.5882,
+      "step": 5941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.040030707370249e-06,
+      "loss": 0.8697,
+      "step": 5942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.039331382446366e-06,
+      "loss": 0.7772,
+      "step": 5943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.038632013257638e-06,
+      "loss": 0.7449,
+      "step": 5944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0379325998614646e-06,
+      "loss": 0.6904,
+      "step": 5945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.037233142315249e-06,
+      "loss": 0.6739,
+      "step": 5946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0365336406763974e-06,
+      "loss": 0.6787,
+      "step": 5947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0358340950023212e-06,
+      "loss": 0.637,
+      "step": 5948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0351345053504354e-06,
+      "loss": 0.7999,
+      "step": 5949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.034434871778156e-06,
+      "loss": 0.6018,
+      "step": 5950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0337351943429054e-06,
+      "loss": 0.7822,
+      "step": 5951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.033035473102109e-06,
+      "loss": 0.7949,
+      "step": 5952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.032335708113195e-06,
+      "loss": 0.747,
+      "step": 5953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0316358994335955e-06,
+      "loss": 0.787,
+      "step": 5954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.030936047120746e-06,
+      "loss": 0.5854,
+      "step": 5955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0302361512320868e-06,
+      "loss": 0.7684,
+      "step": 5956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0295362118250594e-06,
+      "loss": 0.7564,
+      "step": 5957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0288362289571123e-06,
+      "loss": 0.6719,
+      "step": 5958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0281362026856936e-06,
+      "loss": 0.6647,
+      "step": 5959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.027436133068258e-06,
+      "loss": 0.7775,
+      "step": 5960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.026736020162262e-06,
+      "loss": 0.635,
+      "step": 5961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0260358640251675e-06,
+      "loss": 0.699,
+      "step": 5962
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0253356647144383e-06,
+      "loss": 0.7458,
+      "step": 5963
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0246354222875416e-06,
+      "loss": 0.7209,
+      "step": 5964
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0239351368019504e-06,
+      "loss": 0.9034,
+      "step": 5965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0232348083151375e-06,
+      "loss": 0.784,
+      "step": 5966
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.022534436884582e-06,
+      "loss": 0.6817,
+      "step": 5967
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.021834022567768e-06,
+      "loss": 0.7539,
+      "step": 5968
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0211335654221776e-06,
+      "loss": 0.7771,
+      "step": 5969
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.020433065505302e-06,
+      "loss": 0.642,
+      "step": 5970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.019732522874633e-06,
+      "loss": 0.7195,
+      "step": 5971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0190319375876676e-06,
+      "loss": 0.6358,
+      "step": 5972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0183313097019035e-06,
+      "loss": 0.8083,
+      "step": 5973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0176306392748457e-06,
+      "loss": 0.7616,
+      "step": 5974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0169299263639995e-06,
+      "loss": 0.6729,
+      "step": 5975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0162291710268747e-06,
+      "loss": 0.7405,
+      "step": 5976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.015528373320986e-06,
+      "loss": 0.7585,
+      "step": 5977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.014827533303849e-06,
+      "loss": 0.676,
+      "step": 5978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0141266510329846e-06,
+      "loss": 0.5809,
+      "step": 5979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0134257265659173e-06,
+      "loss": 0.6539,
+      "step": 5980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0127247599601734e-06,
+      "loss": 0.711,
+      "step": 5981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0120237512732854e-06,
+      "loss": 0.617,
+      "step": 5982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.011322700562786e-06,
+      "loss": 0.6431,
+      "step": 5983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.010621607886214e-06,
+      "loss": 0.7359,
+      "step": 5984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.009920473301109e-06,
+      "loss": 0.6804,
+      "step": 5985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0092192968650168e-06,
+      "loss": 0.6201,
+      "step": 5986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0085180786354863e-06,
+      "loss": 0.7325,
+      "step": 5987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0078168186700666e-06,
+      "loss": 0.7572,
+      "step": 5988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0071155170263146e-06,
+      "loss": 0.6747,
+      "step": 5989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.006414173761788e-06,
+      "loss": 0.8044,
+      "step": 5990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0057127889340487e-06,
+      "loss": 0.6459,
+      "step": 5991
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.005011362600661e-06,
+      "loss": 0.6197,
+      "step": 5992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0043098948191947e-06,
+      "loss": 0.6805,
+      "step": 5993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.003608385647221e-06,
+      "loss": 0.739,
+      "step": 5994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.002906835142315e-06,
+      "loss": 0.7811,
+      "step": 5995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.002205243362057e-06,
+      "loss": 0.7294,
+      "step": 5996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.001503610364027e-06,
+      "loss": 0.5945,
+      "step": 5997
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0008019362058116e-06,
+      "loss": 0.7834,
+      "step": 5998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0001002209449998e-06,
+      "loss": 0.7422,
+      "step": 5999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9993984646391833e-06,
+      "loss": 0.7071,
+      "step": 6000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.998696667345959e-06,
+      "loss": 0.8595,
+      "step": 6001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9979948291229245e-06,
+      "loss": 0.7875,
+      "step": 6002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9972929500276824e-06,
+      "loss": 0.7761,
+      "step": 6003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9965910301178393e-06,
+      "loss": 0.7085,
+      "step": 6004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9958890694510027e-06,
+      "loss": 0.8636,
+      "step": 6005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9951870680847874e-06,
+      "loss": 0.6631,
+      "step": 6006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9944850260768064e-06,
+      "loss": 0.7336,
+      "step": 6007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9937829434846804e-06,
+      "loss": 0.7005,
+      "step": 6008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.993080820366032e-06,
+      "loss": 0.751,
+      "step": 6009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9923786567784864e-06,
+      "loss": 0.798,
+      "step": 6010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.991676452779672e-06,
+      "loss": 0.6891,
+      "step": 6011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.990974208427222e-06,
+      "loss": 0.6296,
+      "step": 6012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.990271923778773e-06,
+      "loss": 0.615,
+      "step": 6013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.989569598891962e-06,
+      "loss": 0.5844,
+      "step": 6014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9888672338244328e-06,
+      "loss": 0.6641,
+      "step": 6015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9881648286338304e-06,
+      "loss": 0.6975,
+      "step": 6016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.987462383377804e-06,
+      "loss": 0.6308,
+      "step": 6017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9867598981140046e-06,
+      "loss": 0.7093,
+      "step": 6018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.986057372900089e-06,
+      "loss": 0.7807,
+      "step": 6019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9853548077937168e-06,
+      "loss": 0.6882,
+      "step": 6020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9846522028525476e-06,
+      "loss": 0.69,
+      "step": 6021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9839495581342483e-06,
+      "loss": 0.7138,
+      "step": 6022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.983246873696487e-06,
+      "loss": 0.6654,
+      "step": 6023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9825441495969352e-06,
+      "loss": 0.7187,
+      "step": 6024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9818413858932687e-06,
+      "loss": 0.808,
+      "step": 6025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9811385826431654e-06,
+      "loss": 0.7341,
+      "step": 6026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9804357399043064e-06,
+      "loss": 0.7715,
+      "step": 6027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9797328577343772e-06,
+      "loss": 0.7625,
+      "step": 6028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9790299361910664e-06,
+      "loss": 0.655,
+      "step": 6029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9783269753320637e-06,
+      "loss": 0.7383,
+      "step": 6030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.977623975215064e-06,
+      "loss": 0.7098,
+      "step": 6031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.976920935897767e-06,
+      "loss": 0.7823,
+      "step": 6032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.976217857437871e-06,
+      "loss": 0.7226,
+      "step": 6033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.975514739893081e-06,
+      "loss": 0.7773,
+      "step": 6034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.974811583321105e-06,
+      "loss": 0.5758,
+      "step": 6035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.974108387779653e-06,
+      "loss": 0.7444,
+      "step": 6036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9734051533264386e-06,
+      "loss": 0.635,
+      "step": 6037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9727018800191793e-06,
+      "loss": 0.7023,
+      "step": 6038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9719985679155957e-06,
+      "loss": 0.6576,
+      "step": 6039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9712952170734095e-06,
+      "loss": 0.705,
+      "step": 6040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9705918275503487e-06,
+      "loss": 0.7612,
+      "step": 6041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9698883994041424e-06,
+      "loss": 0.7334,
+      "step": 6042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9691849326925226e-06,
+      "loss": 0.6589,
+      "step": 6043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9684814274732273e-06,
+      "loss": 0.6288,
+      "step": 6044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.967777883803994e-06,
+      "loss": 0.7784,
+      "step": 6045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.967074301742565e-06,
+      "loss": 0.6728,
+      "step": 6046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.966370681346687e-06,
+      "loss": 0.6507,
+      "step": 6047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.965667022674108e-06,
+      "loss": 0.7205,
+      "step": 6048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9649633257825795e-06,
+      "loss": 0.9685,
+      "step": 6049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9642595907298564e-06,
+      "loss": 0.6387,
+      "step": 6050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9635558175736984e-06,
+      "loss": 0.6182,
+      "step": 6051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.962852006371864e-06,
+      "loss": 0.6742,
+      "step": 6052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9621481571821194e-06,
+      "loss": 0.789,
+      "step": 6053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.961444270062231e-06,
+      "loss": 0.6524,
+      "step": 6054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9607403450699696e-06,
+      "loss": 0.6334,
+      "step": 6055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.960036382263109e-06,
+      "loss": 0.6719,
+      "step": 6056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.959332381699426e-06,
+      "loss": 0.6593,
+      "step": 6057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9586283434367e-06,
+      "loss": 0.7521,
+      "step": 6058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.957924267532714e-06,
+      "loss": 0.7185,
+      "step": 6059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9572201540452556e-06,
+      "loss": 0.7689,
+      "step": 6060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.956516003032111e-06,
+      "loss": 0.7606,
+      "step": 6061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9558118145510746e-06,
+      "loss": 0.729,
+      "step": 6062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.955107588659941e-06,
+      "loss": 0.7703,
+      "step": 6063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9544033254165083e-06,
+      "loss": 0.655,
+      "step": 6064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9536990248785778e-06,
+      "loss": 0.6436,
+      "step": 6065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9529946871039545e-06,
+      "loss": 0.6625,
+      "step": 6066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952290312150446e-06,
+      "loss": 0.631,
+      "step": 6067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.951585900075862e-06,
+      "loss": 0.7663,
+      "step": 6068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9508814509380164e-06,
+      "loss": 0.7532,
+      "step": 6069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9501769647947266e-06,
+      "loss": 0.5993,
+      "step": 6070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9494724417038113e-06,
+      "loss": 0.6683,
+      "step": 6071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9487678817230937e-06,
+      "loss": 0.7441,
+      "step": 6072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9480632849103995e-06,
+      "loss": 0.6426,
+      "step": 6073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9473586513235577e-06,
+      "loss": 0.5941,
+      "step": 6074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9466539810203994e-06,
+      "loss": 0.7749,
+      "step": 6075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9459492740587598e-06,
+      "loss": 0.8429,
+      "step": 6076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.945244530496477e-06,
+      "loss": 0.7837,
+      "step": 6077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9445397503913914e-06,
+      "loss": 0.9136,
+      "step": 6078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.943834933801348e-06,
+      "loss": 0.9992,
+      "step": 6079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9431300807841912e-06,
+      "loss": 0.6957,
+      "step": 6080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9424251913977726e-06,
+      "loss": 0.8248,
+      "step": 6081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.941720265699945e-06,
+      "loss": 0.6544,
+      "step": 6082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.941015303748563e-06,
+      "loss": 0.756,
+      "step": 6083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9403103056014858e-06,
+      "loss": 0.6993,
+      "step": 6084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9396052713165755e-06,
+      "loss": 0.7152,
+      "step": 6085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9389002009516977e-06,
+      "loss": 0.6926,
+      "step": 6086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.938195094564718e-06,
+      "loss": 0.6777,
+      "step": 6087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.937489952213508e-06,
+      "loss": 0.7115,
+      "step": 6088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.936784773955942e-06,
+      "loss": 0.6262,
+      "step": 6089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.936079559849895e-06,
+      "loss": 0.7042,
+      "step": 6090
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.8121260404586792,
+      "eval_runtime": 437.0349,
+      "eval_samples_per_second": 6.862,
+      "eval_steps_per_second": 0.858,
+      "step": 6090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9353743099532472e-06,
+      "loss": 0.7456,
+      "step": 6091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9346690243238807e-06,
+      "loss": 0.6847,
+      "step": 6092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9339637030196803e-06,
+      "loss": 0.6917,
+      "step": 6093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.933258346098536e-06,
+      "loss": 0.7872,
+      "step": 6094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9325529536183374e-06,
+      "loss": 0.627,
+      "step": 6095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.931847525636979e-06,
+      "loss": 0.7526,
+      "step": 6096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9311420622123576e-06,
+      "loss": 0.6763,
+      "step": 6097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9304365634023742e-06,
+      "loss": 0.6109,
+      "step": 6098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.92973102926493e-06,
+      "loss": 0.6846,
+      "step": 6099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.929025459857931e-06,
+      "loss": 0.7777,
+      "step": 6100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9283198552392873e-06,
+      "loss": 0.712,
+      "step": 6101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9276142154669084e-06,
+      "loss": 0.6175,
+      "step": 6102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9269085405987095e-06,
+      "loss": 0.5999,
+      "step": 6103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9262028306926078e-06,
+      "loss": 0.6818,
+      "step": 6104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.925497085806524e-06,
+      "loss": 0.64,
+      "step": 6105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.924791305998381e-06,
+      "loss": 0.759,
+      "step": 6106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.924085491326103e-06,
+      "loss": 0.6457,
+      "step": 6107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9233796418476208e-06,
+      "loss": 0.7757,
+      "step": 6108
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9226737576208645e-06,
+      "loss": 0.6767,
+      "step": 6109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.92196783870377e-06,
+      "loss": 0.6176,
+      "step": 6110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9212618851542735e-06,
+      "loss": 0.6652,
+      "step": 6111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.920555897030315e-06,
+      "loss": 0.6879,
+      "step": 6112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9198498743898383e-06,
+      "loss": 0.7173,
+      "step": 6113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9191438172907883e-06,
+      "loss": 0.7976,
+      "step": 6114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9184377257911146e-06,
+      "loss": 0.6892,
+      "step": 6115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9177315999487675e-06,
+      "loss": 0.6407,
+      "step": 6116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9170254398217023e-06,
+      "loss": 0.6956,
+      "step": 6117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9163192454678758e-06,
+      "loss": 0.6841,
+      "step": 6118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9156130169452467e-06,
+      "loss": 0.6849,
+      "step": 6119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9149067543117797e-06,
+      "loss": 0.6835,
+      "step": 6120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9142004576254392e-06,
+      "loss": 0.7109,
+      "step": 6121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9134941269441925e-06,
+      "loss": 0.7647,
+      "step": 6122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9127877623260126e-06,
+      "loss": 0.5653,
+      "step": 6123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9120813638288724e-06,
+      "loss": 0.6606,
+      "step": 6124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9113749315107486e-06,
+      "loss": 0.6619,
+      "step": 6125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9106684654296204e-06,
+      "loss": 0.671,
+      "step": 6126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9099619656434707e-06,
+      "loss": 0.708,
+      "step": 6127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9092554322102835e-06,
+      "loss": 0.6924,
+      "step": 6128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.908548865188048e-06,
+      "loss": 0.746,
+      "step": 6129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.907842264634752e-06,
+      "loss": 0.7327,
+      "step": 6130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9071356306083915e-06,
+      "loss": 0.7127,
+      "step": 6131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9064289631669607e-06,
+      "loss": 0.6716,
+      "step": 6132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.905722262368459e-06,
+      "loss": 0.7826,
+      "step": 6133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9050155282708885e-06,
+      "loss": 0.6058,
+      "step": 6134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9043087609322524e-06,
+      "loss": 0.6912,
+      "step": 6135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.903601960410558e-06,
+      "loss": 0.7217,
+      "step": 6136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9028951267638146e-06,
+      "loss": 0.6425,
+      "step": 6137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9021882600500344e-06,
+      "loss": 0.6617,
+      "step": 6138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9014813603272336e-06,
+      "loss": 0.6945,
+      "step": 6139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.900774427653429e-06,
+      "loss": 0.7157,
+      "step": 6140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9000674620866408e-06,
+      "loss": 0.6213,
+      "step": 6141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8993604636848926e-06,
+      "loss": 0.7042,
+      "step": 6142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.898653432506211e-06,
+      "loss": 0.7182,
+      "step": 6143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.897946368608624e-06,
+      "loss": 0.7286,
+      "step": 6144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.897239272050162e-06,
+      "loss": 0.7133,
+      "step": 6145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8965321428888603e-06,
+      "loss": 0.7318,
+      "step": 6146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8958249811827547e-06,
+      "loss": 0.6894,
+      "step": 6147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8951177869898854e-06,
+      "loss": 0.709,
+      "step": 6148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8944105603682924e-06,
+      "loss": 0.6386,
+      "step": 6149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8937033013760224e-06,
+      "loss": 0.7321,
+      "step": 6150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8929960100711213e-06,
+      "loss": 0.6219,
+      "step": 6151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8922886865116393e-06,
+      "loss": 0.7559,
+      "step": 6152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8915813307556303e-06,
+      "loss": 0.7831,
+      "step": 6153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8908739428611477e-06,
+      "loss": 0.7298,
+      "step": 6154
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8901665228862503e-06,
+      "loss": 0.6908,
+      "step": 6155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8894590708889975e-06,
+      "loss": 0.7123,
+      "step": 6156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8887515869274533e-06,
+      "loss": 0.6625,
+      "step": 6157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.888044071059684e-06,
+      "loss": 0.6886,
+      "step": 6158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8873365233437568e-06,
+      "loss": 0.7932,
+      "step": 6159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.886628943837743e-06,
+      "loss": 0.704,
+      "step": 6160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8859213325997163e-06,
+      "loss": 0.6949,
+      "step": 6161
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8852136896877534e-06,
+      "loss": 0.6593,
+      "step": 6162
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.884506015159932e-06,
+      "loss": 0.7988,
+      "step": 6163
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8837983090743343e-06,
+      "loss": 0.7883,
+      "step": 6164
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.883090571489044e-06,
+      "loss": 0.631,
+      "step": 6165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.882382802462147e-06,
+      "loss": 0.7088,
+      "step": 6166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8816750020517343e-06,
+      "loss": 0.7981,
+      "step": 6167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8809671703158947e-06,
+      "loss": 0.7286,
+      "step": 6168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.880259307312725e-06,
+      "loss": 0.6914,
+      "step": 6169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8795514131003206e-06,
+      "loss": 0.5862,
+      "step": 6170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8788434877367816e-06,
+      "loss": 0.6907,
+      "step": 6171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.87813553128021e-06,
+      "loss": 0.7376,
+      "step": 6172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.87742754378871e-06,
+      "loss": 0.6285,
+      "step": 6173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.876719525320389e-06,
+      "loss": 0.5446,
+      "step": 6174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8760114759333556e-06,
+      "loss": 0.6639,
+      "step": 6175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.875303395685723e-06,
+      "loss": 0.7261,
+      "step": 6176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.874595284635606e-06,
+      "loss": 0.698,
+      "step": 6177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.873887142841121e-06,
+      "loss": 0.7158,
+      "step": 6178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8731789703603876e-06,
+      "loss": 0.7558,
+      "step": 6179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8724707672515285e-06,
+      "loss": 0.6753,
+      "step": 6180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.871762533572669e-06,
+      "loss": 0.6136,
+      "step": 6181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8710542693819347e-06,
+      "loss": 0.7176,
+      "step": 6182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8703459747374574e-06,
+      "loss": 0.5764,
+      "step": 6183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8696376496973683e-06,
+      "loss": 0.7165,
+      "step": 6184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8689292943198023e-06,
+      "loss": 0.857,
+      "step": 6185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8682209086628964e-06,
+      "loss": 0.7197,
+      "step": 6186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.86751249278479e-06,
+      "loss": 0.7314,
+      "step": 6187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8668040467436266e-06,
+      "loss": 0.6919,
+      "step": 6188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.866095570597549e-06,
+      "loss": 0.692,
+      "step": 6189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8653870644047066e-06,
+      "loss": 0.6844,
+      "step": 6190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8646785282232477e-06,
+      "loss": 0.7403,
+      "step": 6191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8639699621113244e-06,
+      "loss": 0.8604,
+      "step": 6192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.863261366127092e-06,
+      "loss": 0.622,
+      "step": 6193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8625527403287064e-06,
+      "loss": 0.7095,
+      "step": 6194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8618440847743267e-06,
+      "loss": 0.7203,
+      "step": 6195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.861135399522117e-06,
+      "loss": 0.6796,
+      "step": 6196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.860426684630239e-06,
+      "loss": 0.6309,
+      "step": 6197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.859717940156862e-06,
+      "loss": 0.6291,
+      "step": 6198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8590091661601533e-06,
+      "loss": 0.6766,
+      "step": 6199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.858300362698285e-06,
+      "loss": 0.5998,
+      "step": 6200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8575915298294317e-06,
+      "loss": 0.6778,
+      "step": 6201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8568826676117687e-06,
+      "loss": 0.6209,
+      "step": 6202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8561737761034765e-06,
+      "loss": 0.6848,
+      "step": 6203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8554648553627344e-06,
+      "loss": 0.7092,
+      "step": 6204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8547559054477274e-06,
+      "loss": 0.8027,
+      "step": 6205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8540469264166406e-06,
+      "loss": 0.6696,
+      "step": 6206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8533379183276637e-06,
+      "loss": 0.7691,
+      "step": 6207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.852628881238986e-06,
+      "loss": 0.795,
+      "step": 6208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8519198152088023e-06,
+      "loss": 0.7939,
+      "step": 6209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8512107202953073e-06,
+      "loss": 0.7408,
+      "step": 6210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8505015965566984e-06,
+      "loss": 0.754,
+      "step": 6211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8497924440511772e-06,
+      "loss": 0.6982,
+      "step": 6212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8490832628369452e-06,
+      "loss": 0.6466,
+      "step": 6213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8483740529722075e-06,
+      "loss": 0.7604,
+      "step": 6214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8476648145151725e-06,
+      "loss": 0.8524,
+      "step": 6215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.846955547524049e-06,
+      "loss": 0.7457,
+      "step": 6216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.84624625205705e-06,
+      "loss": 0.7279,
+      "step": 6217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8455369281723884e-06,
+      "loss": 0.8509,
+      "step": 6218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8448275759282825e-06,
+      "loss": 0.8175,
+      "step": 6219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.84411819538295e-06,
+      "loss": 0.7024,
+      "step": 6220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.843408786594613e-06,
+      "loss": 0.6907,
+      "step": 6221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.842699349621496e-06,
+      "loss": 0.6605,
+      "step": 6222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8419898845218238e-06,
+      "loss": 0.6195,
+      "step": 6223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.841280391353825e-06,
+      "loss": 0.7382,
+      "step": 6224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8405708701757296e-06,
+      "loss": 0.6745,
+      "step": 6225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.839861321045772e-06,
+      "loss": 0.5914,
+      "step": 6226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8391517440221865e-06,
+      "loss": 0.7652,
+      "step": 6227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8384421391632104e-06,
+      "loss": 0.7591,
+      "step": 6228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.837732506527085e-06,
+      "loss": 0.6551,
+      "step": 6229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.83702284617205e-06,
+      "loss": 0.8156,
+      "step": 6230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.836313158156352e-06,
+      "loss": 0.7233,
+      "step": 6231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.835603442538236e-06,
+      "loss": 0.664,
+      "step": 6232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.834893699375951e-06,
+      "loss": 0.5802,
+      "step": 6233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8341839287277496e-06,
+      "loss": 0.6888,
+      "step": 6234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8334741306518838e-06,
+      "loss": 0.6706,
+      "step": 6235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.83276430520661e-06,
+      "loss": 0.6578,
+      "step": 6236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8320544524501856e-06,
+      "loss": 0.5742,
+      "step": 6237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8313445724408716e-06,
+      "loss": 0.7285,
+      "step": 6238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8306346652369286e-06,
+      "loss": 0.8002,
+      "step": 6239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8299247308966233e-06,
+      "loss": 0.6525,
+      "step": 6240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8292147694782213e-06,
+      "loss": 0.6896,
+      "step": 6241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8285047810399914e-06,
+      "loss": 0.5932,
+      "step": 6242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.827794765640206e-06,
+      "loss": 0.6393,
+      "step": 6243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8270847233371375e-06,
+      "loss": 0.7077,
+      "step": 6244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8263746541890623e-06,
+      "loss": 0.5774,
+      "step": 6245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8256645582542584e-06,
+      "loss": 0.6892,
+      "step": 6246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8249544355910054e-06,
+      "loss": 0.6995,
+      "step": 6247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.824244286257586e-06,
+      "loss": 1.0583,
+      "step": 6248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8235341103122837e-06,
+      "loss": 0.8015,
+      "step": 6249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8228239078133874e-06,
+      "loss": 0.7309,
+      "step": 6250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8221136788191833e-06,
+      "loss": 0.6653,
+      "step": 6251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.821403423387964e-06,
+      "loss": 0.6229,
+      "step": 6252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8206931415780216e-06,
+      "loss": 0.7644,
+      "step": 6253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8199828334476527e-06,
+      "loss": 0.743,
+      "step": 6254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8192724990551547e-06,
+      "loss": 0.8206,
+      "step": 6255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8185621384588264e-06,
+      "loss": 0.7437,
+      "step": 6256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.81785175171697e-06,
+      "loss": 0.6978,
+      "step": 6257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.81714133888789e-06,
+      "loss": 0.6492,
+      "step": 6258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.816430900029891e-06,
+      "loss": 0.8861,
+      "step": 6259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8157204352012835e-06,
+      "loss": 0.6397,
+      "step": 6260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8150099444603758e-06,
+      "loss": 0.7469,
+      "step": 6261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8142994278654814e-06,
+      "loss": 0.5966,
+      "step": 6262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.813588885474915e-06,
+      "loss": 0.6986,
+      "step": 6263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.812878317346993e-06,
+      "loss": 0.8502,
+      "step": 6264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8121677235400346e-06,
+      "loss": 0.6836,
+      "step": 6265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.81145710411236e-06,
+      "loss": 0.6764,
+      "step": 6266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.810746459122294e-06,
+      "loss": 0.6553,
+      "step": 6267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8100357886281592e-06,
+      "loss": 0.7545,
+      "step": 6268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8093250926882853e-06,
+      "loss": 0.6604,
+      "step": 6269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8086143713609998e-06,
+      "loss": 0.7229,
+      "step": 6270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8079036247046347e-06,
+      "loss": 0.6813,
+      "step": 6271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8071928527775245e-06,
+      "loss": 0.6373,
+      "step": 6272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.806482055638003e-06,
+      "loss": 0.7528,
+      "step": 6273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.80577123334441e-06,
+      "loss": 0.6626,
+      "step": 6274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8050603859550832e-06,
+      "loss": 0.7718,
+      "step": 6275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.804349513528366e-06,
+      "loss": 0.7315,
+      "step": 6276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.803638616122601e-06,
+      "loss": 0.6769,
+      "step": 6277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.802927693796134e-06,
+      "loss": 0.7339,
+      "step": 6278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8022167466073146e-06,
+      "loss": 0.6973,
+      "step": 6279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.801505774614491e-06,
+      "loss": 0.7169,
+      "step": 6280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8007947778760154e-06,
+      "loss": 0.7005,
+      "step": 6281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8000837564502425e-06,
+      "loss": 0.7542,
+      "step": 6282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.799372710395528e-06,
+      "loss": 0.71,
+      "step": 6283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7986616397702306e-06,
+      "loss": 0.6328,
+      "step": 6284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7979505446327094e-06,
+      "loss": 0.691,
+      "step": 6285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7972394250413274e-06,
+      "loss": 0.723,
+      "step": 6286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.796528281054448e-06,
+      "loss": 0.7078,
+      "step": 6287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7958171127304382e-06,
+      "loss": 0.7042,
+      "step": 6288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7951059201276652e-06,
+      "loss": 0.7754,
+      "step": 6289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.794394703304499e-06,
+      "loss": 0.7392,
+      "step": 6290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.793683462319312e-06,
+      "loss": 0.7435,
+      "step": 6291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.792972197230479e-06,
+      "loss": 0.6529,
+      "step": 6292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.792260908096376e-06,
+      "loss": 0.7448,
+      "step": 6293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7915495949753796e-06,
+      "loss": 0.6903,
+      "step": 6294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.790838257925871e-06,
+      "loss": 0.7987,
+      "step": 6295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7901268970062323e-06,
+      "loss": 0.7968,
+      "step": 6296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7894155122748464e-06,
+      "loss": 0.935,
+      "step": 6297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7887041037901e-06,
+      "loss": 0.6937,
+      "step": 6298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7879926716103804e-06,
+      "loss": 0.6235,
+      "step": 6299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7872812157940774e-06,
+      "loss": 0.6643,
+      "step": 6300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7865697363995836e-06,
+      "loss": 0.6634,
+      "step": 6301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7858582334852926e-06,
+      "loss": 0.7952,
+      "step": 6302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.785146707109598e-06,
+      "loss": 0.6126,
+      "step": 6303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7844351573308993e-06,
+      "loss": 0.7348,
+      "step": 6304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7837235842075956e-06,
+      "loss": 0.6982,
+      "step": 6305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.783011987798088e-06,
+      "loss": 0.6656,
+      "step": 6306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7823003681607792e-06,
+      "loss": 0.7092,
+      "step": 6307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.781588725354075e-06,
+      "loss": 0.7041,
+      "step": 6308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7808770594363823e-06,
+      "loss": 0.6274,
+      "step": 6309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7801653704661096e-06,
+      "loss": 0.6943,
+      "step": 6310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7794536585016684e-06,
+      "loss": 0.6657,
+      "step": 6311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7787419236014725e-06,
+      "loss": 0.7424,
+      "step": 6312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.778030165823934e-06,
+      "loss": 0.6502,
+      "step": 6313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.777318385227471e-06,
+      "loss": 0.6154,
+      "step": 6314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.776606581870501e-06,
+      "loss": 0.6996,
+      "step": 6315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.775894755811446e-06,
+      "loss": 0.6646,
+      "step": 6316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.775182907108726e-06,
+      "loss": 0.684,
+      "step": 6317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.774471035820766e-06,
+      "loss": 0.7214,
+      "step": 6318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7737591420059913e-06,
+      "loss": 0.6674,
+      "step": 6319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7730472257228296e-06,
+      "loss": 0.6728,
+      "step": 6320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7723352870297117e-06,
+      "loss": 0.7484,
+      "step": 6321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7716233259850672e-06,
+      "loss": 0.7238,
+      "step": 6322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.77091134264733e-06,
+      "loss": 0.6236,
+      "step": 6323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.770199337074936e-06,
+      "loss": 0.7482,
+      "step": 6324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.76948730932632e-06,
+      "loss": 0.6125,
+      "step": 6325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.768775259459922e-06,
+      "loss": 0.6542,
+      "step": 6326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7680631875341824e-06,
+      "loss": 0.6918,
+      "step": 6327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.767351093607543e-06,
+      "loss": 0.6834,
+      "step": 6328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.766638977738449e-06,
+      "loss": 0.6901,
+      "step": 6329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7659268399853447e-06,
+      "loss": 0.6463,
+      "step": 6330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7652146804066793e-06,
+      "loss": 0.7005,
+      "step": 6331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.764502499060901e-06,
+      "loss": 0.8979,
+      "step": 6332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7637902960064617e-06,
+      "loss": 0.6558,
+      "step": 6333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.763078071301814e-06,
+      "loss": 0.6809,
+      "step": 6334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7623658250054134e-06,
+      "loss": 0.7613,
+      "step": 6335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7616535571757163e-06,
+      "loss": 0.6241,
+      "step": 6336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7609412678711806e-06,
+      "loss": 0.7077,
+      "step": 6337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7602289571502668e-06,
+      "loss": 0.7765,
+      "step": 6338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7595166250714367e-06,
+      "loss": 0.7354,
+      "step": 6339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.758804271693154e-06,
+      "loss": 0.6856,
+      "step": 6340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7580918970738833e-06,
+      "loss": 0.7617,
+      "step": 6341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7573795012720934e-06,
+      "loss": 0.64,
+      "step": 6342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7566670843462516e-06,
+      "loss": 0.7173,
+      "step": 6343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7559546463548293e-06,
+      "loss": 0.7494,
+      "step": 6344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.755242187356299e-06,
+      "loss": 0.6818,
+      "step": 6345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7545297074091338e-06,
+      "loss": 0.8182,
+      "step": 6346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7538172065718092e-06,
+      "loss": 0.7338,
+      "step": 6347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.753104684902804e-06,
+      "loss": 0.8005,
+      "step": 6348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7523921424605972e-06,
+      "loss": 0.5998,
+      "step": 6349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7516795793036693e-06,
+      "loss": 0.7121,
+      "step": 6350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7509669954905026e-06,
+      "loss": 0.7559,
+      "step": 6351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7502543910795826e-06,
+      "loss": 0.722,
+      "step": 6352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.749541766129394e-06,
+      "loss": 0.7054,
+      "step": 6353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.748829120698424e-06,
+      "loss": 0.7941,
+      "step": 6354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7481164548451637e-06,
+      "loss": 0.6894,
+      "step": 6355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.747403768628103e-06,
+      "loss": 0.726,
+      "step": 6356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7466910621057346e-06,
+      "loss": 0.8562,
+      "step": 6357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7459783353365532e-06,
+      "loss": 0.7866,
+      "step": 6358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.745265588379055e-06,
+      "loss": 0.5998,
+      "step": 6359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.744552821291737e-06,
+      "loss": 0.6378,
+      "step": 6360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.743840034133099e-06,
+      "loss": 0.6725,
+      "step": 6361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7431272269616428e-06,
+      "loss": 0.992,
+      "step": 6362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7424143998358687e-06,
+      "loss": 0.6205,
+      "step": 6363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7417015528142833e-06,
+      "loss": 0.6417,
+      "step": 6364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7409886859553914e-06,
+      "loss": 0.6544,
+      "step": 6365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7402757993177003e-06,
+      "loss": 0.7426,
+      "step": 6366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7395628929597197e-06,
+      "loss": 0.8423,
+      "step": 6367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.73884996693996e-06,
+      "loss": 0.6584,
+      "step": 6368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7381370213169343e-06,
+      "loss": 0.7985,
+      "step": 6369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7374240561491556e-06,
+      "loss": 0.6159,
+      "step": 6370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7367110714951408e-06,
+      "loss": 0.7127,
+      "step": 6371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.735998067413405e-06,
+      "loss": 0.7964,
+      "step": 6372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.735285043962469e-06,
+      "loss": 0.6334,
+      "step": 6373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.734572001200852e-06,
+      "loss": 0.6926,
+      "step": 6374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.733858939187076e-06,
+      "loss": 0.5541,
+      "step": 6375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7331458579796655e-06,
+      "loss": 0.8178,
+      "step": 6376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7324327576371445e-06,
+      "loss": 0.8299,
+      "step": 6377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.731719638218041e-06,
+      "loss": 0.6941,
+      "step": 6378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7310064997808815e-06,
+      "loss": 0.8764,
+      "step": 6379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.730293342384197e-06,
+      "loss": 0.6836,
+      "step": 6380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7295801660865197e-06,
+      "loss": 0.7338,
+      "step": 6381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.72886697094638e-06,
+      "loss": 0.7109,
+      "step": 6382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7281537570223145e-06,
+      "loss": 0.6003,
+      "step": 6383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7274405243728575e-06,
+      "loss": 0.6422,
+      "step": 6384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.726727273056548e-06,
+      "loss": 0.7775,
+      "step": 6385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.726014003131925e-06,
+      "loss": 0.6183,
+      "step": 6386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7253007146575278e-06,
+      "loss": 0.7555,
+      "step": 6387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7245874076919004e-06,
+      "loss": 0.6436,
+      "step": 6388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.723874082293585e-06,
+      "loss": 0.6801,
+      "step": 6389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7231607385211274e-06,
+      "loss": 0.6012,
+      "step": 6390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.722447376433074e-06,
+      "loss": 0.6756,
+      "step": 6391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.721733996087972e-06,
+      "loss": 0.741,
+      "step": 6392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7210205975443736e-06,
+      "loss": 0.6927,
+      "step": 6393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.720307180860828e-06,
+      "loss": 0.794,
+      "step": 6394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7195937460958877e-06,
+      "loss": 0.8032,
+      "step": 6395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7188802933081077e-06,
+      "loss": 0.6822,
+      "step": 6396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7181668225560443e-06,
+      "loss": 0.7377,
+      "step": 6397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7174533338982524e-06,
+      "loss": 0.6832,
+      "step": 6398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7167398273932917e-06,
+      "loss": 0.7389,
+      "step": 6399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.716026303099723e-06,
+      "loss": 0.6482,
+      "step": 6400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7153127610761064e-06,
+      "loss": 0.7494,
+      "step": 6401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.714599201381006e-06,
+      "loss": 0.6198,
+      "step": 6402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.713885624072986e-06,
+      "loss": 0.7724,
+      "step": 6403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.713172029210611e-06,
+      "loss": 0.7591,
+      "step": 6404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.71245841685245e-06,
+      "loss": 0.6774,
+      "step": 6405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7117447870570703e-06,
+      "loss": 0.657,
+      "step": 6406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7110311398830424e-06,
+      "loss": 0.6634,
+      "step": 6407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.710317475388938e-06,
+      "loss": 0.72,
+      "step": 6408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7096037936333316e-06,
+      "loss": 0.739,
+      "step": 6409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7088900946747946e-06,
+      "loss": 0.857,
+      "step": 6410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.708176378571905e-06,
+      "loss": 0.6029,
+      "step": 6411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.70746264538324e-06,
+      "loss": 0.5934,
+      "step": 6412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7067488951673765e-06,
+      "loss": 0.8119,
+      "step": 6413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7060351279828954e-06,
+      "loss": 0.7007,
+      "step": 6414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.705321343888379e-06,
+      "loss": 0.7101,
+      "step": 6415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.70460754294241e-06,
+      "loss": 0.7149,
+      "step": 6416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7038937252035713e-06,
+      "loss": 0.6375,
+      "step": 6417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.703179890730449e-06,
+      "loss": 0.7706,
+      "step": 6418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.702466039581631e-06,
+      "loss": 0.5633,
+      "step": 6419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.701752171815704e-06,
+      "loss": 0.7388,
+      "step": 6420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7010382874912593e-06,
+      "loss": 0.6562,
+      "step": 6421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7003243866668864e-06,
+      "loss": 0.6759,
+      "step": 6422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6996104694011787e-06,
+      "loss": 0.6673,
+      "step": 6423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6988965357527297e-06,
+      "loss": 0.6919,
+      "step": 6424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.698182585780134e-06,
+      "loss": 0.7988,
+      "step": 6425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.697468619541989e-06,
+      "loss": 0.8539,
+      "step": 6426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6967546370968915e-06,
+      "loss": 0.6635,
+      "step": 6427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6960406385034416e-06,
+      "loss": 0.6765,
+      "step": 6428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6953266238202387e-06,
+      "loss": 0.7926,
+      "step": 6429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.694612593105884e-06,
+      "loss": 0.8061,
+      "step": 6430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6938985464189825e-06,
+      "loss": 0.7505,
+      "step": 6431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.693184483818137e-06,
+      "loss": 0.7168,
+      "step": 6432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6924704053619534e-06,
+      "loss": 0.7555,
+      "step": 6433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6917563111090388e-06,
+      "loss": 0.7042,
+      "step": 6434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6910422011180025e-06,
+      "loss": 0.821,
+      "step": 6435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6903280754474525e-06,
+      "loss": 0.8681,
+      "step": 6436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.689613934156e-06,
+      "loss": 0.6511,
+      "step": 6437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.688899777302258e-06,
+      "loss": 0.7467,
+      "step": 6438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6881856049448384e-06,
+      "loss": 0.6981,
+      "step": 6439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6874714171423575e-06,
+      "loss": 0.7439,
+      "step": 6440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6867572139534304e-06,
+      "loss": 0.7396,
+      "step": 6441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6860429954366734e-06,
+      "loss": 0.7124,
+      "step": 6442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6853287616507064e-06,
+      "loss": 0.5813,
+      "step": 6443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6846145126541485e-06,
+      "loss": 0.7795,
+      "step": 6444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6839002485056214e-06,
+      "loss": 0.6181,
+      "step": 6445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6831859692637455e-06,
+      "loss": 0.6775,
+      "step": 6446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6824716749871467e-06,
+      "loss": 0.6165,
+      "step": 6447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6817573657344476e-06,
+      "loss": 0.7135,
+      "step": 6448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6810430415642747e-06,
+      "loss": 0.7401,
+      "step": 6449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.680328702535256e-06,
+      "loss": 0.6712,
+      "step": 6450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.679614348706018e-06,
+      "loss": 0.7521,
+      "step": 6451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.678899980135192e-06,
+      "loss": 0.7145,
+      "step": 6452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6781855968814084e-06,
+      "loss": 0.762,
+      "step": 6453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.677471199003299e-06,
+      "loss": 0.7245,
+      "step": 6454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6767567865594967e-06,
+      "loss": 0.766,
+      "step": 6455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.676042359608636e-06,
+      "loss": 0.5918,
+      "step": 6456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.675327918209353e-06,
+      "loss": 0.7564,
+      "step": 6457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6746134624202835e-06,
+      "loss": 0.8139,
+      "step": 6458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.673898992300066e-06,
+      "loss": 0.7274,
+      "step": 6459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.67318450790734e-06,
+      "loss": 0.6911,
+      "step": 6460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6724700093007443e-06,
+      "loss": 0.7253,
+      "step": 6461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.671755496538922e-06,
+      "loss": 0.6563,
+      "step": 6462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6710409696805145e-06,
+      "loss": 0.7553,
+      "step": 6463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6703264287841672e-06,
+      "loss": 0.5673,
+      "step": 6464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.669611873908523e-06,
+      "loss": 0.712,
+      "step": 6465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6688973051122296e-06,
+      "loss": 0.6571,
+      "step": 6466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.668182722453933e-06,
+      "loss": 0.7618,
+      "step": 6467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.667468125992282e-06,
+      "loss": 0.7566,
+      "step": 6468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6667535157859264e-06,
+      "loss": 0.6277,
+      "step": 6469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6660388918935165e-06,
+      "loss": 0.7833,
+      "step": 6470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6653242543737035e-06,
+      "loss": 0.6029,
+      "step": 6471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.664609603285141e-06,
+      "loss": 0.6466,
+      "step": 6472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.663894938686483e-06,
+      "loss": 0.7305,
+      "step": 6473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6631802606363845e-06,
+      "loss": 0.7455,
+      "step": 6474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6624655691935007e-06,
+      "loss": 0.7211,
+      "step": 6475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.66175086441649e-06,
+      "loss": 0.6283,
+      "step": 6476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6610361463640104e-06,
+      "loss": 0.6976,
+      "step": 6477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.660321415094722e-06,
+      "loss": 0.7944,
+      "step": 6478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6596066706672843e-06,
+      "loss": 0.6586,
+      "step": 6479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6588919131403583e-06,
+      "loss": 0.6866,
+      "step": 6480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6581771425726088e-06,
+      "loss": 0.5887,
+      "step": 6481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.657462359022698e-06,
+      "loss": 0.7113,
+      "step": 6482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.656747562549292e-06,
+      "loss": 0.7433,
+      "step": 6483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6560327532110545e-06,
+      "loss": 0.7341,
+      "step": 6484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6553179310666553e-06,
+      "loss": 0.7127,
+      "step": 6485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.65460309617476e-06,
+      "loss": 0.7772,
+      "step": 6486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.653888248594039e-06,
+      "loss": 0.6679,
+      "step": 6487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6531733883831616e-06,
+      "loss": 0.5585,
+      "step": 6488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6524585156007996e-06,
+      "loss": 0.8044,
+      "step": 6489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6517436303056243e-06,
+      "loss": 0.614,
+      "step": 6490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.65102873255631e-06,
+      "loss": 0.7249,
+      "step": 6491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6503138224115303e-06,
+      "loss": 0.7219,
+      "step": 6492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.64959889992996e-06,
+      "loss": 0.7882,
+      "step": 6493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.648883965170276e-06,
+      "loss": 0.6449,
+      "step": 6494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.648169018191156e-06,
+      "loss": 0.7702,
+      "step": 6495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.647454059051276e-06,
+      "loss": 0.6306,
+      "step": 6496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.646739087809318e-06,
+      "loss": 0.8586,
+      "step": 6497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6460241045239605e-06,
+      "loss": 0.7155,
+      "step": 6498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6453091092538852e-06,
+      "loss": 0.673,
+      "step": 6499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.644594102057774e-06,
+      "loss": 0.6452,
+      "step": 6500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.643879082994311e-06,
+      "loss": 0.8086,
+      "step": 6501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6431640521221803e-06,
+      "loss": 0.6882,
+      "step": 6502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6424490095000655e-06,
+      "loss": 0.65,
+      "step": 6503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6417339551866544e-06,
+      "loss": 0.6026,
+      "step": 6504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.641018889240633e-06,
+      "loss": 0.6363,
+      "step": 6505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.64030381172069e-06,
+      "loss": 0.6983,
+      "step": 6506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.639588722685514e-06,
+      "loss": 0.7675,
+      "step": 6507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6388736221937944e-06,
+      "loss": 0.6571,
+      "step": 6508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6381585103042225e-06,
+      "loss": 0.6945,
+      "step": 6509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.63744338707549e-06,
+      "loss": 0.6877,
+      "step": 6510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6367282525662907e-06,
+      "loss": 1.0139,
+      "step": 6511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.636013106835317e-06,
+      "loss": 0.7283,
+      "step": 6512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6352979499412633e-06,
+      "loss": 0.6582,
+      "step": 6513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.634582781942826e-06,
+      "loss": 0.6755,
+      "step": 6514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6338676028987e-06,
+      "loss": 0.6476,
+      "step": 6515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.633152412867585e-06,
+      "loss": 0.6122,
+      "step": 6516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.632437211908177e-06,
+      "loss": 0.7358,
+      "step": 6517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.631722000079176e-06,
+      "loss": 0.6996,
+      "step": 6518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6310067774392812e-06,
+      "loss": 0.7881,
+      "step": 6519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6302915440471944e-06,
+      "loss": 0.7895,
+      "step": 6520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.629576299961618e-06,
+      "loss": 0.6815,
+      "step": 6521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6288610452412527e-06,
+      "loss": 0.6663,
+      "step": 6522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6281457799448034e-06,
+      "loss": 0.5775,
+      "step": 6523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.627430504130974e-06,
+      "loss": 0.8329,
+      "step": 6524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6267152178584694e-06,
+      "loss": 0.6354,
+      "step": 6525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6259999211859965e-06,
+      "loss": 0.7848,
+      "step": 6526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6252846141722615e-06,
+      "loss": 0.6202,
+      "step": 6527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.624569296875972e-06,
+      "loss": 0.6888,
+      "step": 6528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.623853969355838e-06,
+      "loss": 0.655,
+      "step": 6529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.623138631670568e-06,
+      "loss": 0.7258,
+      "step": 6530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.622423283878872e-06,
+      "loss": 0.7527,
+      "step": 6531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.621707926039462e-06,
+      "loss": 0.6864,
+      "step": 6532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6209925582110496e-06,
+      "loss": 0.6385,
+      "step": 6533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.620277180452347e-06,
+      "loss": 0.6558,
+      "step": 6534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.619561792822069e-06,
+      "loss": 0.6453,
+      "step": 6535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6188463953789283e-06,
+      "loss": 0.6563,
+      "step": 6536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6181309881816413e-06,
+      "loss": 0.8377,
+      "step": 6537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.617415571288924e-06,
+      "loss": 0.6376,
+      "step": 6538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.616700144759493e-06,
+      "loss": 0.6441,
+      "step": 6539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6159847086520667e-06,
+      "loss": 0.6978,
+      "step": 6540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6152692630253616e-06,
+      "loss": 0.7144,
+      "step": 6541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6145538079380993e-06,
+      "loss": 0.7026,
+      "step": 6542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.613838343448997e-06,
+      "loss": 0.6519,
+      "step": 6543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6131228696167775e-06,
+      "loss": 0.7339,
+      "step": 6544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6124073865001624e-06,
+      "loss": 0.7384,
+      "step": 6545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6116918941578725e-06,
+      "loss": 0.7905,
+      "step": 6546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6109763926486307e-06,
+      "loss": 0.7139,
+      "step": 6547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6102608820311628e-06,
+      "loss": 0.6881,
+      "step": 6548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6095453623641913e-06,
+      "loss": 0.6292,
+      "step": 6549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.608829833706442e-06,
+      "loss": 0.614,
+      "step": 6550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6081142961166418e-06,
+      "loss": 0.6964,
+      "step": 6551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6073987496535162e-06,
+      "loss": 0.805,
+      "step": 6552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6066831943757936e-06,
+      "loss": 0.7,
+      "step": 6553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6059676303422015e-06,
+      "loss": 0.6064,
+      "step": 6554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6052520576114688e-06,
+      "loss": 0.6843,
+      "step": 6555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6045364762423255e-06,
+      "loss": 0.6555,
+      "step": 6556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.603820886293501e-06,
+      "loss": 0.5978,
+      "step": 6557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.603105287823727e-06,
+      "loss": 0.7258,
+      "step": 6558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6023896808917365e-06,
+      "loss": 0.6768,
+      "step": 6559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6016740655562595e-06,
+      "loss": 0.7802,
+      "step": 6560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6009584418760308e-06,
+      "loss": 0.6767,
+      "step": 6561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6002428099097836e-06,
+      "loss": 0.8508,
+      "step": 6562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5995271697162515e-06,
+      "loss": 0.7807,
+      "step": 6563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.598811521354172e-06,
+      "loss": 0.5856,
+      "step": 6564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.598095864882278e-06,
+      "loss": 0.7518,
+      "step": 6565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.597380200359308e-06,
+      "loss": 0.7399,
+      "step": 6566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5966645278439983e-06,
+      "loss": 0.8267,
+      "step": 6567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5959488473950873e-06,
+      "loss": 0.7494,
+      "step": 6568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5952331590713124e-06,
+      "loss": 0.6065,
+      "step": 6569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5945174629314134e-06,
+      "loss": 0.6825,
+      "step": 6570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.593801759034131e-06,
+      "loss": 0.6329,
+      "step": 6571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5930860474382037e-06,
+      "loss": 0.8188,
+      "step": 6572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.592370328202374e-06,
+      "loss": 0.8625,
+      "step": 6573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5916546013853814e-06,
+      "loss": 0.7324,
+      "step": 6574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5909388670459707e-06,
+      "loss": 0.6179,
+      "step": 6575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5902231252428826e-06,
+      "loss": 0.6088,
+      "step": 6576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5895073760348626e-06,
+      "loss": 0.6178,
+      "step": 6577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.588791619480654e-06,
+      "loss": 0.7179,
+      "step": 6578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.588075855639001e-06,
+      "loss": 0.8449,
+      "step": 6579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5873600845686493e-06,
+      "loss": 0.7855,
+      "step": 6580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.586644306328344e-06,
+      "loss": 0.7403,
+      "step": 6581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5859285209768325e-06,
+      "loss": 0.6272,
+      "step": 6582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.585212728572862e-06,
+      "loss": 0.7969,
+      "step": 6583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5844969291751793e-06,
+      "loss": 0.7652,
+      "step": 6584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.583781122842533e-06,
+      "loss": 0.8514,
+      "step": 6585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5830653096336723e-06,
+      "loss": 0.8774,
+      "step": 6586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.582349489607346e-06,
+      "loss": 0.7139,
+      "step": 6587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5816336628223037e-06,
+      "loss": 0.7862,
+      "step": 6588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5809178293372965e-06,
+      "loss": 0.675,
+      "step": 6589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5802019892110753e-06,
+      "loss": 0.7313,
+      "step": 6590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5794861425023913e-06,
+      "loss": 0.7196,
+      "step": 6591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.578770289269997e-06,
+      "loss": 0.7927,
+      "step": 6592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.578054429572645e-06,
+      "loss": 0.6897,
+      "step": 6593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.577338563469088e-06,
+      "loss": 0.6534,
+      "step": 6594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5766226910180798e-06,
+      "loss": 0.6269,
+      "step": 6595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5759068122783747e-06,
+      "loss": 0.7232,
+      "step": 6596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5751909273087276e-06,
+      "loss": 0.7086,
+      "step": 6597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.574475036167894e-06,
+      "loss": 0.6676,
+      "step": 6598
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.573759138914629e-06,
+      "loss": 0.5715,
+      "step": 6599
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.573043235607689e-06,
+      "loss": 0.7082,
+      "step": 6600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5723273263058307e-06,
+      "loss": 0.6694,
+      "step": 6601
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5716114110678114e-06,
+      "loss": 0.7985,
+      "step": 6602
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5708954899523893e-06,
+      "loss": 0.5793,
+      "step": 6603
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5701795630183214e-06,
+      "loss": 0.7137,
+      "step": 6604
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.569463630324367e-06,
+      "loss": 0.9624,
+      "step": 6605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5687476919292866e-06,
+      "loss": 0.8145,
+      "step": 6606
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.568031747891837e-06,
+      "loss": 0.7292,
+      "step": 6607
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5673157982707802e-06,
+      "loss": 0.8012,
+      "step": 6608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5665998431248773e-06,
+      "loss": 0.829,
+      "step": 6609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5658838825128874e-06,
+      "loss": 0.7059,
+      "step": 6610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5651679164935734e-06,
+      "loss": 0.6926,
+      "step": 6611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.564451945125696e-06,
+      "loss": 0.7526,
+      "step": 6612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5637359684680185e-06,
+      "loss": 0.6894,
+      "step": 6613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5630199865793033e-06,
+      "loss": 0.8012,
+      "step": 6614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5623039995183137e-06,
+      "loss": 0.64,
+      "step": 6615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.561588007343813e-06,
+      "loss": 0.5866,
+      "step": 6616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5608720101145657e-06,
+      "loss": 0.6535,
+      "step": 6617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.560156007889336e-06,
+      "loss": 0.697,
+      "step": 6618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.559440000726889e-06,
+      "loss": 0.724,
+      "step": 6619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5587239886859886e-06,
+      "loss": 0.6226,
+      "step": 6620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5580079718254032e-06,
+      "loss": 0.6855,
+      "step": 6621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5572919502038955e-06,
+      "loss": 0.6997,
+      "step": 6622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.556575923880234e-06,
+      "loss": 0.6927,
+      "step": 6623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5558598929131855e-06,
+      "loss": 0.7261,
+      "step": 6624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.555143857361517e-06,
+      "loss": 0.8136,
+      "step": 6625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.554427817283996e-06,
+      "loss": 0.6442,
+      "step": 6626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5537117727393896e-06,
+      "loss": 0.7321,
+      "step": 6627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5529957237864678e-06,
+      "loss": 0.7104,
+      "step": 6628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.552279670483998e-06,
+      "loss": 0.6163,
+      "step": 6629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.55156361289075e-06,
+      "loss": 0.7306,
+      "step": 6630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.550847551065492e-06,
+      "loss": 0.6597,
+      "step": 6631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5501314850669958e-06,
+      "loss": 0.6505,
+      "step": 6632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.549415414954029e-06,
+      "loss": 0.6729,
+      "step": 6633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.548699340785364e-06,
+      "loss": 0.6379,
+      "step": 6634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.547983262619771e-06,
+      "loss": 0.7451,
+      "step": 6635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.54726718051602e-06,
+      "loss": 0.7899,
+      "step": 6636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5465510945328847e-06,
+      "loss": 0.7983,
+      "step": 6637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5458350047291346e-06,
+      "loss": 0.8199,
+      "step": 6638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5451189111635427e-06,
+      "loss": 0.6877,
+      "step": 6639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.544402813894882e-06,
+      "loss": 0.6661,
+      "step": 6640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5436867129819233e-06,
+      "loss": 0.6876,
+      "step": 6641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.542970608483441e-06,
+      "loss": 0.6777,
+      "step": 6642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.542254500458208e-06,
+      "loss": 0.7154,
+      "step": 6643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.541538388964998e-06,
+      "loss": 0.6623,
+      "step": 6644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5408222740625843e-06,
+      "loss": 0.6229,
+      "step": 6645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.540106155809741e-06,
+      "loss": 0.7534,
+      "step": 6646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.539390034265244e-06,
+      "loss": 0.8405,
+      "step": 6647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.538673909487866e-06,
+      "loss": 0.8002,
+      "step": 6648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5379577815363833e-06,
+      "loss": 0.8036,
+      "step": 6649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.53724165046957e-06,
+      "loss": 0.7654,
+      "step": 6650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.536525516346201e-06,
+      "loss": 0.6834,
+      "step": 6651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.535809379225054e-06,
+      "loss": 0.7736,
+      "step": 6652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.535093239164903e-06,
+      "loss": 0.6872,
+      "step": 6653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.534377096224526e-06,
+      "loss": 0.7447,
+      "step": 6654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.533660950462698e-06,
+      "loss": 0.618,
+      "step": 6655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5329448019381963e-06,
+      "loss": 0.716,
+      "step": 6656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.532228650709797e-06,
+      "loss": 0.6828,
+      "step": 6657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.531512496836278e-06,
+      "loss": 0.7405,
+      "step": 6658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.530796340376416e-06,
+      "loss": 0.7757,
+      "step": 6659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.530080181388989e-06,
+      "loss": 0.7547,
+      "step": 6660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5293640199327746e-06,
+      "loss": 0.7111,
+      "step": 6661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.52864785606655e-06,
+      "loss": 0.5994,
+      "step": 6662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.527931689849095e-06,
+      "loss": 0.6507,
+      "step": 6663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5272155213391863e-06,
+      "loss": 0.7781,
+      "step": 6664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5264993505956025e-06,
+      "loss": 0.6673,
+      "step": 6665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5257831776771235e-06,
+      "loss": 0.6823,
+      "step": 6666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.525067002642527e-06,
+      "loss": 0.6728,
+      "step": 6667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5243508255505927e-06,
+      "loss": 0.5736,
+      "step": 6668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.523634646460099e-06,
+      "loss": 0.7977,
+      "step": 6669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.522918465429826e-06,
+      "loss": 0.702,
+      "step": 6670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5222022825185533e-06,
+      "loss": 0.6746,
+      "step": 6671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5214860977850596e-06,
+      "loss": 0.7406,
+      "step": 6672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.520769911288126e-06,
+      "loss": 0.6995,
+      "step": 6673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.520053723086532e-06,
+      "loss": 0.6931,
+      "step": 6674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5193375332390575e-06,
+      "loss": 0.7644,
+      "step": 6675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5186213418044824e-06,
+      "loss": 0.755,
+      "step": 6676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.517905148841588e-06,
+      "loss": 0.6222,
+      "step": 6677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5171889544091544e-06,
+      "loss": 0.783,
+      "step": 6678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5164727585659616e-06,
+      "loss": 0.9174,
+      "step": 6679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5157565613707908e-06,
+      "loss": 0.7989,
+      "step": 6680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5150403628824234e-06,
+      "loss": 0.7521,
+      "step": 6681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5143241631596406e-06,
+      "loss": 0.7134,
+      "step": 6682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5136079622612214e-06,
+      "loss": 0.7146,
+      "step": 6683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.512891760245949e-06,
+      "loss": 0.8155,
+      "step": 6684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.512175557172604e-06,
+      "loss": 0.6498,
+      "step": 6685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5114593530999675e-06,
+      "loss": 0.6369,
+      "step": 6686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5107431480868215e-06,
+      "loss": 0.6123,
+      "step": 6687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.510026942191947e-06,
+      "loss": 0.7082,
+      "step": 6688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5093107354741253e-06,
+      "loss": 0.7399,
+      "step": 6689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5085945279921387e-06,
+      "loss": 0.5997,
+      "step": 6690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5078783198047685e-06,
+      "loss": 0.6833,
+      "step": 6691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5071621109707966e-06,
+      "loss": 0.7956,
+      "step": 6692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.506445901549005e-06,
+      "loss": 0.6739,
+      "step": 6693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5057296915981753e-06,
+      "loss": 0.6205,
+      "step": 6694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5050134811770894e-06,
+      "loss": 0.6417,
+      "step": 6695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.504297270344529e-06,
+      "loss": 0.7468,
+      "step": 6696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.503581059159276e-06,
+      "loss": 0.7637,
+      "step": 6697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.502864847680113e-06,
+      "loss": 0.6778,
+      "step": 6698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5021486359658226e-06,
+      "loss": 0.702,
+      "step": 6699
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.808079719543457,
+      "eval_runtime": 438.0771,
+      "eval_samples_per_second": 6.846,
+      "eval_steps_per_second": 0.856,
+      "step": 6699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5014324240751845e-06,
+      "loss": 0.7757,
+      "step": 6700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.500716212066984e-06,
+      "loss": 0.6995,
+      "step": 6701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5e-06,
+      "loss": 0.6514,
+      "step": 6702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4992837879330175e-06,
+      "loss": 0.9109,
+      "step": 6703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.498567575924815e-06,
+      "loss": 0.6073,
+      "step": 6704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4978513640341782e-06,
+      "loss": 0.7042,
+      "step": 6705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4971351523198877e-06,
+      "loss": 0.7988,
+      "step": 6706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4964189408407247e-06,
+      "loss": 0.676,
+      "step": 6707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4957027296554715e-06,
+      "loss": 0.8057,
+      "step": 6708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.494986518822912e-06,
+      "loss": 0.8201,
+      "step": 6709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.494270308401826e-06,
+      "loss": 0.9168,
+      "step": 6710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4935540984509954e-06,
+      "loss": 0.5987,
+      "step": 6711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.492837889029204e-06,
+      "loss": 0.7073,
+      "step": 6712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4921216801952323e-06,
+      "loss": 0.7072,
+      "step": 6713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.491405472007863e-06,
+      "loss": 0.6501,
+      "step": 6714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4906892645258755e-06,
+      "loss": 0.5622,
+      "step": 6715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.489973057808054e-06,
+      "loss": 0.6954,
+      "step": 6716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4892568519131798e-06,
+      "loss": 0.6022,
+      "step": 6717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4885406469000325e-06,
+      "loss": 0.6588,
+      "step": 6718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4878244428273967e-06,
+      "loss": 0.7185,
+      "step": 6719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.487108239754052e-06,
+      "loss": 0.6171,
+      "step": 6720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.48639203773878e-06,
+      "loss": 0.6695,
+      "step": 6721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4856758368403606e-06,
+      "loss": 0.7453,
+      "step": 6722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4849596371175774e-06,
+      "loss": 0.6661,
+      "step": 6723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.48424343862921e-06,
+      "loss": 0.8011,
+      "step": 6724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4835272414340392e-06,
+      "loss": 0.6228,
+      "step": 6725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4828110455908465e-06,
+      "loss": 0.6491,
+      "step": 6726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.482094851158413e-06,
+      "loss": 0.68,
+      "step": 6727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4813786581955176e-06,
+      "loss": 0.5963,
+      "step": 6728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4806624667609433e-06,
+      "loss": 0.6489,
+      "step": 6729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.479946276913469e-06,
+      "loss": 0.7119,
+      "step": 6730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.479230088711875e-06,
+      "loss": 0.7385,
+      "step": 6731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.478513902214941e-06,
+      "loss": 0.6396,
+      "step": 6732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.477797717481448e-06,
+      "loss": 0.7335,
+      "step": 6733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4770815345701753e-06,
+      "loss": 0.6847,
+      "step": 6734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4763653535399015e-06,
+      "loss": 0.6458,
+      "step": 6735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.475649174449408e-06,
+      "loss": 0.5862,
+      "step": 6736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4749329973574737e-06,
+      "loss": 0.7114,
+      "step": 6737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4742168223228773e-06,
+      "loss": 0.5963,
+      "step": 6738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4735006494043975e-06,
+      "loss": 0.6805,
+      "step": 6739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4727844786608145e-06,
+      "loss": 0.7106,
+      "step": 6740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.472068310150906e-06,
+      "loss": 0.8178,
+      "step": 6741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4713521439334497e-06,
+      "loss": 0.6234,
+      "step": 6742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4706359800672262e-06,
+      "loss": 0.8109,
+      "step": 6743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4699198186110114e-06,
+      "loss": 0.7408,
+      "step": 6744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.469203659623585e-06,
+      "loss": 0.6896,
+      "step": 6745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4684875031637224e-06,
+      "loss": 0.7189,
+      "step": 6746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4677713492902037e-06,
+      "loss": 0.919,
+      "step": 6747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.467055198061805e-06,
+      "loss": 0.7106,
+      "step": 6748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4663390495373023e-06,
+      "loss": 0.6442,
+      "step": 6749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4656229037754747e-06,
+      "loss": 0.6708,
+      "step": 6750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4649067608350977e-06,
+      "loss": 0.6093,
+      "step": 6751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.464190620774948e-06,
+      "loss": 0.6594,
+      "step": 6752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4634744836538e-06,
+      "loss": 0.7577,
+      "step": 6753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4627583495304315e-06,
+      "loss": 0.6415,
+      "step": 6754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4620422184636183e-06,
+      "loss": 0.7644,
+      "step": 6755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4613260905121344e-06,
+      "loss": 0.683,
+      "step": 6756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.460609965734757e-06,
+      "loss": 0.5452,
+      "step": 6757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4598938441902593e-06,
+      "loss": 0.7511,
+      "step": 6758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4591777259374174e-06,
+      "loss": 0.6599,
+      "step": 6759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.458461611035003e-06,
+      "loss": 0.5449,
+      "step": 6760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4577454995417934e-06,
+      "loss": 0.8245,
+      "step": 6761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.45702939151656e-06,
+      "loss": 0.6644,
+      "step": 6762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4563132870180775e-06,
+      "loss": 0.759,
+      "step": 6763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.455597186105119e-06,
+      "loss": 0.7712,
+      "step": 6764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.454881088836458e-06,
+      "loss": 0.6404,
+      "step": 6765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4541649952708654e-06,
+      "loss": 0.7247,
+      "step": 6766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.453448905467116e-06,
+      "loss": 0.8144,
+      "step": 6767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.45273281948398e-06,
+      "loss": 0.7286,
+      "step": 6768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.45201673738023e-06,
+      "loss": 0.6979,
+      "step": 6769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.451300659214636e-06,
+      "loss": 0.7938,
+      "step": 6770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4505845850459716e-06,
+      "loss": 0.755,
+      "step": 6771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4498685149330055e-06,
+      "loss": 0.7307,
+      "step": 6772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4491524489345082e-06,
+      "loss": 0.6256,
+      "step": 6773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.448436387109251e-06,
+      "loss": 0.7198,
+      "step": 6774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4477203295160026e-06,
+      "loss": 0.7904,
+      "step": 6775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4470042762135335e-06,
+      "loss": 0.6581,
+      "step": 6776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.446288227260611e-06,
+      "loss": 0.6197,
+      "step": 6777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.445572182716005e-06,
+      "loss": 0.5639,
+      "step": 6778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.444856142638484e-06,
+      "loss": 0.8046,
+      "step": 6779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4441401070868145e-06,
+      "loss": 0.6621,
+      "step": 6780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4434240761197665e-06,
+      "loss": 0.9319,
+      "step": 6781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.442708049796105e-06,
+      "loss": 0.6287,
+      "step": 6782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4419920281745984e-06,
+      "loss": 0.7884,
+      "step": 6783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4412760113140114e-06,
+      "loss": 0.6752,
+      "step": 6784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.440559999273112e-06,
+      "loss": 0.8148,
+      "step": 6785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4398439921106647e-06,
+      "loss": 0.7132,
+      "step": 6786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4391279898854343e-06,
+      "loss": 0.7606,
+      "step": 6787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4384119926561872e-06,
+      "loss": 0.5688,
+      "step": 6788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.437696000481687e-06,
+      "loss": 0.6984,
+      "step": 6789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.436980013420697e-06,
+      "loss": 0.6738,
+      "step": 6790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.436264031531982e-06,
+      "loss": 0.909,
+      "step": 6791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4355480548743043e-06,
+      "loss": 0.6653,
+      "step": 6792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4348320835064274e-06,
+      "loss": 0.8262,
+      "step": 6793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4341161174871126e-06,
+      "loss": 0.7774,
+      "step": 6794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4334001568751235e-06,
+      "loss": 0.6485,
+      "step": 6795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4326842017292206e-06,
+      "loss": 0.6689,
+      "step": 6796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.431968252108164e-06,
+      "loss": 0.6631,
+      "step": 6797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4312523080707147e-06,
+      "loss": 0.5801,
+      "step": 6798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4305363696756337e-06,
+      "loss": 0.7778,
+      "step": 6799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4298204369816795e-06,
+      "loss": 0.534,
+      "step": 6800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.429104510047612e-06,
+      "loss": 0.6869,
+      "step": 6801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.428388588932189e-06,
+      "loss": 0.6769,
+      "step": 6802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.42767267369417e-06,
+      "loss": 0.6017,
+      "step": 6803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.426956764392311e-06,
+      "loss": 0.6973,
+      "step": 6804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4262408610853717e-06,
+      "loss": 0.6933,
+      "step": 6805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.425524963832107e-06,
+      "loss": 0.6491,
+      "step": 6806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4248090726912733e-06,
+      "loss": 0.676,
+      "step": 6807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4240931877216257e-06,
+      "loss": 0.7438,
+      "step": 6808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.423377308981921e-06,
+      "loss": 0.6871,
+      "step": 6809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.422661436530913e-06,
+      "loss": 0.6082,
+      "step": 6810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4219455704273556e-06,
+      "loss": 0.6273,
+      "step": 6811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4212297107300033e-06,
+      "loss": 0.6903,
+      "step": 6812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.420513857497609e-06,
+      "loss": 0.6428,
+      "step": 6813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.419798010788926e-06,
+      "loss": 0.7179,
+      "step": 6814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.419082170662704e-06,
+      "loss": 0.7609,
+      "step": 6815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.418366337177697e-06,
+      "loss": 0.8137,
+      "step": 6816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4176505103926552e-06,
+      "loss": 0.6785,
+      "step": 6817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.416934690366328e-06,
+      "loss": 0.5992,
+      "step": 6818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4162188771574677e-06,
+      "loss": 0.6848,
+      "step": 6819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4155030708248215e-06,
+      "loss": 0.7653,
+      "step": 6820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.414787271427139e-06,
+      "loss": 0.6704,
+      "step": 6821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4140714790231675e-06,
+      "loss": 0.6528,
+      "step": 6822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4133556936716566e-06,
+      "loss": 0.761,
+      "step": 6823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4126399154313524e-06,
+      "loss": 0.7835,
+      "step": 6824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4119241443609996e-06,
+      "loss": 0.7861,
+      "step": 6825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4112083805193466e-06,
+      "loss": 0.7941,
+      "step": 6826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.410492623965138e-06,
+      "loss": 0.6917,
+      "step": 6827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.409776874757118e-06,
+      "loss": 0.6217,
+      "step": 6828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.40906113295403e-06,
+      "loss": 0.6941,
+      "step": 6829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.408345398614619e-06,
+      "loss": 0.7792,
+      "step": 6830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.407629671797628e-06,
+      "loss": 0.6716,
+      "step": 6831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4069139525617967e-06,
+      "loss": 0.7493,
+      "step": 6832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.40619824096587e-06,
+      "loss": 0.6722,
+      "step": 6833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4054825370685874e-06,
+      "loss": 0.6377,
+      "step": 6834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.404766840928689e-06,
+      "loss": 0.666,
+      "step": 6835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4040511526049135e-06,
+      "loss": 0.6167,
+      "step": 6836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4033354721560026e-06,
+      "loss": 0.8192,
+      "step": 6837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4026197996406928e-06,
+      "loss": 0.7743,
+      "step": 6838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4019041351177224e-06,
+      "loss": 0.6962,
+      "step": 6839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.401188478645829e-06,
+      "loss": 0.6712,
+      "step": 6840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.400472830283749e-06,
+      "loss": 0.8749,
+      "step": 6841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.399757190090217e-06,
+      "loss": 0.6415,
+      "step": 6842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3990415581239696e-06,
+      "loss": 0.7197,
+      "step": 6843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.398325934443741e-06,
+      "loss": 0.7159,
+      "step": 6844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3976103191082648e-06,
+      "loss": 0.6467,
+      "step": 6845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.396894712176273e-06,
+      "loss": 0.6633,
+      "step": 6846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3961791137064996e-06,
+      "loss": 0.8349,
+      "step": 6847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3954635237576757e-06,
+      "loss": 0.6141,
+      "step": 6848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.394747942388532e-06,
+      "loss": 0.7084,
+      "step": 6849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3940323696577993e-06,
+      "loss": 0.6531,
+      "step": 6850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3933168056242073e-06,
+      "loss": 0.7428,
+      "step": 6851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3926012503464846e-06,
+      "loss": 0.68,
+      "step": 6852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3918857038833586e-06,
+      "loss": 0.7482,
+      "step": 6853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3911701662935584e-06,
+      "loss": 0.7762,
+      "step": 6854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.39045463763581e-06,
+      "loss": 0.7042,
+      "step": 6855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3897391179688376e-06,
+      "loss": 0.6247,
+      "step": 6856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3890236073513697e-06,
+      "loss": 0.7588,
+      "step": 6857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3883081058421283e-06,
+      "loss": 0.7569,
+      "step": 6858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.387592613499839e-06,
+      "loss": 0.6673,
+      "step": 6859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.386877130383222e-06,
+      "loss": 0.7674,
+      "step": 6860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3861616565510033e-06,
+      "loss": 0.7024,
+      "step": 6861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.385446192061902e-06,
+      "loss": 0.7453,
+      "step": 6862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.384730736974638e-06,
+      "loss": 0.6648,
+      "step": 6863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.384015291347934e-06,
+      "loss": 0.7885,
+      "step": 6864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3832998552405075e-06,
+      "loss": 0.702,
+      "step": 6865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3825844287110767e-06,
+      "loss": 0.6864,
+      "step": 6866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.381869011818359e-06,
+      "loss": 0.7465,
+      "step": 6867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3811536046210725e-06,
+      "loss": 0.818,
+      "step": 6868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3804382071779323e-06,
+      "loss": 0.7338,
+      "step": 6869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.379722819547653e-06,
+      "loss": 0.7255,
+      "step": 6870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3790074417889513e-06,
+      "loss": 0.7492,
+      "step": 6871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.378292073960539e-06,
+      "loss": 0.6976,
+      "step": 6872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3775767161211292e-06,
+      "loss": 0.7901,
+      "step": 6873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3768613683294324e-06,
+      "loss": 0.7165,
+      "step": 6874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.376146030644163e-06,
+      "loss": 0.6045,
+      "step": 6875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3754307031240283e-06,
+      "loss": 0.7502,
+      "step": 6876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3747153858277393e-06,
+      "loss": 0.6399,
+      "step": 6877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3740000788140043e-06,
+      "loss": 0.898,
+      "step": 6878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3732847821415314e-06,
+      "loss": 0.6571,
+      "step": 6879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.372569495869026e-06,
+      "loss": 0.7132,
+      "step": 6880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3718542200551974e-06,
+      "loss": 0.692,
+      "step": 6881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.371138954758748e-06,
+      "loss": 0.7734,
+      "step": 6882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3704237000383835e-06,
+      "loss": 0.6544,
+      "step": 6883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.3697084559528056e-06,
+      "loss": 0.6219,
+      "step": 6884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.368993222560719e-06,
+      "loss": 0.6625,
+      "step": 6885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.368277999920825e-06,
+      "loss": 0.6737,
+      "step": 6886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3675627880918236e-06,
+      "loss": 0.7185,
+      "step": 6887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3668475871324157e-06,
+      "loss": 0.6968,
+      "step": 6888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3661323971013003e-06,
+      "loss": 0.613,
+      "step": 6889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3654172180571754e-06,
+      "loss": 0.7497,
+      "step": 6890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.364702050058737e-06,
+      "loss": 0.7228,
+      "step": 6891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.363986893164684e-06,
+      "loss": 0.8945,
+      "step": 6892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3632717474337105e-06,
+      "loss": 0.6689,
+      "step": 6893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.36255661292451e-06,
+      "loss": 0.6745,
+      "step": 6894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3618414896957783e-06,
+      "loss": 0.754,
+      "step": 6895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3611263778062064e-06,
+      "loss": 0.8247,
+      "step": 6896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3604112773144873e-06,
+      "loss": 0.6952,
+      "step": 6897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3596961882793106e-06,
+      "loss": 0.6856,
+      "step": 6898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3589811107593677e-06,
+      "loss": 0.6109,
+      "step": 6899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.358266044813347e-06,
+      "loss": 0.7293,
+      "step": 6900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3575509904999345e-06,
+      "loss": 0.8996,
+      "step": 6901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3568359478778206e-06,
+      "loss": 0.7384,
+      "step": 6902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3561209170056897e-06,
+      "loss": 0.6756,
+      "step": 6903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3554058979422264e-06,
+      "loss": 0.5963,
+      "step": 6904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3546908907461156e-06,
+      "loss": 0.5999,
+      "step": 6905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3539758954760403e-06,
+      "loss": 0.8088,
+      "step": 6906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3532609121906833e-06,
+      "loss": 0.6972,
+      "step": 6907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.352545940948724e-06,
+      "loss": 0.7365,
+      "step": 6908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3518309818088454e-06,
+      "loss": 0.7517,
+      "step": 6909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.351116034829725e-06,
+      "loss": 0.6687,
+      "step": 6910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3504011000700413e-06,
+      "loss": 0.6408,
+      "step": 6911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3496861775884705e-06,
+      "loss": 0.6259,
+      "step": 6912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.348971267443691e-06,
+      "loss": 0.7447,
+      "step": 6913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3482563696943765e-06,
+      "loss": 0.6919,
+      "step": 6914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3475414843992012e-06,
+      "loss": 0.6234,
+      "step": 6915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.346826611616839e-06,
+      "loss": 0.7824,
+      "step": 6916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.346111751405962e-06,
+      "loss": 0.7025,
+      "step": 6917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.34539690382524e-06,
+      "loss": 0.7681,
+      "step": 6918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3446820689333455e-06,
+      "loss": 0.7214,
+      "step": 6919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.343967246788946e-06,
+      "loss": 0.7151,
+      "step": 6920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3432524374507094e-06,
+      "loss": 0.6744,
+      "step": 6921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.342537640977302e-06,
+      "loss": 0.6325,
+      "step": 6922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.341822857427392e-06,
+      "loss": 0.6226,
+      "step": 6923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.341108086859642e-06,
+      "loss": 0.5808,
+      "step": 6924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3403933293327166e-06,
+      "loss": 0.8238,
+      "step": 6925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.339678584905279e-06,
+      "loss": 0.7018,
+      "step": 6926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.33896385363599e-06,
+      "loss": 0.8159,
+      "step": 6927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3382491355835108e-06,
+      "loss": 0.6644,
+      "step": 6928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3375344308064997e-06,
+      "loss": 0.7953,
+      "step": 6929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3368197393636168e-06,
+      "loss": 0.6979,
+      "step": 6930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3361050613135182e-06,
+      "loss": 0.7476,
+      "step": 6931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.335390396714859e-06,
+      "loss": 0.7233,
+      "step": 6932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3346757456262973e-06,
+      "loss": 0.7167,
+      "step": 6933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3339611081064843e-06,
+      "loss": 0.79,
+      "step": 6934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.333246484214075e-06,
+      "loss": 0.7373,
+      "step": 6935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.332531874007718e-06,
+      "loss": 0.685,
+      "step": 6936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3318172775460675e-06,
+      "loss": 0.7824,
+      "step": 6937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3311026948877717e-06,
+      "loss": 0.8331,
+      "step": 6938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.330388126091477e-06,
+      "loss": 0.736,
+      "step": 6939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3296735712158336e-06,
+      "loss": 0.6986,
+      "step": 6940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.328959030319486e-06,
+      "loss": 0.6903,
+      "step": 6941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3282445034610788e-06,
+      "loss": 0.6717,
+      "step": 6942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.327529990699256e-06,
+      "loss": 0.7617,
+      "step": 6943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.326815492092661e-06,
+      "loss": 0.7164,
+      "step": 6944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.326101007699935e-06,
+      "loss": 0.6248,
+      "step": 6945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3253865375797165e-06,
+      "loss": 0.6324,
+      "step": 6946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3246720817906474e-06,
+      "loss": 0.6913,
+      "step": 6947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3239576403913646e-06,
+      "loss": 0.982,
+      "step": 6948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3232432134405046e-06,
+      "loss": 0.6424,
+      "step": 6949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3225288009967016e-06,
+      "loss": 0.8622,
+      "step": 6950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.321814403118592e-06,
+      "loss": 0.825,
+      "step": 6951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3211000198648082e-06,
+      "loss": 0.5805,
+      "step": 6952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3203856512939823e-06,
+      "loss": 0.6722,
+      "step": 6953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.319671297464745e-06,
+      "loss": 0.6462,
+      "step": 6954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.318956958435726e-06,
+      "loss": 0.772,
+      "step": 6955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3182426342655524e-06,
+      "loss": 0.647,
+      "step": 6956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3175283250128537e-06,
+      "loss": 0.6055,
+      "step": 6957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.316814030736255e-06,
+      "loss": 0.6603,
+      "step": 6958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.31609975149438e-06,
+      "loss": 0.9304,
+      "step": 6959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.315385487345852e-06,
+      "loss": 0.6131,
+      "step": 6960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.314671238349294e-06,
+      "loss": 0.6775,
+      "step": 6961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3139570045633274e-06,
+      "loss": 0.698,
+      "step": 6962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3132427860465704e-06,
+      "loss": 0.7777,
+      "step": 6963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.312528582857643e-06,
+      "loss": 0.6171,
+      "step": 6964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.311814395055162e-06,
+      "loss": 0.7916,
+      "step": 6965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3111002226977433e-06,
+      "loss": 0.6545,
+      "step": 6966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3103860658440002e-06,
+      "loss": 0.5828,
+      "step": 6967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3096719245525483e-06,
+      "loss": 0.6557,
+      "step": 6968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3089577988819987e-06,
+      "loss": 0.6791,
+      "step": 6969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3082436888909608e-06,
+      "loss": 0.6216,
+      "step": 6970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.307529594638047e-06,
+      "loss": 0.6227,
+      "step": 6971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3068155161818635e-06,
+      "loss": 0.7428,
+      "step": 6972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3061014535810188e-06,
+      "loss": 0.6486,
+      "step": 6973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.305387406894116e-06,
+      "loss": 0.8053,
+      "step": 6974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3046733761797625e-06,
+      "loss": 0.7061,
+      "step": 6975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3039593614965597e-06,
+      "loss": 0.6487,
+      "step": 6976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3032453629031085e-06,
+      "loss": 0.7346,
+      "step": 6977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3025313804580113e-06,
+      "loss": 0.7006,
+      "step": 6978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3018174142198665e-06,
+      "loss": 0.8234,
+      "step": 6979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.301103464247271e-06,
+      "loss": 0.8216,
+      "step": 6980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.3003895305988218e-06,
+      "loss": 0.7246,
+      "step": 6981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.299675613333114e-06,
+      "loss": 0.5869,
+      "step": 6982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2989617125087415e-06,
+      "loss": 0.6691,
+      "step": 6983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.298247828184296e-06,
+      "loss": 0.6263,
+      "step": 6984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.29753396041837e-06,
+      "loss": 0.6085,
+      "step": 6985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2968201092695515e-06,
+      "loss": 0.675,
+      "step": 6986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.29610627479643e-06,
+      "loss": 0.7064,
+      "step": 6987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2953924570575908e-06,
+      "loss": 0.8895,
+      "step": 6988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2946786561116215e-06,
+      "loss": 0.6501,
+      "step": 6989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.293964872017105e-06,
+      "loss": 0.7095,
+      "step": 6990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2932511048326243e-06,
+      "loss": 0.7729,
+      "step": 6991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.292537354616761e-06,
+      "loss": 0.7089,
+      "step": 6992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.291823621428096e-06,
+      "loss": 0.7656,
+      "step": 6993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2911099053252054e-06,
+      "loss": 0.6518,
+      "step": 6994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2903962063666693e-06,
+      "loss": 0.6963,
+      "step": 6995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2896825246110623e-06,
+      "loss": 0.7319,
+      "step": 6996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2889688601169585e-06,
+      "loss": 0.6495,
+      "step": 6997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.28825521294293e-06,
+      "loss": 0.7112,
+      "step": 6998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2875415831475513e-06,
+      "loss": 0.7185,
+      "step": 6999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2868279707893897e-06,
+      "loss": 0.681,
+      "step": 7000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.286114375927015e-06,
+      "loss": 0.7417,
+      "step": 7001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.285400798618994e-06,
+      "loss": 0.9295,
+      "step": 7002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.284687238923894e-06,
+      "loss": 0.7331,
+      "step": 7003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.283973696900278e-06,
+      "loss": 0.6951,
+      "step": 7004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2832601726067083e-06,
+      "loss": 0.6609,
+      "step": 7005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2825466661017484e-06,
+      "loss": 0.6416,
+      "step": 7006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2818331774439574e-06,
+      "loss": 0.624,
+      "step": 7007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2811197066918923e-06,
+      "loss": 0.5582,
+      "step": 7008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2804062539041127e-06,
+      "loss": 0.6491,
+      "step": 7009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2796928191391724e-06,
+      "loss": 0.6958,
+      "step": 7010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2789794024556276e-06,
+      "loss": 0.6624,
+      "step": 7011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2782660039120278e-06,
+      "loss": 0.6577,
+      "step": 7012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.277552623566927e-06,
+      "loss": 0.7191,
+      "step": 7013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.276839261478874e-06,
+      "loss": 0.6484,
+      "step": 7014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2761259177064153e-06,
+      "loss": 0.7104,
+      "step": 7015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2754125923081004e-06,
+      "loss": 0.7656,
+      "step": 7016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.274699285342473e-06,
+      "loss": 0.6164,
+      "step": 7017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.273985996868076e-06,
+      "loss": 0.682,
+      "step": 7018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2732727269434527e-06,
+      "loss": 0.7831,
+      "step": 7019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.272559475627143e-06,
+      "loss": 0.7413,
+      "step": 7020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2718462429776868e-06,
+      "loss": 0.6693,
+      "step": 7021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2711330290536206e-06,
+      "loss": 0.7143,
+      "step": 7022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2704198339134816e-06,
+      "loss": 0.7247,
+      "step": 7023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2697066576158035e-06,
+      "loss": 0.81,
+      "step": 7024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2689935002191193e-06,
+      "loss": 0.6857,
+      "step": 7025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2682803617819594e-06,
+      "loss": 0.7116,
+      "step": 7026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.267567242362856e-06,
+      "loss": 0.5753,
+      "step": 7027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2668541420203353e-06,
+      "loss": 0.5973,
+      "step": 7028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2661410608129243e-06,
+      "loss": 0.5865,
+      "step": 7029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2654279987991486e-06,
+      "loss": 0.7608,
+      "step": 7030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.264714956037532e-06,
+      "loss": 0.6789,
+      "step": 7031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.264001932586595e-06,
+      "loss": 0.6935,
+      "step": 7032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.26328892850486e-06,
+      "loss": 0.6846,
+      "step": 7033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2625759438508453e-06,
+      "loss": 0.6245,
+      "step": 7034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.261862978683067e-06,
+      "loss": 0.8226,
+      "step": 7035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.26115003306004e-06,
+      "loss": 0.7187,
+      "step": 7036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.260437107040281e-06,
+      "loss": 0.6461,
+      "step": 7037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2597242006823006e-06,
+      "loss": 0.6945,
+      "step": 7038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2590113140446095e-06,
+      "loss": 0.6837,
+      "step": 7039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.258298447185717e-06,
+      "loss": 0.6483,
+      "step": 7040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2575856001641318e-06,
+      "loss": 0.7034,
+      "step": 7041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.256872773038359e-06,
+      "loss": 0.8343,
+      "step": 7042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.256159965866901e-06,
+      "loss": 0.6348,
+      "step": 7043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2554471787082634e-06,
+      "loss": 0.6659,
+      "step": 7044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2547344116209465e-06,
+      "loss": 0.6363,
+      "step": 7045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.254021664663447e-06,
+      "loss": 0.6642,
+      "step": 7046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.253308937894266e-06,
+      "loss": 0.642,
+      "step": 7047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2525962313718977e-06,
+      "loss": 0.7103,
+      "step": 7048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2518835451548375e-06,
+      "loss": 0.7488,
+      "step": 7049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.251170879301576e-06,
+      "loss": 0.8099,
+      "step": 7050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.250458233870607e-06,
+      "loss": 0.6682,
+      "step": 7051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2497456089204187e-06,
+      "loss": 0.7983,
+      "step": 7052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.249033004509497e-06,
+      "loss": 0.8265,
+      "step": 7053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.248320420696331e-06,
+      "loss": 0.6523,
+      "step": 7054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2476078575394036e-06,
+      "loss": 0.7463,
+      "step": 7055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2468953150971963e-06,
+      "loss": 0.5653,
+      "step": 7056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.246182793428191e-06,
+      "loss": 0.799,
+      "step": 7057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.245470292590867e-06,
+      "loss": 0.6444,
+      "step": 7058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2447578126437026e-06,
+      "loss": 0.6859,
+      "step": 7059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2440453536451707e-06,
+      "loss": 0.7367,
+      "step": 7060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2433329156537488e-06,
+      "loss": 0.7327,
+      "step": 7061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2426204987279075e-06,
+      "loss": 0.7473,
+      "step": 7062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2419081029261175e-06,
+      "loss": 0.7394,
+      "step": 7063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2411957283068466e-06,
+      "loss": 0.614,
+      "step": 7064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.240483374928564e-06,
+      "loss": 0.7103,
+      "step": 7065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2397710428497336e-06,
+      "loss": 0.6308,
+      "step": 7066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.23905873212882e-06,
+      "loss": 0.6013,
+      "step": 7067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.238346442824284e-06,
+      "loss": 0.6547,
+      "step": 7068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2376341749945874e-06,
+      "loss": 0.6808,
+      "step": 7069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2369219286981856e-06,
+      "loss": 0.6268,
+      "step": 7070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2362097039935387e-06,
+      "loss": 0.6139,
+      "step": 7071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2354975009391e-06,
+      "loss": 0.7144,
+      "step": 7072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2347853195933224e-06,
+      "loss": 0.7839,
+      "step": 7073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2340731600146558e-06,
+      "loss": 0.8828,
+      "step": 7074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.233361022261552e-06,
+      "loss": 0.6526,
+      "step": 7075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2326489063924572e-06,
+      "loss": 0.7066,
+      "step": 7076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.231936812465818e-06,
+      "loss": 0.7516,
+      "step": 7077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2312247405400782e-06,
+      "loss": 0.6547,
+      "step": 7078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.230512690673681e-06,
+      "loss": 0.8123,
+      "step": 7079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2298006629250657e-06,
+      "loss": 0.7246,
+      "step": 7080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.22908865735267e-06,
+      "loss": 0.5389,
+      "step": 7081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2283766740149336e-06,
+      "loss": 0.6357,
+      "step": 7082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2276647129702896e-06,
+      "loss": 0.6802,
+      "step": 7083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2269527742771704e-06,
+      "loss": 0.6582,
+      "step": 7084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2262408579940095e-06,
+      "loss": 0.6921,
+      "step": 7085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.225528964179235e-06,
+      "loss": 0.7001,
+      "step": 7086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.224817092891275e-06,
+      "loss": 0.7264,
+      "step": 7087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2241052441885546e-06,
+      "loss": 0.7746,
+      "step": 7088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2233934181294993e-06,
+      "loss": 0.7542,
+      "step": 7089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.22268161477253e-06,
+      "loss": 0.6578,
+      "step": 7090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2219698341760665e-06,
+      "loss": 0.6795,
+      "step": 7091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2212580763985288e-06,
+      "loss": 0.7296,
+      "step": 7092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2205463414983324e-06,
+      "loss": 0.685,
+      "step": 7093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.219834629533891e-06,
+      "loss": 0.6138,
+      "step": 7094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2191229405636185e-06,
+      "loss": 0.6805,
+      "step": 7095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2184112746459256e-06,
+      "loss": 0.7157,
+      "step": 7096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.217699631839222e-06,
+      "loss": 0.7567,
+      "step": 7097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2169880122019126e-06,
+      "loss": 0.7127,
+      "step": 7098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.216276415792405e-06,
+      "loss": 0.6263,
+      "step": 7099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2155648426691016e-06,
+      "loss": 0.6521,
+      "step": 7100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2148532928904033e-06,
+      "loss": 0.7958,
+      "step": 7101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2141417665147087e-06,
+      "loss": 0.695,
+      "step": 7102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2134302636004172e-06,
+      "loss": 0.7329,
+      "step": 7103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.212718784205923e-06,
+      "loss": 0.731,
+      "step": 7104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2120073283896204e-06,
+      "loss": 0.7718,
+      "step": 7105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2112958962099005e-06,
+      "loss": 0.6111,
+      "step": 7106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.210584487725155e-06,
+      "loss": 0.6595,
+      "step": 7107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.209873102993768e-06,
+      "loss": 0.637,
+      "step": 7108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2091617420741292e-06,
+      "loss": 0.6667,
+      "step": 7109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2084504050246212e-06,
+      "loss": 0.7733,
+      "step": 7110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2077390919036254e-06,
+      "loss": 0.7322,
+      "step": 7111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.207027802769521e-06,
+      "loss": 0.6521,
+      "step": 7112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2063165376806885e-06,
+      "loss": 0.8169,
+      "step": 7113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2056052966955016e-06,
+      "loss": 0.7689,
+      "step": 7114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2048940798723356e-06,
+      "loss": 0.8202,
+      "step": 7115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.204182887269562e-06,
+      "loss": 0.8078,
+      "step": 7116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2034717189455525e-06,
+      "loss": 0.7852,
+      "step": 7117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.202760574958674e-06,
+      "loss": 0.7282,
+      "step": 7118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2020494553672906e-06,
+      "loss": 0.7252,
+      "step": 7119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2013383602297698e-06,
+      "loss": 0.6071,
+      "step": 7120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2006272896044727e-06,
+      "loss": 0.6281,
+      "step": 7121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.199916243549758e-06,
+      "loss": 0.8595,
+      "step": 7122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1992052221239854e-06,
+      "loss": 0.6201,
+      "step": 7123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.19849422538551e-06,
+      "loss": 0.747,
+      "step": 7124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1977832533926867e-06,
+      "loss": 0.6258,
+      "step": 7125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.197072306203866e-06,
+      "loss": 0.679,
+      "step": 7126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1963613838774e-06,
+      "loss": 0.8004,
+      "step": 7127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1956504864716353e-06,
+      "loss": 0.7085,
+      "step": 7128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.194939614044917e-06,
+      "loss": 0.7123,
+      "step": 7129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.194228766655591e-06,
+      "loss": 0.612,
+      "step": 7130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1935179443619974e-06,
+      "loss": 0.7988,
+      "step": 7131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1928071472224764e-06,
+      "loss": 0.6675,
+      "step": 7132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1920963752953657e-06,
+      "loss": 0.7576,
+      "step": 7133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.191385628639001e-06,
+      "loss": 0.6948,
+      "step": 7134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1906749073117164e-06,
+      "loss": 0.5813,
+      "step": 7135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1899642113718408e-06,
+      "loss": 0.778,
+      "step": 7136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.189253540877707e-06,
+      "loss": 0.6131,
+      "step": 7137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1885428958876407e-06,
+      "loss": 0.6862,
+      "step": 7138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.187832276459967e-06,
+      "loss": 0.7415,
+      "step": 7139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1871216826530076e-06,
+      "loss": 0.7744,
+      "step": 7140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.186411114525086e-06,
+      "loss": 0.6509,
+      "step": 7141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.185700572134519e-06,
+      "loss": 0.7725,
+      "step": 7142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1849900555396246e-06,
+      "loss": 0.7004,
+      "step": 7143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1842795647987173e-06,
+      "loss": 0.6919,
+      "step": 7144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1835690999701093e-06,
+      "loss": 0.6601,
+      "step": 7145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1828586611121105e-06,
+      "loss": 0.7725,
+      "step": 7146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1821482482830305e-06,
+      "loss": 0.6881,
+      "step": 7147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.181437861541175e-06,
+      "loss": 0.6838,
+      "step": 7148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1807275009448466e-06,
+      "loss": 0.6685,
+      "step": 7149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1800171665523477e-06,
+      "loss": 0.7043,
+      "step": 7150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.179306858421979e-06,
+      "loss": 0.718,
+      "step": 7151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.178596576612037e-06,
+      "loss": 0.7732,
+      "step": 7152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.177886321180817e-06,
+      "loss": 0.7484,
+      "step": 7153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.177176092186614e-06,
+      "loss": 0.7162,
+      "step": 7154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1764658896877167e-06,
+      "loss": 0.7145,
+      "step": 7155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1757557137424152e-06,
+      "loss": 0.8776,
+      "step": 7156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.175045564408995e-06,
+      "loss": 0.7001,
+      "step": 7157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1743354417457424e-06,
+      "loss": 0.6232,
+      "step": 7158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1736253458109385e-06,
+      "loss": 0.852,
+      "step": 7159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.172915276662863e-06,
+      "loss": 0.7263,
+      "step": 7160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.172205234359795e-06,
+      "loss": 0.6183,
+      "step": 7161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.171495218960009e-06,
+      "loss": 0.6461,
+      "step": 7162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.17078523052178e-06,
+      "loss": 0.6868,
+      "step": 7163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.170075269103377e-06,
+      "loss": 0.717,
+      "step": 7164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.169365334763072e-06,
+      "loss": 0.7097,
+      "step": 7165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1686554275591297e-06,
+      "loss": 0.6364,
+      "step": 7166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1679455475498148e-06,
+      "loss": 0.8609,
+      "step": 7167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.16723569479339e-06,
+      "loss": 0.7267,
+      "step": 7168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.166525869348117e-06,
+      "loss": 0.7983,
+      "step": 7169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1658160712722513e-06,
+      "loss": 0.6969,
+      "step": 7170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1651063006240488e-06,
+      "loss": 0.6057,
+      "step": 7171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1643965574617648e-06,
+      "loss": 0.6917,
+      "step": 7172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.163686841843649e-06,
+      "loss": 0.6074,
+      "step": 7173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.16297715382795e-06,
+      "loss": 0.7461,
+      "step": 7174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.162267493472916e-06,
+      "loss": 0.6842,
+      "step": 7175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.16155786083679e-06,
+      "loss": 0.7251,
+      "step": 7176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1608482559778148e-06,
+      "loss": 0.6903,
+      "step": 7177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1601386789542286e-06,
+      "loss": 0.7139,
+      "step": 7178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1594291298242716e-06,
+      "loss": 0.6564,
+      "step": 7179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1587196086461766e-06,
+      "loss": 0.6666,
+      "step": 7180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1580101154781767e-06,
+      "loss": 0.6772,
+      "step": 7181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1573006503785044e-06,
+      "loss": 0.8056,
+      "step": 7182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1565912134053875e-06,
+      "loss": 0.759,
+      "step": 7183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.15588180461705e-06,
+      "loss": 0.7399,
+      "step": 7184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.155172424071718e-06,
+      "loss": 0.7506,
+      "step": 7185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1544630718276124e-06,
+      "loss": 0.7002,
+      "step": 7186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1537537479429514e-06,
+      "loss": 0.8398,
+      "step": 7187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1530444524759515e-06,
+      "loss": 0.861,
+      "step": 7188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1523351854848283e-06,
+      "loss": 0.6968,
+      "step": 7189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.151625947027793e-06,
+      "loss": 0.6164,
+      "step": 7190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.150916737163055e-06,
+      "loss": 0.7531,
+      "step": 7191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.150207555948823e-06,
+      "loss": 0.7061,
+      "step": 7192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.149498403443302e-06,
+      "loss": 0.6287,
+      "step": 7193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.148789279704694e-06,
+      "loss": 0.6786,
+      "step": 7194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.148080184791198e-06,
+      "loss": 0.7211,
+      "step": 7195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1473711187610142e-06,
+      "loss": 0.6613,
+      "step": 7196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1466620816723376e-06,
+      "loss": 0.7805,
+      "step": 7197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1459530735833594e-06,
+      "loss": 0.6559,
+      "step": 7198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.145244094552273e-06,
+      "loss": 0.9305,
+      "step": 7199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1445351446372665e-06,
+      "loss": 0.6888,
+      "step": 7200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1438262238965247e-06,
+      "loss": 0.7973,
+      "step": 7201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1431173323882313e-06,
+      "loss": 0.6563,
+      "step": 7202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1424084701705688e-06,
+      "loss": 0.6105,
+      "step": 7203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1416996373017157e-06,
+      "loss": 0.6449,
+      "step": 7204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1409908338398467e-06,
+      "loss": 0.632,
+      "step": 7205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1402820598431386e-06,
+      "loss": 0.6663,
+      "step": 7206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1395733153697613e-06,
+      "loss": 0.6788,
+      "step": 7207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.138864600477884e-06,
+      "loss": 0.7985,
+      "step": 7208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1381559152256733e-06,
+      "loss": 0.7111,
+      "step": 7209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.137447259671295e-06,
+      "loss": 0.6382,
+      "step": 7210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1367386338729095e-06,
+      "loss": 0.7154,
+      "step": 7211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.136030037888676e-06,
+      "loss": 0.7645,
+      "step": 7212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1353214717767527e-06,
+      "loss": 0.6463,
+      "step": 7213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.134612935595294e-06,
+      "loss": 0.7625,
+      "step": 7214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.133904429402452e-06,
+      "loss": 0.6714,
+      "step": 7215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1331959532563746e-06,
+      "loss": 0.6634,
+      "step": 7216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1324875072152108e-06,
+      "loss": 0.6674,
+      "step": 7217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.131779091337105e-06,
+      "loss": 0.6992,
+      "step": 7218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.131070705680198e-06,
+      "loss": 0.6523,
+      "step": 7219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1303623503026326e-06,
+      "loss": 0.7162,
+      "step": 7220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1296540252625435e-06,
+      "loss": 0.741,
+      "step": 7221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.128945730618065e-06,
+      "loss": 0.5805,
+      "step": 7222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.128237466427332e-06,
+      "loss": 0.7638,
+      "step": 7223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1275292327484727e-06,
+      "loss": 0.713,
+      "step": 7224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1268210296396137e-06,
+      "loss": 0.6288,
+      "step": 7225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1261128571588803e-06,
+      "loss": 0.6174,
+      "step": 7226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1254047153643947e-06,
+      "loss": 0.5846,
+      "step": 7227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1246966043142773e-06,
+      "loss": 0.7708,
+      "step": 7228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1239885240666443e-06,
+      "loss": 0.7222,
+      "step": 7229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.123280474679612e-06,
+      "loss": 0.6924,
+      "step": 7230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1225724562112905e-06,
+      "loss": 0.751,
+      "step": 7231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.121864468719791e-06,
+      "loss": 0.6535,
+      "step": 7232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1211565122632184e-06,
+      "loss": 0.6257,
+      "step": 7233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.12044858689968e-06,
+      "loss": 0.6766,
+      "step": 7234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1197406926872766e-06,
+      "loss": 0.6247,
+      "step": 7235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1190328296841057e-06,
+      "loss": 0.7846,
+      "step": 7236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1183249979482666e-06,
+      "loss": 0.6177,
+      "step": 7237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1176171975378535e-06,
+      "loss": 0.7488,
+      "step": 7238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1169094285109573e-06,
+      "loss": 0.6974,
+      "step": 7239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.116201690925666e-06,
+      "loss": 0.7453,
+      "step": 7240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1154939848400683e-06,
+      "loss": 0.6038,
+      "step": 7241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.114786310312248e-06,
+      "loss": 0.7108,
+      "step": 7242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.114078667400284e-06,
+      "loss": 0.9152,
+      "step": 7243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1133710561622576e-06,
+      "loss": 0.7128,
+      "step": 7244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1126634766562436e-06,
+      "loss": 0.6871,
+      "step": 7245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.111955928940317e-06,
+      "loss": 0.6194,
+      "step": 7246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.111248413072547e-06,
+      "loss": 0.7335,
+      "step": 7247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1105409291110033e-06,
+      "loss": 0.6662,
+      "step": 7248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1098334771137513e-06,
+      "loss": 0.7506,
+      "step": 7249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1091260571388527e-06,
+      "loss": 0.9451,
+      "step": 7250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1084186692443706e-06,
+      "loss": 0.694,
+      "step": 7251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.107711313488361e-06,
+      "loss": 0.7549,
+      "step": 7252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.10700398992888e-06,
+      "loss": 0.6616,
+      "step": 7253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1062966986239784e-06,
+      "loss": 0.7439,
+      "step": 7254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.105589439631708e-06,
+      "loss": 0.7635,
+      "step": 7255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.104882213010116e-06,
+      "loss": 0.8389,
+      "step": 7256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1041750188172453e-06,
+      "loss": 0.6518,
+      "step": 7257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.10346785711114e-06,
+      "loss": 0.6328,
+      "step": 7258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1027607279498386e-06,
+      "loss": 0.5963,
+      "step": 7259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1020536313913764e-06,
+      "loss": 0.6766,
+      "step": 7260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1013465674937894e-06,
+      "loss": 0.7679,
+      "step": 7261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1006395363151082e-06,
+      "loss": 0.7964,
+      "step": 7262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.09993253791336e-06,
+      "loss": 0.6962,
+      "step": 7263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.099225572346572e-06,
+      "loss": 0.7235,
+      "step": 7264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0985186396727668e-06,
+      "loss": 0.6573,
+      "step": 7265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0978117399499665e-06,
+      "loss": 0.673,
+      "step": 7266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.097104873236186e-06,
+      "loss": 0.6666,
+      "step": 7267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0963980395894427e-06,
+      "loss": 0.7592,
+      "step": 7268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0956912390677484e-06,
+      "loss": 0.7577,
+      "step": 7269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0949844717291128e-06,
+      "loss": 0.8062,
+      "step": 7270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.094277737631541e-06,
+      "loss": 0.8017,
+      "step": 7271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.09357103683304e-06,
+      "loss": 0.6775,
+      "step": 7272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0928643693916097e-06,
+      "loss": 0.6078,
+      "step": 7273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0921577353652484e-06,
+      "loss": 0.6163,
+      "step": 7274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0914511348119533e-06,
+      "loss": 0.6551,
+      "step": 7275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.090744567789717e-06,
+      "loss": 0.7497,
+      "step": 7276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0900380343565306e-06,
+      "loss": 0.7462,
+      "step": 7277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.08933153457038e-06,
+      "loss": 0.7172,
+      "step": 7278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0886250684892523e-06,
+      "loss": 0.6382,
+      "step": 7279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.087918636171129e-06,
+      "loss": 0.7596,
+      "step": 7280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.087212237673988e-06,
+      "loss": 0.7134,
+      "step": 7281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.086505873055808e-06,
+      "loss": 0.6793,
+      "step": 7282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.085799542374562e-06,
+      "loss": 0.7594,
+      "step": 7283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0850932456882216e-06,
+      "loss": 0.5864,
+      "step": 7284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0843869830547533e-06,
+      "loss": 0.6554,
+      "step": 7285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.083680754532125e-06,
+      "loss": 0.6946,
+      "step": 7286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0829745601782986e-06,
+      "loss": 0.8463,
+      "step": 7287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.082268400051233e-06,
+      "loss": 0.743,
+      "step": 7288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0815622742088862e-06,
+      "loss": 0.7047,
+      "step": 7289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0808561827092126e-06,
+      "loss": 0.6939,
+      "step": 7290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0801501256101625e-06,
+      "loss": 0.7148,
+      "step": 7291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0794441029696855e-06,
+      "loss": 0.6484,
+      "step": 7292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.078738114845727e-06,
+      "loss": 0.7335,
+      "step": 7293
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.078032161296231e-06,
+      "loss": 0.8307,
+      "step": 7294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.077326242379135e-06,
+      "loss": 0.7193,
+      "step": 7295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0766203581523796e-06,
+      "loss": 0.6614,
+      "step": 7296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0759145086738977e-06,
+      "loss": 0.604,
+      "step": 7297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0752086940016196e-06,
+      "loss": 0.7417,
+      "step": 7298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0745029141934763e-06,
+      "loss": 0.6268,
+      "step": 7299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0737971693073927e-06,
+      "loss": 0.7704,
+      "step": 7300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0730914594012913e-06,
+      "loss": 0.6478,
+      "step": 7301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0723857845330924e-06,
+      "loss": 0.6843,
+      "step": 7302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0716801447607135e-06,
+      "loss": 0.7128,
+      "step": 7303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0709745401420696e-06,
+      "loss": 0.6745,
+      "step": 7304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0702689707350702e-06,
+      "loss": 0.7111,
+      "step": 7305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0695634365976266e-06,
+      "loss": 0.6346,
+      "step": 7306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.068857937787643e-06,
+      "loss": 0.6497,
+      "step": 7307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.068152474363022e-06,
+      "loss": 0.7255,
+      "step": 7308
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 0.8053967356681824,
+      "eval_runtime": 437.746,
+      "eval_samples_per_second": 6.851,
+      "eval_steps_per_second": 0.857,
+      "step": 7308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.067447046381663e-06,
+      "loss": 0.7212,
+      "step": 7309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0667416539014652e-06,
+      "loss": 0.6972,
+      "step": 7310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.06603629698032e-06,
+      "loss": 0.7209,
+      "step": 7311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.06533097567612e-06,
+      "loss": 0.6934,
+      "step": 7312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0646256900467536e-06,
+      "loss": 0.6513,
+      "step": 7313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.063920440150106e-06,
+      "loss": 0.5607,
+      "step": 7314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0632152260440597e-06,
+      "loss": 0.7495,
+      "step": 7315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.062510047786492e-06,
+      "loss": 0.7003,
+      "step": 7316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0618049054352824e-06,
+      "loss": 0.6996,
+      "step": 7317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.061099799048303e-06,
+      "loss": 0.6556,
+      "step": 7318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.060394728683424e-06,
+      "loss": 0.6644,
+      "step": 7319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0596896943985146e-06,
+      "loss": 0.6668,
+      "step": 7320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.058984696251438e-06,
+      "loss": 0.7674,
+      "step": 7321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0582797343000566e-06,
+      "loss": 0.6998,
+      "step": 7322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.057574808602228e-06,
+      "loss": 0.6831,
+      "step": 7323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0568699192158096e-06,
+      "loss": 0.6589,
+      "step": 7324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.056165066198654e-06,
+      "loss": 0.6613,
+      "step": 7325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0554602496086086e-06,
+      "loss": 0.7895,
+      "step": 7326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0547554695035233e-06,
+      "loss": 0.7775,
+      "step": 7327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.054050725941241e-06,
+      "loss": 0.6868,
+      "step": 7328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.053346018979602e-06,
+      "loss": 0.6019,
+      "step": 7329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0526413486764436e-06,
+      "loss": 0.6371,
+      "step": 7330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.051936715089601e-06,
+      "loss": 0.7208,
+      "step": 7331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0512321182769075e-06,
+      "loss": 0.8156,
+      "step": 7332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.050527558296189e-06,
+      "loss": 0.7316,
+      "step": 7333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0498230352052743e-06,
+      "loss": 0.701,
+      "step": 7334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.049118549061985e-06,
+      "loss": 0.7752,
+      "step": 7335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0484140999241385e-06,
+      "loss": 0.8,
+      "step": 7336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.047709687849555e-06,
+      "loss": 0.8745,
+      "step": 7337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0470053128960463e-06,
+      "loss": 0.855,
+      "step": 7338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.046300975121423e-06,
+      "loss": 0.5909,
+      "step": 7339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0455966745834926e-06,
+      "loss": 0.6188,
+      "step": 7340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.04489241134006e-06,
+      "loss": 0.5974,
+      "step": 7341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0441881854489263e-06,
+      "loss": 0.5748,
+      "step": 7342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0434839969678893e-06,
+      "loss": 0.7355,
+      "step": 7343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0427798459547453e-06,
+      "loss": 0.942,
+      "step": 7344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0420757324672864e-06,
+      "loss": 0.6736,
+      "step": 7345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.041371656563301e-06,
+      "loss": 0.712,
+      "step": 7346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0406676183005746e-06,
+      "loss": 0.6691,
+      "step": 7347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.039963617736892e-06,
+      "loss": 0.7666,
+      "step": 7348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0392596549300312e-06,
+      "loss": 0.7384,
+      "step": 7349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0385557299377697e-06,
+      "loss": 0.6116,
+      "step": 7350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.037851842817881e-06,
+      "loss": 0.8011,
+      "step": 7351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.037147993628137e-06,
+      "loss": 0.7862,
+      "step": 7352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0364441824263033e-06,
+      "loss": 0.7217,
+      "step": 7353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.035740409270143e-06,
+      "loss": 0.62,
+      "step": 7354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.035036674217421e-06,
+      "loss": 0.6818,
+      "step": 7355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.034332977325893e-06,
+      "loss": 0.6931,
+      "step": 7356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0336293186533133e-06,
+      "loss": 0.6434,
+      "step": 7357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0329256982574356e-06,
+      "loss": 0.7097,
+      "step": 7358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.032222116196007e-06,
+      "loss": 0.6144,
+      "step": 7359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.031518572526774e-06,
+      "loss": 0.6989,
+      "step": 7360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0308150673074774e-06,
+      "loss": 0.7338,
+      "step": 7361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.030111600595859e-06,
+      "loss": 0.6593,
+      "step": 7362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0294081724496525e-06,
+      "loss": 0.6442,
+      "step": 7363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.028704782926591e-06,
+      "loss": 0.6901,
+      "step": 7364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.028001432084405e-06,
+      "loss": 0.6867,
+      "step": 7365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0272981199808215e-06,
+      "loss": 0.7219,
+      "step": 7366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.026594846673562e-06,
+      "loss": 0.7603,
+      "step": 7367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0258916122203476e-06,
+      "loss": 0.7028,
+      "step": 7368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0251884166788952e-06,
+      "loss": 0.8234,
+      "step": 7369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.02448526010692e-06,
+      "loss": 0.7163,
+      "step": 7370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0237821425621294e-06,
+      "loss": 0.7108,
+      "step": 7371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.023079064102234e-06,
+      "loss": 0.6822,
+      "step": 7372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0223760247849363e-06,
+      "loss": 0.6388,
+      "step": 7373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0216730246679368e-06,
+      "loss": 0.8184,
+      "step": 7374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0209700638089344e-06,
+      "loss": 0.7284,
+      "step": 7375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.020267142265623e-06,
+      "loss": 0.686,
+      "step": 7376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0195642600956945e-06,
+      "loss": 0.7286,
+      "step": 7377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0188614173568355e-06,
+      "loss": 0.8242,
+      "step": 7378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.018158614106732e-06,
+      "loss": 0.7795,
+      "step": 7379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0174558504030656e-06,
+      "loss": 0.7072,
+      "step": 7380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0167531263035134e-06,
+      "loss": 0.898,
+      "step": 7381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.016050441865752e-06,
+      "loss": 0.688,
+      "step": 7382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0153477971474532e-06,
+      "loss": 0.8113,
+      "step": 7383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.014645192206285e-06,
+      "loss": 0.5888,
+      "step": 7384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0139426270999105e-06,
+      "loss": 0.7312,
+      "step": 7385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0132401018859958e-06,
+      "loss": 0.5816,
+      "step": 7386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0125376166221972e-06,
+      "loss": 0.7553,
+      "step": 7387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0118351713661704e-06,
+      "loss": 0.6677,
+      "step": 7388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0111327661755677e-06,
+      "loss": 0.6607,
+      "step": 7389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0104304011080385e-06,
+      "loss": 0.7234,
+      "step": 7390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0097280762212283e-06,
+      "loss": 0.7244,
+      "step": 7391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.009025791572778e-06,
+      "loss": 0.8262,
+      "step": 7392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0083235472203284e-06,
+      "loss": 0.654,
+      "step": 7393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0076213432215153e-06,
+      "loss": 0.8016,
+      "step": 7394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0069191796339686e-06,
+      "loss": 0.6303,
+      "step": 7395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0062170565153204e-06,
+      "loss": 0.5983,
+      "step": 7396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.005514973923194e-06,
+      "loss": 0.6778,
+      "step": 7397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.004812931915214e-06,
+      "loss": 0.6039,
+      "step": 7398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0041109305489973e-06,
+      "loss": 0.7104,
+      "step": 7399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0034089698821615e-06,
+      "loss": 0.7649,
+      "step": 7400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0027070499723184e-06,
+      "loss": 0.6969,
+      "step": 7401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.002005170877076e-06,
+      "loss": 0.6312,
+      "step": 7402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.001303332654042e-06,
+      "loss": 0.7026,
+      "step": 7403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0006015353608175e-06,
+      "loss": 0.679,
+      "step": 7404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999899779055001e-06,
+      "loss": 0.7675,
+      "step": 7405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9991980637941892e-06,
+      "loss": 0.774,
+      "step": 7406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.998496389635974e-06,
+      "loss": 0.6987,
+      "step": 7407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9977947566379443e-06,
+      "loss": 0.668,
+      "step": 7408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9970931648576845e-06,
+      "loss": 0.722,
+      "step": 7409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9963916143527794e-06,
+      "loss": 0.6911,
+      "step": 7410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9956901051808066e-06,
+      "loss": 0.6537,
+      "step": 7411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.994988637399339e-06,
+      "loss": 0.7601,
+      "step": 7412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.994287211065952e-06,
+      "loss": 0.6737,
+      "step": 7413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.993585826238213e-06,
+      "loss": 0.7386,
+      "step": 7414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.992884482973686e-06,
+      "loss": 0.6657,
+      "step": 7415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9921831813299338e-06,
+      "loss": 0.7152,
+      "step": 7416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.991481921364515e-06,
+      "loss": 0.7332,
+      "step": 7417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9907807031349836e-06,
+      "loss": 0.6475,
+      "step": 7418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9900795266988914e-06,
+      "loss": 0.7487,
+      "step": 7419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.989378392113787e-06,
+      "loss": 0.788,
+      "step": 7420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.988677299437215e-06,
+      "loss": 0.606,
+      "step": 7421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.987976248726716e-06,
+      "loss": 0.6991,
+      "step": 7422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9872752400398266e-06,
+      "loss": 0.6082,
+      "step": 7423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9865742734340835e-06,
+      "loss": 0.6705,
+      "step": 7424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.985873348967016e-06,
+      "loss": 0.701,
+      "step": 7425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9851724666961517e-06,
+      "loss": 0.8596,
+      "step": 7426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9844716266790146e-06,
+      "loss": 0.6287,
+      "step": 7427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.983770828973126e-06,
+      "loss": 0.6917,
+      "step": 7428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.983070073636002e-06,
+      "loss": 0.7147,
+      "step": 7429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9823693607251547e-06,
+      "loss": 0.7621,
+      "step": 7430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.981668690298097e-06,
+      "loss": 0.7733,
+      "step": 7431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9809680624123337e-06,
+      "loss": 0.6888,
+      "step": 7432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9802674771253672e-06,
+      "loss": 0.7329,
+      "step": 7433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9795669344946988e-06,
+      "loss": 0.6317,
+      "step": 7434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.978866434577823e-06,
+      "loss": 0.7902,
+      "step": 7435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.978165977432234e-06,
+      "loss": 0.6841,
+      "step": 7436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.977465563115418e-06,
+      "loss": 0.6606,
+      "step": 7437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9767651916848633e-06,
+      "loss": 0.6397,
+      "step": 7438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9760648631980513e-06,
+      "loss": 0.6665,
+      "step": 7439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9753645777124584e-06,
+      "loss": 0.6706,
+      "step": 7440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9746643352855626e-06,
+      "loss": 0.6843,
+      "step": 7441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9739641359748334e-06,
+      "loss": 0.6735,
+      "step": 7442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9732639798377386e-06,
+      "loss": 0.7759,
+      "step": 7443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.972563866931743e-06,
+      "loss": 0.7033,
+      "step": 7444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9718637973143072e-06,
+      "loss": 0.7329,
+      "step": 7445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.971163771042889e-06,
+      "loss": 0.6629,
+      "step": 7446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9704637881749406e-06,
+      "loss": 0.8161,
+      "step": 7447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.969763848767914e-06,
+      "loss": 0.7391,
+      "step": 7448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.969063952879255e-06,
+      "loss": 0.6722,
+      "step": 7449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.968364100566405e-06,
+      "loss": 0.6676,
+      "step": 7450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9676642918868054e-06,
+      "loss": 0.7539,
+      "step": 7451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9669645268978916e-06,
+      "loss": 0.7212,
+      "step": 7452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.966264805657095e-06,
+      "loss": 0.7358,
+      "step": 7453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9655651282218447e-06,
+      "loss": 0.8016,
+      "step": 7454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9648654946495654e-06,
+      "loss": 0.6519,
+      "step": 7455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.964165904997679e-06,
+      "loss": 0.8116,
+      "step": 7456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9634663593236026e-06,
+      "loss": 0.7567,
+      "step": 7457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.962766857684752e-06,
+      "loss": 0.6967,
+      "step": 7458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9620674001385366e-06,
+      "loss": 0.7121,
+      "step": 7459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.961367986742363e-06,
+      "loss": 0.7307,
+      "step": 7460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9606686175536344e-06,
+      "loss": 0.8978,
+      "step": 7461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.959969292629752e-06,
+      "loss": 0.7486,
+      "step": 7462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9592700120281106e-06,
+      "loss": 0.7006,
+      "step": 7463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9585707758061028e-06,
+      "loss": 0.5769,
+      "step": 7464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.957871584021118e-06,
+      "loss": 0.6494,
+      "step": 7465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.957172436730541e-06,
+      "loss": 0.7306,
+      "step": 7466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956473333991754e-06,
+      "loss": 0.8022,
+      "step": 7467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557742758621325e-06,
+      "loss": 0.6229,
+      "step": 7468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550752623990537e-06,
+      "loss": 0.6688,
+      "step": 7469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.954376293659887e-06,
+      "loss": 0.6746,
+      "step": 7470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.953677369701998e-06,
+      "loss": 0.8104,
+      "step": 7471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9529784905827513e-06,
+      "loss": 0.6265,
+      "step": 7472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.952279656359506e-06,
+      "loss": 0.6762,
+      "step": 7473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9515808670896184e-06,
+      "loss": 0.7715,
+      "step": 7474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9508821228304387e-06,
+      "loss": 0.6493,
+      "step": 7475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.950183423639318e-06,
+      "loss": 0.776,
+      "step": 7476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9494847695736004e-06,
+      "loss": 0.8753,
+      "step": 7477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9487861606906246e-06,
+      "loss": 0.7066,
+      "step": 7478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9480875970477306e-06,
+      "loss": 0.6817,
+      "step": 7479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9473890787022515e-06,
+      "loss": 0.5792,
+      "step": 7480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9466906057115158e-06,
+      "loss": 0.6481,
+      "step": 7481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9459921781328506e-06,
+      "loss": 0.602,
+      "step": 7482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9452937960235786e-06,
+      "loss": 0.5888,
+      "step": 7483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9445954594410184e-06,
+      "loss": 0.7292,
+      "step": 7484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.943897168442484e-06,
+      "loss": 0.6348,
+      "step": 7485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.943198923085288e-06,
+      "loss": 0.706,
+      "step": 7486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.942500723426738e-06,
+      "loss": 0.6074,
+      "step": 7487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.941802569524137e-06,
+      "loss": 0.7953,
+      "step": 7488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9411044614347836e-06,
+      "loss": 0.7564,
+      "step": 7489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.940406399215976e-06,
+      "loss": 0.7553,
+      "step": 7490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.939708382925006e-06,
+      "loss": 0.6958,
+      "step": 7491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.939010412619163e-06,
+      "loss": 0.7817,
+      "step": 7492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9383124883557302e-06,
+      "loss": 0.5845,
+      "step": 7493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.937614610191991e-06,
+      "loss": 0.6792,
+      "step": 7494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9369167781852204e-06,
+      "loss": 0.6284,
+      "step": 7495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.936218992392694e-06,
+      "loss": 0.7862,
+      "step": 7496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9355212528716803e-06,
+      "loss": 0.6534,
+      "step": 7497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.934823559679446e-06,
+      "loss": 0.6599,
+      "step": 7498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.934125912873252e-06,
+      "loss": 0.6856,
+      "step": 7499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9334283125103583e-06,
+      "loss": 0.7155,
+      "step": 7500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9327307586480176e-06,
+      "loss": 0.6779,
+      "step": 7501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9320332513434824e-06,
+      "loss": 0.78,
+      "step": 7502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.931335790653999e-06,
+      "loss": 0.6728,
+      "step": 7503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9306383766368105e-06,
+      "loss": 0.6113,
+      "step": 7504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9299410093491562e-06,
+      "loss": 0.7609,
+      "step": 7505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9292436888482702e-06,
+      "loss": 0.6865,
+      "step": 7506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9285464151913864e-06,
+      "loss": 0.7066,
+      "step": 7507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.927849188435731e-06,
+      "loss": 0.7324,
+      "step": 7508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9271520086385272e-06,
+      "loss": 0.7123,
+      "step": 7509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9264548758569974e-06,
+      "loss": 0.8176,
+      "step": 7510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.925757790148356e-06,
+      "loss": 0.74,
+      "step": 7511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.925060751569816e-06,
+      "loss": 0.6462,
+      "step": 7512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9243637601785846e-06,
+      "loss": 0.7049,
+      "step": 7513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.923666816031868e-06,
+      "loss": 0.8129,
+      "step": 7514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9229699191868668e-06,
+      "loss": 0.6719,
+      "step": 7515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.922273069700776e-06,
+      "loss": 0.7049,
+      "step": 7516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.921576267630791e-06,
+      "loss": 0.6679,
+      "step": 7517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9208795130340992e-06,
+      "loss": 0.747,
+      "step": 7518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9201828059678855e-06,
+      "loss": 0.8864,
+      "step": 7519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9194861464893317e-06,
+      "loss": 0.764,
+      "step": 7520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9187895346556155e-06,
+      "loss": 0.6288,
+      "step": 7521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9180929705239102e-06,
+      "loss": 0.6164,
+      "step": 7522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.917396454151384e-06,
+      "loss": 0.6554,
+      "step": 7523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9166999855952046e-06,
+      "loss": 0.6927,
+      "step": 7524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9160035649125325e-06,
+      "loss": 0.6882,
+      "step": 7525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.915307192160526e-06,
+      "loss": 0.667,
+      "step": 7526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9146108673963364e-06,
+      "loss": 0.7126,
+      "step": 7527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.913914590677117e-06,
+      "loss": 0.7183,
+      "step": 7528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.913218362060012e-06,
+      "loss": 0.6198,
+      "step": 7529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.912522181602163e-06,
+      "loss": 0.7546,
+      "step": 7530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9118260493607094e-06,
+      "loss": 0.7105,
+      "step": 7531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9111299653927844e-06,
+      "loss": 0.7214,
+      "step": 7532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9104339297555175e-06,
+      "loss": 0.6509,
+      "step": 7533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9097379425060364e-06,
+      "loss": 0.7316,
+      "step": 7534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.909042003701462e-06,
+      "loss": 0.7914,
+      "step": 7535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.908346113398913e-06,
+      "loss": 0.6454,
+      "step": 7536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9076502716555024e-06,
+      "loss": 0.8578,
+      "step": 7537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9069544785283423e-06,
+      "loss": 0.7113,
+      "step": 7538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9062587340745375e-06,
+      "loss": 0.8001,
+      "step": 7539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9055630383511909e-06,
+      "loss": 0.7879,
+      "step": 7540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9048673914154004e-06,
+      "loss": 0.8045,
+      "step": 7541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.904171793324261e-06,
+      "loss": 0.6272,
+      "step": 7542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.903476244134862e-06,
+      "loss": 0.6325,
+      "step": 7543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.902780743904289e-06,
+      "loss": 0.6863,
+      "step": 7544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.902085292689626e-06,
+      "loss": 0.7196,
+      "step": 7545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9013898905479506e-06,
+      "loss": 0.69,
+      "step": 7546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.900694537536335e-06,
+      "loss": 0.827,
+      "step": 7547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8999992337118523e-06,
+      "loss": 0.7314,
+      "step": 7548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8993039791315661e-06,
+      "loss": 0.7371,
+      "step": 7549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8986087738525404e-06,
+      "loss": 0.6951,
+      "step": 7550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.897913617931831e-06,
+      "loss": 0.7617,
+      "step": 7551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.897218511426494e-06,
+      "loss": 0.7224,
+      "step": 7552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8965234543935782e-06,
+      "loss": 0.7094,
+      "step": 7553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8958284468901283e-06,
+      "loss": 0.7641,
+      "step": 7554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8951334889731888e-06,
+      "loss": 0.7866,
+      "step": 7555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8944385806997951e-06,
+      "loss": 0.7072,
+      "step": 7556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8937437221269814e-06,
+      "loss": 0.7598,
+      "step": 7557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.893048913311777e-06,
+      "loss": 0.7597,
+      "step": 7558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8923541543112077e-06,
+      "loss": 0.6328,
+      "step": 7559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8916594451822952e-06,
+      "loss": 0.6708,
+      "step": 7560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8909647859820557e-06,
+      "loss": 0.6769,
+      "step": 7561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8902701767675035e-06,
+      "loss": 0.6695,
+      "step": 7562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8895756175956474e-06,
+      "loss": 0.6316,
+      "step": 7563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8888811085234917e-06,
+      "loss": 0.8282,
+      "step": 7564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.888186649608037e-06,
+      "loss": 0.7723,
+      "step": 7565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8874922409062812e-06,
+      "loss": 0.7054,
+      "step": 7566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8867978824752162e-06,
+      "loss": 0.7575,
+      "step": 7567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8861035743718304e-06,
+      "loss": 0.7687,
+      "step": 7568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8854093166531078e-06,
+      "loss": 0.6638,
+      "step": 7569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8847151093760301e-06,
+      "loss": 0.7169,
+      "step": 7570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8840209525975714e-06,
+      "loss": 0.7955,
+      "step": 7571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8833268463747056e-06,
+      "loss": 0.7841,
+      "step": 7572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8826327907643993e-06,
+      "loss": 0.7579,
+      "step": 7573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.881938785823616e-06,
+      "loss": 0.6076,
+      "step": 7574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8812448316093146e-06,
+      "loss": 0.699,
+      "step": 7575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8805509281784524e-06,
+      "loss": 0.6933,
+      "step": 7576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8798570755879785e-06,
+      "loss": 0.7719,
+      "step": 7577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.879163273894841e-06,
+      "loss": 0.6478,
+      "step": 7578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8784695231559822e-06,
+      "loss": 0.6276,
+      "step": 7579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.877775823428341e-06,
+      "loss": 0.7391,
+      "step": 7580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8770821747688521e-06,
+      "loss": 0.7138,
+      "step": 7581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.876388577234444e-06,
+      "loss": 0.681,
+      "step": 7582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8756950308820451e-06,
+      "loss": 0.7488,
+      "step": 7583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.875001535768576e-06,
+      "loss": 0.7441,
+      "step": 7584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8743080919509535e-06,
+      "loss": 0.7109,
+      "step": 7585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8736146994860927e-06,
+      "loss": 0.8767,
+      "step": 7586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8729213584309016e-06,
+      "loss": 0.7814,
+      "step": 7587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8722280688422862e-06,
+      "loss": 0.6336,
+      "step": 7588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8715348307771452e-06,
+      "loss": 0.7426,
+      "step": 7589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.870841644292378e-06,
+      "loss": 0.6226,
+      "step": 7590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.870148509444875e-06,
+      "loss": 0.6322,
+      "step": 7591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8694554262915237e-06,
+      "loss": 0.8779,
+      "step": 7592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.86876239488921e-06,
+      "loss": 0.7604,
+      "step": 7593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8680694152948122e-06,
+      "loss": 0.6615,
+      "step": 7594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.867376487565205e-06,
+      "loss": 0.6435,
+      "step": 7595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8666836117572606e-06,
+      "loss": 0.7441,
+      "step": 7596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.865990787927845e-06,
+      "loss": 0.766,
+      "step": 7597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8652980161338214e-06,
+      "loss": 0.7248,
+      "step": 7598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8646052964320468e-06,
+      "loss": 0.6528,
+      "step": 7599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.863912628879377e-06,
+      "loss": 0.6319,
+      "step": 7600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8632200135326612e-06,
+      "loss": 0.7673,
+      "step": 7601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8625274504487437e-06,
+      "loss": 0.7359,
+      "step": 7602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8618349396844654e-06,
+      "loss": 0.7082,
+      "step": 7603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.861142481296665e-06,
+      "loss": 0.6756,
+      "step": 7604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8604500753421739e-06,
+      "loss": 0.642,
+      "step": 7605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.85975772187782e-06,
+      "loss": 0.6308,
+      "step": 7606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8590654209604276e-06,
+      "loss": 0.7155,
+      "step": 7607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8583731726468168e-06,
+      "loss": 0.6979,
+      "step": 7608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8576809769938018e-06,
+      "loss": 0.6965,
+      "step": 7609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8569888340581949e-06,
+      "loss": 0.7379,
+      "step": 7610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8562967438968021e-06,
+      "loss": 0.7337,
+      "step": 7611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8556047065664257e-06,
+      "loss": 0.6421,
+      "step": 7612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8549127221238623e-06,
+      "loss": 0.6725,
+      "step": 7613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.854220790625908e-06,
+      "loss": 0.7144,
+      "step": 7614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8535289121293504e-06,
+      "loss": 0.7358,
+      "step": 7615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8528370866909749e-06,
+      "loss": 0.7048,
+      "step": 7616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.852145314367562e-06,
+      "loss": 0.6544,
+      "step": 7617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8514535952158885e-06,
+      "loss": 0.6953,
+      "step": 7618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.850761929292726e-06,
+      "loss": 0.5919,
+      "step": 7619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.85007031665484e-06,
+      "loss": 0.8376,
+      "step": 7620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8493787573589973e-06,
+      "loss": 0.71,
+      "step": 7621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8486872514619542e-06,
+      "loss": 0.7466,
+      "step": 7622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8479957990204644e-06,
+      "loss": 0.8205,
+      "step": 7623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8473044000912802e-06,
+      "loss": 0.7199,
+      "step": 7624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8466130547311452e-06,
+      "loss": 0.8374,
+      "step": 7625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8459217629968023e-06,
+      "loss": 0.7512,
+      "step": 7626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8452305249449859e-06,
+      "loss": 0.7635,
+      "step": 7627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8445393406324309e-06,
+      "loss": 0.7405,
+      "step": 7628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8438482101158644e-06,
+      "loss": 0.6759,
+      "step": 7629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8431571334520081e-06,
+      "loss": 0.6406,
+      "step": 7630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8424661106975842e-06,
+      "loss": 0.6746,
+      "step": 7631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8417751419093056e-06,
+      "loss": 0.8836,
+      "step": 7632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8410842271438822e-06,
+      "loss": 0.8262,
+      "step": 7633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8403933664580208e-06,
+      "loss": 0.7851,
+      "step": 7634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8397025599084225e-06,
+      "loss": 0.6841,
+      "step": 7635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8390118075517846e-06,
+      "loss": 0.7488,
+      "step": 7636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8383211094447985e-06,
+      "loss": 0.7282,
+      "step": 7637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376304656441534e-06,
+      "loss": 0.6357,
+      "step": 7638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369398762065326e-06,
+      "loss": 0.6917,
+      "step": 7639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362493411886156e-06,
+      "loss": 0.7485,
+      "step": 7640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355588606470753e-06,
+      "loss": 0.5651,
+      "step": 7641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8348684346385841e-06,
+      "loss": 0.6883,
+      "step": 7642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341780632198064e-06,
+      "loss": 0.6295,
+      "step": 7643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8334877464474038e-06,
+      "loss": 0.7437,
+      "step": 7644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327974843780327e-06,
+      "loss": 0.7797,
+      "step": 7645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321072770683467e-06,
+      "loss": 0.7055,
+      "step": 7646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8314171245749914e-06,
+      "loss": 0.76,
+      "step": 7647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8307270269546124e-06,
+      "loss": 0.6562,
+      "step": 7648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8300369842638473e-06,
+      "loss": 0.6628,
+      "step": 7649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8293469965593306e-06,
+      "loss": 0.7633,
+      "step": 7650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8286570638976906e-06,
+      "loss": 0.7497,
+      "step": 7651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8279671863355547e-06,
+      "loss": 0.6866,
+      "step": 7652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8272773639295424e-06,
+      "loss": 0.7715,
+      "step": 7653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8265875967362695e-06,
+      "loss": 0.6231,
+      "step": 7654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8258978848123494e-06,
+      "loss": 0.6463,
+      "step": 7655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8252082282143882e-06,
+      "loss": 0.7183,
+      "step": 7656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8245186269989884e-06,
+      "loss": 0.6972,
+      "step": 7657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.823829081222747e-06,
+      "loss": 0.7409,
+      "step": 7658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8231395909422594e-06,
+      "loss": 0.6807,
+      "step": 7659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8224501562141136e-06,
+      "loss": 0.852,
+      "step": 7660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.821760777094893e-06,
+      "loss": 0.6795,
+      "step": 7661
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8210714536411795e-06,
+      "loss": 0.7828,
+      "step": 7662
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8203821859095466e-06,
+      "loss": 0.6427,
+      "step": 7663
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8196929739565661e-06,
+      "loss": 0.6891,
+      "step": 7664
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8190038178388026e-06,
+      "loss": 0.6424,
+      "step": 7665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8183147176128194e-06,
+      "loss": 0.6598,
+      "step": 7666
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8176256733351725e-06,
+      "loss": 0.7269,
+      "step": 7667
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.816936685062413e-06,
+      "loss": 0.7588,
+      "step": 7668
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8162477528510914e-06,
+      "loss": 0.8263,
+      "step": 7669
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8155588767577493e-06,
+      "loss": 0.6525,
+      "step": 7670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8148700568389243e-06,
+      "loss": 0.6213,
+      "step": 7671
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.814181293151151e-06,
+      "loss": 0.6803,
+      "step": 7672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8134925857509595e-06,
+      "loss": 0.7367,
+      "step": 7673
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8128039346948742e-06,
+      "loss": 0.6983,
+      "step": 7674
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8121153400394136e-06,
+      "loss": 0.7869,
+      "step": 7675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8114268018410955e-06,
+      "loss": 0.6341,
+      "step": 7676
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8107383201564294e-06,
+      "loss": 0.7166,
+      "step": 7677
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8100498950419214e-06,
+      "loss": 0.7304,
+      "step": 7678
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8093615265540724e-06,
+      "loss": 0.666,
+      "step": 7679
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8086732147493808e-06,
+      "loss": 0.684,
+      "step": 7680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8079849596843375e-06,
+      "loss": 0.5737,
+      "step": 7681
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.80729676141543e-06,
+      "loss": 0.7068,
+      "step": 7682
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.806608619999143e-06,
+      "loss": 0.608,
+      "step": 7683
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.805920535491953e-06,
+      "loss": 0.8128,
+      "step": 7684
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.805232507950333e-06,
+      "loss": 0.6284,
+      "step": 7685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8045445374307539e-06,
+      "loss": 0.8779,
+      "step": 7686
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8038566239896787e-06,
+      "loss": 0.6829,
+      "step": 7687
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.803168767683567e-06,
+      "loss": 0.6671,
+      "step": 7688
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.802480968568873e-06,
+      "loss": 0.7356,
+      "step": 7689
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8017932267020483e-06,
+      "loss": 0.6139,
+      "step": 7690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8011055421395368e-06,
+      "loss": 0.7608,
+      "step": 7691
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8004179149377793e-06,
+      "loss": 0.7322,
+      "step": 7692
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7997303451532136e-06,
+      "loss": 0.7619,
+      "step": 7693
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7990428328422699e-06,
+      "loss": 0.6028,
+      "step": 7694
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7983553780613745e-06,
+      "loss": 0.7436,
+      "step": 7695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7976679808669484e-06,
+      "loss": 0.6459,
+      "step": 7696
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7969806413154108e-06,
+      "loss": 0.6786,
+      "step": 7697
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.796293359463173e-06,
+      "loss": 0.6435,
+      "step": 7698
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7956061353666422e-06,
+      "loss": 0.7602,
+      "step": 7699
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7949189690822222e-06,
+      "loss": 0.7957,
+      "step": 7700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7942318606663111e-06,
+      "loss": 0.6183,
+      "step": 7701
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7935448101753023e-06,
+      "loss": 0.7324,
+      "step": 7702
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.792857817665583e-06,
+      "loss": 0.7645,
+      "step": 7703
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7921708831935395e-06,
+      "loss": 0.6789,
+      "step": 7704
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7914840068155497e-06,
+      "loss": 0.7267,
+      "step": 7705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7907971885879874e-06,
+      "loss": 0.7002,
+      "step": 7706
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7901104285672238e-06,
+      "loss": 0.764,
+      "step": 7707
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7894237268096232e-06,
+      "loss": 0.7802,
+      "step": 7708
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7887370833715445e-06,
+      "loss": 0.7147,
+      "step": 7709
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7880504983093433e-06,
+      "loss": 0.6576,
+      "step": 7710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7873639716793718e-06,
+      "loss": 0.7391,
+      "step": 7711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7866775035379741e-06,
+      "loss": 0.6686,
+      "step": 7712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7859910939414904e-06,
+      "loss": 0.6573,
+      "step": 7713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7853047429462594e-06,
+      "loss": 0.6639,
+      "step": 7714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7846184506086104e-06,
+      "loss": 0.679,
+      "step": 7715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7839322169848701e-06,
+      "loss": 0.6531,
+      "step": 7716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7832460421313596e-06,
+      "loss": 0.7735,
+      "step": 7717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7825599261043968e-06,
+      "loss": 0.6671,
+      "step": 7718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7818738689602939e-06,
+      "loss": 0.714,
+      "step": 7719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7811878707553563e-06,
+      "loss": 0.8228,
+      "step": 7720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7805019315458882e-06,
+      "loss": 0.642,
+      "step": 7721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7798160513881867e-06,
+      "loss": 0.6943,
+      "step": 7722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7791302303385428e-06,
+      "loss": 0.7431,
+      "step": 7723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7784444684532464e-06,
+      "loss": 0.8459,
+      "step": 7724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.77775876578858e-06,
+      "loss": 0.765,
+      "step": 7725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7770731224008209e-06,
+      "loss": 0.5762,
+      "step": 7726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7763875383462414e-06,
+      "loss": 0.8008,
+      "step": 7727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.775702013681112e-06,
+      "loss": 0.5513,
+      "step": 7728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7750165484616953e-06,
+      "loss": 0.7908,
+      "step": 7729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7743311427442487e-06,
+      "loss": 0.6778,
+      "step": 7730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7736457965850278e-06,
+      "loss": 0.6628,
+      "step": 7731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7729605100402808e-06,
+      "loss": 0.6302,
+      "step": 7732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.772275283166251e-06,
+      "loss": 0.7669,
+      "step": 7733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.771590116019177e-06,
+      "loss": 0.6798,
+      "step": 7734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7709050086552946e-06,
+      "loss": 0.7727,
+      "step": 7735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7702199611308318e-06,
+      "loss": 0.7252,
+      "step": 7736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7695349735020124e-06,
+      "loss": 0.6846,
+      "step": 7737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7688500458250565e-06,
+      "loss": 0.6827,
+      "step": 7738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7681651781561793e-06,
+      "loss": 0.6796,
+      "step": 7739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7674803705515897e-06,
+      "loss": 0.6408,
+      "step": 7740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.766795623067491e-06,
+      "loss": 0.7159,
+      "step": 7741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.766110935760085e-06,
+      "loss": 0.6748,
+      "step": 7742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7654263086855656e-06,
+      "loss": 0.6922,
+      "step": 7743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.764741741900121e-06,
+      "loss": 0.7119,
+      "step": 7744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7640572354599388e-06,
+      "loss": 0.7292,
+      "step": 7745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.763372789421197e-06,
+      "loss": 0.7741,
+      "step": 7746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7626884038400717e-06,
+      "loss": 0.7014,
+      "step": 7747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.762004078772731e-06,
+      "loss": 0.7319,
+      "step": 7748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7613198142753419e-06,
+      "loss": 0.6473,
+      "step": 7749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7606356104040639e-06,
+      "loss": 0.8267,
+      "step": 7750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7599514672150508e-06,
+      "loss": 0.7234,
+      "step": 7751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7592673847644543e-06,
+      "loss": 0.6827,
+      "step": 7752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7585833631084192e-06,
+      "loss": 0.5966,
+      "step": 7753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7578994023030848e-06,
+      "loss": 0.6973,
+      "step": 7754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7572155024045861e-06,
+      "loss": 0.7226,
+      "step": 7755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7565316634690538e-06,
+      "loss": 0.6447,
+      "step": 7756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7558478855526133e-06,
+      "loss": 0.7468,
+      "step": 7757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7551641687113838e-06,
+      "loss": 0.6918,
+      "step": 7758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7544805130014813e-06,
+      "loss": 0.6417,
+      "step": 7759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.753796918479016e-06,
+      "loss": 0.7444,
+      "step": 7760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.753113385200091e-06,
+      "loss": 0.6379,
+      "step": 7761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7524299132208087e-06,
+      "loss": 0.7073,
+      "step": 7762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.751746502597263e-06,
+      "loss": 0.6579,
+      "step": 7763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7510631533855433e-06,
+      "loss": 0.6208,
+      "step": 7764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7503798656417352e-06,
+      "loss": 0.641,
+      "step": 7765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7496966394219183e-06,
+      "loss": 0.7648,
+      "step": 7766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.749013474782168e-06,
+      "loss": 0.7441,
+      "step": 7767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7483303717785527e-06,
+      "loss": 0.7654,
+      "step": 7768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7476473304671388e-06,
+      "loss": 0.5949,
+      "step": 7769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7469643509039851e-06,
+      "loss": 0.6477,
+      "step": 7770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7462814331451467e-06,
+      "loss": 0.6409,
+      "step": 7771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7455985772466711e-06,
+      "loss": 0.7977,
+      "step": 7772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7449157832646052e-06,
+      "loss": 0.7024,
+      "step": 7773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7442330512549869e-06,
+      "loss": 0.618,
+      "step": 7774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7435503812738509e-06,
+      "loss": 0.6535,
+      "step": 7775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7428677733772259e-06,
+      "loss": 0.7326,
+      "step": 7776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7421852276211374e-06,
+      "loss": 0.6107,
+      "step": 7777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.741502744061603e-06,
+      "loss": 0.796,
+      "step": 7778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.740820322754636e-06,
+      "loss": 0.6726,
+      "step": 7779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7401379637562471e-06,
+      "loss": 0.7546,
+      "step": 7780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7394556671224389e-06,
+      "loss": 0.6693,
+      "step": 7781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7387734329092085e-06,
+      "loss": 0.7878,
+      "step": 7782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.738091261172552e-06,
+      "loss": 0.6797,
+      "step": 7783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7374091519684555e-06,
+      "loss": 0.7064,
+      "step": 7784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.736727105352904e-06,
+      "loss": 0.7161,
+      "step": 7785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7360451213818726e-06,
+      "loss": 0.672,
+      "step": 7786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7353632001113375e-06,
+      "loss": 0.6292,
+      "step": 7787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.734681341597265e-06,
+      "loss": 0.6432,
+      "step": 7788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7339995458956164e-06,
+      "loss": 0.6287,
+      "step": 7789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7333178130623513e-06,
+      "loss": 0.7088,
+      "step": 7790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7326361431534206e-06,
+      "loss": 0.6914,
+      "step": 7791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7319545362247713e-06,
+      "loss": 0.6778,
+      "step": 7792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7312729923323458e-06,
+      "loss": 0.6261,
+      "step": 7793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7305915115320803e-06,
+      "loss": 0.6353,
+      "step": 7794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.729910093879908e-06,
+      "loss": 0.7708,
+      "step": 7795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7292287394317525e-06,
+      "loss": 0.6608,
+      "step": 7796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7285474482435374e-06,
+      "loss": 0.6501,
+      "step": 7797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7278662203711782e-06,
+      "loss": 0.7402,
+      "step": 7798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.727185055870584e-06,
+      "loss": 0.54,
+      "step": 7799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7265039547976626e-06,
+      "loss": 0.6637,
+      "step": 7800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7258229172083134e-06,
+      "loss": 0.7344,
+      "step": 7801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7251419431584316e-06,
+      "loss": 0.6708,
+      "step": 7802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7244610327039069e-06,
+      "loss": 0.6334,
+      "step": 7803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7237801859006244e-06,
+      "loss": 0.6697,
+      "step": 7804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7230994028044643e-06,
+      "loss": 0.7678,
+      "step": 7805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7224186834712988e-06,
+      "loss": 0.7118,
+      "step": 7806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7217380279569996e-06,
+      "loss": 0.7503,
+      "step": 7807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7210574363174293e-06,
+      "loss": 0.8555,
+      "step": 7808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.720376908608446e-06,
+      "loss": 0.641,
+      "step": 7809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7196964448859027e-06,
+      "loss": 0.741,
+      "step": 7810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7190160452056493e-06,
+      "loss": 0.713,
+      "step": 7811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.718335709623527e-06,
+      "loss": 0.736,
+      "step": 7812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7176554381953739e-06,
+      "loss": 0.6987,
+      "step": 7813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7169752309770222e-06,
+      "loss": 0.7759,
+      "step": 7814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7162950880242998e-06,
+      "loss": 0.6712,
+      "step": 7815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7156150093930274e-06,
+      "loss": 0.6268,
+      "step": 7816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7149349951390208e-06,
+      "loss": 0.7017,
+      "step": 7817
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7142550453180934e-06,
+      "loss": 0.688,
+      "step": 7818
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7135751599860499e-06,
+      "loss": 0.6748,
+      "step": 7819
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7128953391986894e-06,
+      "loss": 0.8697,
+      "step": 7820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7122155830118097e-06,
+      "loss": 0.7805,
+      "step": 7821
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7115358914811997e-06,
+      "loss": 0.6787,
+      "step": 7822
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7108562646626444e-06,
+      "loss": 0.7445,
+      "step": 7823
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.710176702611922e-06,
+      "loss": 0.7449,
+      "step": 7824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7094972053848086e-06,
+      "loss": 0.756,
+      "step": 7825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.708817773037072e-06,
+      "loss": 0.6861,
+      "step": 7826
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7081384056244743e-06,
+      "loss": 0.7257,
+      "step": 7827
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.707459103202776e-06,
+      "loss": 0.6966,
+      "step": 7828
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7067798658277287e-06,
+      "loss": 0.6523,
+      "step": 7829
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7061006935550793e-06,
+      "loss": 0.592,
+      "step": 7830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7054215864405704e-06,
+      "loss": 0.7194,
+      "step": 7831
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7047425445399385e-06,
+      "loss": 0.7066,
+      "step": 7832
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7040635679089163e-06,
+      "loss": 0.658,
+      "step": 7833
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7033846566032271e-06,
+      "loss": 0.7061,
+      "step": 7834
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7027058106785946e-06,
+      "loss": 0.618,
+      "step": 7835
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7020270301907327e-06,
+      "loss": 0.6083,
+      "step": 7836
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7013483151953503e-06,
+      "loss": 0.6861,
+      "step": 7837
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7006696657481542e-06,
+      "loss": 0.7214,
+      "step": 7838
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.699991081904842e-06,
+      "loss": 0.7325,
+      "step": 7839
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6993125637211077e-06,
+      "loss": 0.6716,
+      "step": 7840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6986341112526395e-06,
+      "loss": 0.5918,
+      "step": 7841
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6979557245551208e-06,
+      "loss": 0.805,
+      "step": 7842
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6972774036842304e-06,
+      "loss": 0.6474,
+      "step": 7843
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6965991486956377e-06,
+      "loss": 0.6865,
+      "step": 7844
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.695920959645012e-06,
+      "loss": 0.6286,
+      "step": 7845
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6952428365880141e-06,
+      "loss": 0.6784,
+      "step": 7846
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6945647795802996e-06,
+      "loss": 0.723,
+      "step": 7847
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.693886788677518e-06,
+      "loss": 0.8451,
+      "step": 7848
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6932088639353167e-06,
+      "loss": 0.7357,
+      "step": 7849
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6925310054093336e-06,
+      "loss": 0.8686,
+      "step": 7850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6918532131552038e-06,
+      "loss": 0.6841,
+      "step": 7851
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6911754872285557e-06,
+      "loss": 0.6636,
+      "step": 7852
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6904978276850133e-06,
+      "loss": 0.6795,
+      "step": 7853
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6898202345801945e-06,
+      "loss": 0.6013,
+      "step": 7854
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.68914270796971e-06,
+      "loss": 0.6399,
+      "step": 7855
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6884652479091696e-06,
+      "loss": 0.6361,
+      "step": 7856
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.687787854454173e-06,
+      "loss": 0.6555,
+      "step": 7857
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.687110527660316e-06,
+      "loss": 0.7314,
+      "step": 7858
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6864332675831912e-06,
+      "loss": 0.7358,
+      "step": 7859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6857560742783817e-06,
+      "loss": 0.7016,
+      "step": 7860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6850789478014687e-06,
+      "loss": 0.7628,
+      "step": 7861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6844018882080243e-06,
+      "loss": 0.8245,
+      "step": 7862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6837248955536202e-06,
+      "loss": 0.7427,
+      "step": 7863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6830479698938174e-06,
+      "loss": 0.665,
+      "step": 7864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6823711112841733e-06,
+      "loss": 0.6344,
+      "step": 7865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.681694319780242e-06,
+      "loss": 0.7421,
+      "step": 7866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6810175954375692e-06,
+      "loss": 0.9011,
+      "step": 7867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6803409383116955e-06,
+      "loss": 0.6808,
+      "step": 7868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.679664348458157e-06,
+      "loss": 0.6128,
+      "step": 7869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.678987825932484e-06,
+      "loss": 0.747,
+      "step": 7870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6783113707902015e-06,
+      "loss": 0.6878,
+      "step": 7871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6776349830868272e-06,
+      "loss": 0.7197,
+      "step": 7872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6769586628778761e-06,
+      "loss": 0.8789,
+      "step": 7873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.676282410218856e-06,
+      "loss": 0.8806,
+      "step": 7874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6756062251652677e-06,
+      "loss": 0.7549,
+      "step": 7875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.674930107772611e-06,
+      "loss": 0.7059,
+      "step": 7876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.674254058096375e-06,
+      "loss": 0.7368,
+      "step": 7877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6735780761920461e-06,
+      "loss": 0.704,
+      "step": 7878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6729021621151045e-06,
+      "loss": 0.7182,
+      "step": 7879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.672226315921025e-06,
+      "loss": 0.5803,
+      "step": 7880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6715505376652775e-06,
+      "loss": 0.6693,
+      "step": 7881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6708748274033232e-06,
+      "loss": 0.6246,
+      "step": 7882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6701991851906229e-06,
+      "loss": 0.9156,
+      "step": 7883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6695236110826275e-06,
+      "loss": 0.6825,
+      "step": 7884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6688481051347838e-06,
+      "loss": 0.6099,
+      "step": 7885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6681726674025323e-06,
+      "loss": 0.7501,
+      "step": 7886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.66749729794131e-06,
+      "loss": 0.8491,
+      "step": 7887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6668219968065463e-06,
+      "loss": 0.5975,
+      "step": 7888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6661467640536649e-06,
+      "loss": 0.6638,
+      "step": 7889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6654715997380854e-06,
+      "loss": 0.6183,
+      "step": 7890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6647965039152215e-06,
+      "loss": 0.7398,
+      "step": 7891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6641214766404796e-06,
+      "loss": 0.7007,
+      "step": 7892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6634465179692608e-06,
+      "loss": 0.7307,
+      "step": 7893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6627716279569633e-06,
+      "loss": 0.6734,
+      "step": 7894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.662096806658977e-06,
+      "loss": 0.7835,
+      "step": 7895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.661422054130686e-06,
+      "loss": 0.6801,
+      "step": 7896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6607473704274712e-06,
+      "loss": 0.7181,
+      "step": 7897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6600727556047052e-06,
+      "loss": 0.7013,
+      "step": 7898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.659398209717757e-06,
+      "loss": 0.648,
+      "step": 7899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6587237328219868e-06,
+      "loss": 0.6683,
+      "step": 7900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6580493249727538e-06,
+      "loss": 0.675,
+      "step": 7901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6573749862254085e-06,
+      "loss": 0.66,
+      "step": 7902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6567007166352944e-06,
+      "loss": 0.7212,
+      "step": 7903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.656026516257754e-06,
+      "loss": 0.7205,
+      "step": 7904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.65535238514812e-06,
+      "loss": 0.6738,
+      "step": 7905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6546783233617204e-06,
+      "loss": 0.6934,
+      "step": 7906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6540043309538779e-06,
+      "loss": 0.6901,
+      "step": 7907
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6533304079799095e-06,
+      "loss": 0.7467,
+      "step": 7908
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6526565544951272e-06,
+      "loss": 0.8238,
+      "step": 7909
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6519827705548352e-06,
+      "loss": 0.6616,
+      "step": 7910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6513090562143355e-06,
+      "loss": 0.5131,
+      "step": 7911
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6506354115289208e-06,
+      "loss": 0.7371,
+      "step": 7912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6499618365538784e-06,
+      "loss": 0.7248,
+      "step": 7913
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6492883313444933e-06,
+      "loss": 0.6206,
+      "step": 7914
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6486148959560413e-06,
+      "loss": 0.7065,
+      "step": 7915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6479415304437934e-06,
+      "loss": 0.5885,
+      "step": 7916
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6472682348630152e-06,
+      "loss": 0.7558,
+      "step": 7917
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 0.8035831451416016,
+      "eval_runtime": 437.9664,
+      "eval_samples_per_second": 6.848,
+      "eval_steps_per_second": 0.856,
+      "step": 7917
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6465950092689669e-06,
+      "loss": 0.6574,
+      "step": 7918
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.645921853716903e-06,
+      "loss": 0.6988,
+      "step": 7919
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.64524876826207e-06,
+      "loss": 0.6565,
+      "step": 7920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6445757529597123e-06,
+      "loss": 0.8107,
+      "step": 7921
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.643902807865066e-06,
+      "loss": 0.7993,
+      "step": 7922
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6432299330333617e-06,
+      "loss": 0.6759,
+      "step": 7923
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6425571285198244e-06,
+      "loss": 0.6534,
+      "step": 7924
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6418843943796747e-06,
+      "loss": 0.8197,
+      "step": 7925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6412117306681252e-06,
+      "loss": 0.6297,
+      "step": 7926
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6405391374403845e-06,
+      "loss": 0.8093,
+      "step": 7927
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.639866614751654e-06,
+      "loss": 0.6132,
+      "step": 7928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6391941626571312e-06,
+      "loss": 0.664,
+      "step": 7929
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6385217812120064e-06,
+      "loss": 0.802,
+      "step": 7930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.637849470471462e-06,
+      "loss": 0.6837,
+      "step": 7931
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6371772304906802e-06,
+      "loss": 0.6812,
+      "step": 7932
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6365050613248329e-06,
+      "loss": 0.6779,
+      "step": 7933
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6358329630290862e-06,
+      "loss": 0.6304,
+      "step": 7934
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6351609356586034e-06,
+      "loss": 0.7967,
+      "step": 7935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6344889792685392e-06,
+      "loss": 0.6295,
+      "step": 7936
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.633817093914044e-06,
+      "loss": 0.7011,
+      "step": 7937
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6331452796502605e-06,
+      "loss": 0.6972,
+      "step": 7938
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.632473536532329e-06,
+      "loss": 0.7293,
+      "step": 7939
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6318018646153808e-06,
+      "loss": 0.6606,
+      "step": 7940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6311302639545413e-06,
+      "loss": 0.6996,
+      "step": 7941
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6304587346049328e-06,
+      "loss": 0.5799,
+      "step": 7942
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.62978727662167e-06,
+      "loss": 0.7663,
+      "step": 7943
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6291158900598605e-06,
+      "loss": 0.8055,
+      "step": 7944
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6284445749746082e-06,
+      "loss": 0.6221,
+      "step": 7945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6277733314210104e-06,
+      "loss": 0.7451,
+      "step": 7946
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6271021594541591e-06,
+      "loss": 0.6851,
+      "step": 7947
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6264310591291377e-06,
+      "loss": 0.6458,
+      "step": 7948
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6257600305010285e-06,
+      "loss": 0.7041,
+      "step": 7949
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6250890736249036e-06,
+      "loss": 0.7456,
+      "step": 7950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6244181885558302e-06,
+      "loss": 0.8543,
+      "step": 7951
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.623747375348872e-06,
+      "loss": 0.8169,
+      "step": 7952
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6230766340590843e-06,
+      "loss": 0.8438,
+      "step": 7953
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6224059647415163e-06,
+      "loss": 0.7179,
+      "step": 7954
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6217353674512131e-06,
+      "loss": 0.703,
+      "step": 7955
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.621064842243213e-06,
+      "loss": 0.8124,
+      "step": 7956
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6203943891725485e-06,
+      "loss": 0.643,
+      "step": 7957
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6197240082942448e-06,
+      "loss": 0.8247,
+      "step": 7958
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6190536996633246e-06,
+      "loss": 0.6289,
+      "step": 7959
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6183834633348013e-06,
+      "loss": 0.6318,
+      "step": 7960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6177132993636838e-06,
+      "loss": 0.6947,
+      "step": 7961
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6170432078049736e-06,
+      "loss": 0.7094,
+      "step": 7962
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6163731887136698e-06,
+      "loss": 0.7496,
+      "step": 7963
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6157032421447613e-06,
+      "loss": 0.6434,
+      "step": 7964
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6150333681532337e-06,
+      "loss": 0.7602,
+      "step": 7965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.614363566794066e-06,
+      "loss": 0.7035,
+      "step": 7966
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.613693838122232e-06,
+      "loss": 0.6678,
+      "step": 7967
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6130241821926978e-06,
+      "loss": 0.7124,
+      "step": 7968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6123545990604233e-06,
+      "loss": 0.7099,
+      "step": 7969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6116850887803658e-06,
+      "loss": 0.7596,
+      "step": 7970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6110156514074737e-06,
+      "loss": 0.6985,
+      "step": 7971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6103462869966885e-06,
+      "loss": 0.8293,
+      "step": 7972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.60967699560295e-06,
+      "loss": 0.6412,
+      "step": 7973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.609007777281187e-06,
+      "loss": 0.7144,
+      "step": 7974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.608338632086326e-06,
+      "loss": 0.7696,
+      "step": 7975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.607669560073285e-06,
+      "loss": 0.6107,
+      "step": 7976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6070005612969791e-06,
+      "loss": 0.7783,
+      "step": 7977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6063316358123138e-06,
+      "loss": 0.7127,
+      "step": 7978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6056627836741894e-06,
+      "loss": 0.7216,
+      "step": 7979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6049940049375032e-06,
+      "loss": 0.8198,
+      "step": 7980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6043252996571432e-06,
+      "loss": 0.6759,
+      "step": 7981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6036566678879917e-06,
+      "loss": 0.6085,
+      "step": 7982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6029881096849264e-06,
+      "loss": 0.5557,
+      "step": 7983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6023196251028184e-06,
+      "loss": 0.7602,
+      "step": 7984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.601651214196533e-06,
+      "loss": 0.8278,
+      "step": 7985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.600982877020927e-06,
+      "loss": 0.6708,
+      "step": 7986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.600314613630856e-06,
+      "loss": 0.6749,
+      "step": 7987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5996464240811654e-06,
+      "loss": 0.6652,
+      "step": 7988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5989783084266951e-06,
+      "loss": 0.7238,
+      "step": 7989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5983102667222813e-06,
+      "loss": 0.6615,
+      "step": 7990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.597642299022752e-06,
+      "loss": 0.7282,
+      "step": 7991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.596974405382929e-06,
+      "loss": 0.8494,
+      "step": 7992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.596306585857629e-06,
+      "loss": 0.5333,
+      "step": 7993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5956388405016626e-06,
+      "loss": 0.7818,
+      "step": 7994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5949711693698348e-06,
+      "loss": 0.6468,
+      "step": 7995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5943035725169414e-06,
+      "loss": 0.8775,
+      "step": 7996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5936360499977776e-06,
+      "loss": 0.7377,
+      "step": 7997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5929686018671273e-06,
+      "loss": 0.728,
+      "step": 7998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5923012281797707e-06,
+      "loss": 0.7437,
+      "step": 7999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.591633928990481e-06,
+      "loss": 0.6988,
+      "step": 8000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.590966704354027e-06,
+      "loss": 0.614,
+      "step": 8001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5902995543251693e-06,
+      "loss": 0.7704,
+      "step": 8002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5896324789586635e-06,
+      "loss": 0.6683,
+      "step": 8003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.588965478309259e-06,
+      "loss": 0.7275,
+      "step": 8004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5882985524316994e-06,
+      "loss": 0.6878,
+      "step": 8005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5876317013807207e-06,
+      "loss": 0.6547,
+      "step": 8006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5869649252110537e-06,
+      "loss": 0.7209,
+      "step": 8007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5862982239774243e-06,
+      "loss": 0.6993,
+      "step": 8008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5856315977345504e-06,
+      "loss": 0.7752,
+      "step": 8009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5849650465371435e-06,
+      "loss": 0.6591,
+      "step": 8010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5842985704399113e-06,
+      "loss": 0.6228,
+      "step": 8011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.583632169497553e-06,
+      "loss": 0.6296,
+      "step": 8012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5829658437647628e-06,
+      "loss": 0.7326,
+      "step": 8013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5822995932962275e-06,
+      "loss": 0.6611,
+      "step": 8014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5816334181466302e-06,
+      "loss": 0.7318,
+      "step": 8015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.580967318370646e-06,
+      "loss": 0.7638,
+      "step": 8016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5803012940229423e-06,
+      "loss": 0.7141,
+      "step": 8017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5796353451581845e-06,
+      "loss": 0.651,
+      "step": 8018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5789694718310284e-06,
+      "loss": 0.6643,
+      "step": 8019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5783036740961236e-06,
+      "loss": 0.6578,
+      "step": 8020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5776379520081154e-06,
+      "loss": 0.7307,
+      "step": 8021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.576972305621642e-06,
+      "loss": 0.6445,
+      "step": 8022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5763067349913357e-06,
+      "loss": 0.7664,
+      "step": 8023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5756412401718214e-06,
+      "loss": 0.6929,
+      "step": 8024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5749758212177197e-06,
+      "loss": 0.675,
+      "step": 8025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5743104781836433e-06,
+      "loss": 0.6763,
+      "step": 8026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.573645211124198e-06,
+      "loss": 0.65,
+      "step": 8027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5729800200939872e-06,
+      "loss": 0.6902,
+      "step": 8028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5723149051476039e-06,
+      "loss": 0.7253,
+      "step": 8029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5716498663396365e-06,
+      "loss": 0.744,
+      "step": 8030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5709849037246672e-06,
+      "loss": 0.6175,
+      "step": 8031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5703200173572718e-06,
+      "loss": 0.7459,
+      "step": 8032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.569655207292021e-06,
+      "loss": 0.8251,
+      "step": 8033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.568990473583476e-06,
+      "loss": 0.7595,
+      "step": 8034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5683258162861958e-06,
+      "loss": 0.6996,
+      "step": 8035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.567661235454731e-06,
+      "loss": 0.8267,
+      "step": 8036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5669967311436251e-06,
+      "loss": 0.6959,
+      "step": 8037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.566332303407416e-06,
+      "loss": 0.7132,
+      "step": 8038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5656679523006373e-06,
+      "loss": 0.7089,
+      "step": 8039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5650036778778132e-06,
+      "loss": 0.8202,
+      "step": 8040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.564339480193464e-06,
+      "loss": 0.6078,
+      "step": 8041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5636753593021024e-06,
+      "loss": 0.7072,
+      "step": 8042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5630113152582355e-06,
+      "loss": 0.7296,
+      "step": 8043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5623473481163633e-06,
+      "loss": 0.6013,
+      "step": 8044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5616834579309792e-06,
+      "loss": 0.7018,
+      "step": 8045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.561019644756573e-06,
+      "loss": 0.7956,
+      "step": 8046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5603559086476256e-06,
+      "loss": 0.8099,
+      "step": 8047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.55969224965861e-06,
+      "loss": 0.7802,
+      "step": 8048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5590286678439982e-06,
+      "loss": 0.6891,
+      "step": 8049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5583651632582505e-06,
+      "loss": 0.6706,
+      "step": 8050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5577017359558243e-06,
+      "loss": 0.6976,
+      "step": 8051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5570383859911684e-06,
+      "loss": 0.7551,
+      "step": 8052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5563751134187276e-06,
+      "loss": 0.7168,
+      "step": 8053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5557119182929387e-06,
+      "loss": 0.667,
+      "step": 8054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5550488006682308e-06,
+      "loss": 0.6859,
+      "step": 8055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.554385760599031e-06,
+      "loss": 0.6875,
+      "step": 8056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5537227981397557e-06,
+      "loss": 0.7343,
+      "step": 8057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5530599133448163e-06,
+      "loss": 0.6058,
+      "step": 8058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5523971062686188e-06,
+      "loss": 0.7489,
+      "step": 8059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5517343769655622e-06,
+      "loss": 0.6249,
+      "step": 8060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5510717254900394e-06,
+      "loss": 0.7191,
+      "step": 8061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5504091518964347e-06,
+      "loss": 0.7094,
+      "step": 8062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5497466562391305e-06,
+      "loss": 0.6634,
+      "step": 8063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5490842385724991e-06,
+      "loss": 0.7186,
+      "step": 8064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.548421898950906e-06,
+      "loss": 0.7442,
+      "step": 8065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5477596374287141e-06,
+      "loss": 0.6357,
+      "step": 8066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5470974540602764e-06,
+      "loss": 0.722,
+      "step": 8067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5464353488999406e-06,
+      "loss": 0.6103,
+      "step": 8068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.545773322002048e-06,
+      "loss": 0.6657,
+      "step": 8069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5451113734209336e-06,
+      "loss": 0.853,
+      "step": 8070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5444495032109263e-06,
+      "loss": 0.7404,
+      "step": 8071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5437877114263472e-06,
+      "loss": 0.7578,
+      "step": 8072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5431259981215136e-06,
+      "loss": 0.6198,
+      "step": 8073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5424643633507331e-06,
+      "loss": 0.8021,
+      "step": 8074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.541802807168309e-06,
+      "loss": 0.6572,
+      "step": 8075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5411413296285366e-06,
+      "loss": 0.6032,
+      "step": 8076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5404799307857077e-06,
+      "loss": 0.6699,
+      "step": 8077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5398186106941037e-06,
+      "loss": 0.605,
+      "step": 8078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5391573694080025e-06,
+      "loss": 0.6693,
+      "step": 8079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5384962069816745e-06,
+      "loss": 0.741,
+      "step": 8080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5378351234693837e-06,
+      "loss": 0.755,
+      "step": 8081
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5371741189253875e-06,
+      "loss": 0.7683,
+      "step": 8082
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5365131934039357e-06,
+      "loss": 0.6461,
+      "step": 8083
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.535852346959275e-06,
+      "loss": 0.8005,
+      "step": 8084
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5351915796456424e-06,
+      "loss": 0.6956,
+      "step": 8085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5345308915172682e-06,
+      "loss": 0.667,
+      "step": 8086
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5338702826283797e-06,
+      "loss": 0.7763,
+      "step": 8087
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5332097530331934e-06,
+      "loss": 0.689,
+      "step": 8088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5325493027859228e-06,
+      "loss": 0.7546,
+      "step": 8089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.531888931940772e-06,
+      "loss": 0.6651,
+      "step": 8090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5312286405519417e-06,
+      "loss": 0.7176,
+      "step": 8091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5305684286736233e-06,
+      "loss": 0.8522,
+      "step": 8092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5299082963600021e-06,
+      "loss": 0.8271,
+      "step": 8093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5292482436652591e-06,
+      "loss": 0.7244,
+      "step": 8094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5285882706435668e-06,
+      "loss": 0.6738,
+      "step": 8095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.52792837734909e-06,
+      "loss": 0.6424,
+      "step": 8096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5272685638359897e-06,
+      "loss": 0.6551,
+      "step": 8097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5266088301584193e-06,
+      "loss": 0.6743,
+      "step": 8098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5259491763705254e-06,
+      "loss": 0.7493,
+      "step": 8099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5252896025264469e-06,
+      "loss": 0.7195,
+      "step": 8100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5246301086803196e-06,
+      "loss": 0.567,
+      "step": 8101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5239706948862693e-06,
+      "loss": 0.7055,
+      "step": 8102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5233113611984157e-06,
+      "loss": 0.7191,
+      "step": 8103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5226521076708739e-06,
+      "loss": 0.6793,
+      "step": 8104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5219929343577511e-06,
+      "loss": 0.6772,
+      "step": 8105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.521333841313147e-06,
+      "loss": 0.7277,
+      "step": 8106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5206748285911562e-06,
+      "loss": 0.8265,
+      "step": 8107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5200158962458667e-06,
+      "loss": 0.6822,
+      "step": 8108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5193570443313593e-06,
+      "loss": 0.6769,
+      "step": 8109
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5186982729017069e-06,
+      "loss": 0.8907,
+      "step": 8110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5180395820109794e-06,
+      "loss": 0.7528,
+      "step": 8111
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.517380971713237e-06,
+      "loss": 0.8066,
+      "step": 8112
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5167224420625342e-06,
+      "loss": 0.6438,
+      "step": 8113
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5160639931129179e-06,
+      "loss": 0.6148,
+      "step": 8114
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5154056249184306e-06,
+      "loss": 0.7325,
+      "step": 8115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5147473375331064e-06,
+      "loss": 0.6085,
+      "step": 8116
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5140891310109734e-06,
+      "loss": 0.6129,
+      "step": 8117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.513431005406053e-06,
+      "loss": 0.6831,
+      "step": 8118
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5127729607723603e-06,
+      "loss": 0.6617,
+      "step": 8119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5121149971639032e-06,
+      "loss": 0.6947,
+      "step": 8120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5114571146346815e-06,
+      "loss": 0.6356,
+      "step": 8121
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5107993132386921e-06,
+      "loss": 0.6482,
+      "step": 8122
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5101415930299229e-06,
+      "loss": 0.6726,
+      "step": 8123
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5094839540623534e-06,
+      "loss": 0.7225,
+      "step": 8124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5088263963899607e-06,
+      "loss": 0.7148,
+      "step": 8125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5081689200667116e-06,
+      "loss": 0.7207,
+      "step": 8126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5075115251465682e-06,
+      "loss": 0.8072,
+      "step": 8127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.506854211683484e-06,
+      "loss": 0.704,
+      "step": 8128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5061969797314087e-06,
+      "loss": 0.7944,
+      "step": 8129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5055398293442833e-06,
+      "loss": 0.6676,
+      "step": 8130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5048827605760408e-06,
+      "loss": 0.5908,
+      "step": 8131
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5042257734806115e-06,
+      "loss": 0.76,
+      "step": 8132
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5035688681119153e-06,
+      "loss": 0.6195,
+      "step": 8133
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5029120445238671e-06,
+      "loss": 0.7418,
+      "step": 8134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.502255302770374e-06,
+      "loss": 0.775,
+      "step": 8135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5015986429053386e-06,
+      "loss": 0.7924,
+      "step": 8136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5009420649826546e-06,
+      "loss": 0.822,
+      "step": 8137
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5002855690562088e-06,
+      "loss": 0.6379,
+      "step": 8138
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.499629155179884e-06,
+      "loss": 0.6464,
+      "step": 8139
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4989728234075532e-06,
+      "loss": 0.7498,
+      "step": 8140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4983165737930833e-06,
+      "loss": 0.6259,
+      "step": 8141
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.497660406390337e-06,
+      "loss": 0.6785,
+      "step": 8142
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4970043212531668e-06,
+      "loss": 0.6804,
+      "step": 8143
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.49634831843542e-06,
+      "loss": 0.6888,
+      "step": 8144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.495692397990937e-06,
+      "loss": 0.7822,
+      "step": 8145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4950365599735522e-06,
+      "loss": 0.6389,
+      "step": 8146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.494380804437093e-06,
+      "loss": 0.8097,
+      "step": 8147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4937251314353777e-06,
+      "loss": 0.6339,
+      "step": 8148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4930695410222218e-06,
+      "loss": 0.6749,
+      "step": 8149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4924140332514308e-06,
+      "loss": 0.6253,
+      "step": 8150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4917586081768048e-06,
+      "loss": 0.6249,
+      "step": 8151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4911032658521358e-06,
+      "loss": 0.7317,
+      "step": 8152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4904480063312121e-06,
+      "loss": 0.7388,
+      "step": 8153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4897928296678118e-06,
+      "loss": 0.6563,
+      "step": 8154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4891377359157075e-06,
+      "loss": 0.6628,
+      "step": 8155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4884827251286666e-06,
+      "loss": 0.8274,
+      "step": 8156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4878277973604465e-06,
+      "loss": 0.7249,
+      "step": 8157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4871729526648006e-06,
+      "loss": 0.5929,
+      "step": 8158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4865181910954726e-06,
+      "loss": 0.8616,
+      "step": 8159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4858635127062032e-06,
+      "loss": 0.6772,
+      "step": 8160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4852089175507237e-06,
+      "loss": 0.6551,
+      "step": 8161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.484554405682757e-06,
+      "loss": 0.6625,
+      "step": 8162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4838999771560242e-06,
+      "loss": 0.7127,
+      "step": 8163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.483245632024235e-06,
+      "loss": 0.6776,
+      "step": 8164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.482591370341094e-06,
+      "loss": 0.6201,
+      "step": 8165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4819371921602982e-06,
+      "loss": 0.7332,
+      "step": 8166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4812830975355396e-06,
+      "loss": 0.8257,
+      "step": 8167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4806290865205019e-06,
+      "loss": 0.7016,
+      "step": 8168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4799751591688605e-06,
+      "loss": 0.6848,
+      "step": 8169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4793213155342875e-06,
+      "loss": 0.632,
+      "step": 8170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4786675556704455e-06,
+      "loss": 0.6837,
+      "step": 8171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4780138796309901e-06,
+      "loss": 0.8266,
+      "step": 8172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4773602874695714e-06,
+      "loss": 0.6955,
+      "step": 8173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4767067792398326e-06,
+      "loss": 0.6492,
+      "step": 8174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4760533549954093e-06,
+      "loss": 0.6334,
+      "step": 8175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4754000147899289e-06,
+      "loss": 0.594,
+      "step": 8176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4747467586770157e-06,
+      "loss": 0.7013,
+      "step": 8177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.474093586710283e-06,
+      "loss": 0.6969,
+      "step": 8178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.473440498943339e-06,
+      "loss": 0.7395,
+      "step": 8179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4727874954297864e-06,
+      "loss": 0.8405,
+      "step": 8180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.472134576223218e-06,
+      "loss": 0.8133,
+      "step": 8181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.471481741377222e-06,
+      "loss": 0.6645,
+      "step": 8182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4708289909453776e-06,
+      "loss": 0.7349,
+      "step": 8183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4701763249812604e-06,
+      "loss": 0.7172,
+      "step": 8184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.469523743538436e-06,
+      "loss": 0.6622,
+      "step": 8185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.468871246670463e-06,
+      "loss": 0.618,
+      "step": 8186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4682188344308961e-06,
+      "loss": 0.6984,
+      "step": 8187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4675665068732806e-06,
+      "loss": 0.5523,
+      "step": 8188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4669142640511546e-06,
+      "loss": 0.6467,
+      "step": 8189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4662621060180497e-06,
+      "loss": 0.7684,
+      "step": 8190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4656100328274919e-06,
+      "loss": 0.6532,
+      "step": 8191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4649580445329987e-06,
+      "loss": 0.6866,
+      "step": 8192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4643061411880805e-06,
+      "loss": 0.6939,
+      "step": 8193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.463654322846243e-06,
+      "loss": 0.6439,
+      "step": 8194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4630025895609816e-06,
+      "loss": 0.6909,
+      "step": 8195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.462350941385788e-06,
+      "loss": 0.6708,
+      "step": 8196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4616993783741423e-06,
+      "loss": 0.6888,
+      "step": 8197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4610479005795242e-06,
+      "loss": 0.8061,
+      "step": 8198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4603965080554005e-06,
+      "loss": 0.547,
+      "step": 8199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4597452008552335e-06,
+      "loss": 0.6765,
+      "step": 8200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4590939790324795e-06,
+      "loss": 0.7274,
+      "step": 8201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.458442842640586e-06,
+      "loss": 0.6482,
+      "step": 8202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4577917917329937e-06,
+      "loss": 0.7232,
+      "step": 8203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4571408263631362e-06,
+      "loss": 0.8288,
+      "step": 8204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.456489946584441e-06,
+      "loss": 0.7604,
+      "step": 8205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4558391524503293e-06,
+      "loss": 0.7614,
+      "step": 8206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4551884440142122e-06,
+      "loss": 0.6841,
+      "step": 8207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4545378213294976e-06,
+      "loss": 0.6729,
+      "step": 8208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4538872844495835e-06,
+      "loss": 0.7821,
+      "step": 8209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4532368334278618e-06,
+      "loss": 0.7724,
+      "step": 8210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.452586468317716e-06,
+      "loss": 0.6788,
+      "step": 8211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.451936189172526e-06,
+      "loss": 0.6273,
+      "step": 8212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4512859960456616e-06,
+      "loss": 0.5922,
+      "step": 8213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4506358889904857e-06,
+      "loss": 0.7428,
+      "step": 8214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4499858680603565e-06,
+      "loss": 0.8706,
+      "step": 8215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4493359333086226e-06,
+      "loss": 0.7382,
+      "step": 8216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.448686084788626e-06,
+      "loss": 0.64,
+      "step": 8217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4480363225537037e-06,
+      "loss": 0.7343,
+      "step": 8218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4473866466571823e-06,
+      "loss": 0.5681,
+      "step": 8219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4467370571523843e-06,
+      "loss": 0.7754,
+      "step": 8220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4460875540926218e-06,
+      "loss": 0.6561,
+      "step": 8221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4454381375312043e-06,
+      "loss": 0.6352,
+      "step": 8222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4447888075214306e-06,
+      "loss": 0.628,
+      "step": 8223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4441395641165922e-06,
+      "loss": 0.717,
+      "step": 8224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4434904073699769e-06,
+      "loss": 0.7408,
+      "step": 8225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4428413373348631e-06,
+      "loss": 0.7041,
+      "step": 8226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4421923540645218e-06,
+      "loss": 0.7139,
+      "step": 8227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4415434576122166e-06,
+      "loss": 0.6215,
+      "step": 8228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4408946480312063e-06,
+      "loss": 0.6253,
+      "step": 8229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4402459253747403e-06,
+      "loss": 0.6188,
+      "step": 8230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4395972896960603e-06,
+      "loss": 0.6987,
+      "step": 8231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4389487410484043e-06,
+      "loss": 0.6896,
+      "step": 8232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4383002794849998e-06,
+      "loss": 0.8221,
+      "step": 8233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4376519050590686e-06,
+      "loss": 0.7272,
+      "step": 8234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4370036178238245e-06,
+      "loss": 0.5935,
+      "step": 8235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4363554178324757e-06,
+      "loss": 0.7053,
+      "step": 8236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4357073051382222e-06,
+      "loss": 0.6745,
+      "step": 8237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4350592797942556e-06,
+      "loss": 0.8124,
+      "step": 8238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4344113418537636e-06,
+      "loss": 0.7103,
+      "step": 8239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4337634913699234e-06,
+      "loss": 0.5831,
+      "step": 8240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4331157283959072e-06,
+      "loss": 0.7261,
+      "step": 8241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4324680529848776e-06,
+      "loss": 0.662,
+      "step": 8242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4318204651899929e-06,
+      "loss": 0.7455,
+      "step": 8243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4311729650644035e-06,
+      "loss": 0.9387,
+      "step": 8244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4305255526612511e-06,
+      "loss": 0.7712,
+      "step": 8245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4298782280336718e-06,
+      "loss": 0.6352,
+      "step": 8246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4292309912347934e-06,
+      "loss": 0.7093,
+      "step": 8247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4285838423177375e-06,
+      "loss": 0.7623,
+      "step": 8248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4279367813356163e-06,
+      "loss": 0.6992,
+      "step": 8249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4272898083415383e-06,
+      "loss": 0.793,
+      "step": 8250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.426642923388602e-06,
+      "loss": 0.7554,
+      "step": 8251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4259961265298988e-06,
+      "loss": 0.8594,
+      "step": 8252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4253494178185156e-06,
+      "loss": 0.7342,
+      "step": 8253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4247027973075286e-06,
+      "loss": 0.7511,
+      "step": 8254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4240562650500079e-06,
+      "loss": 0.7633,
+      "step": 8255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4234098210990179e-06,
+      "loss": 0.6145,
+      "step": 8256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4227634655076144e-06,
+      "loss": 0.647,
+      "step": 8257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4221171983288458e-06,
+      "loss": 0.7209,
+      "step": 8258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4214710196157522e-06,
+      "loss": 0.7472,
+      "step": 8259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4208249294213702e-06,
+      "loss": 0.7327,
+      "step": 8260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4201789277987246e-06,
+      "loss": 0.6274,
+      "step": 8261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4195330148008369e-06,
+      "loss": 0.6604,
+      "step": 8262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4188871904807179e-06,
+      "loss": 0.7862,
+      "step": 8263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4182414548913741e-06,
+      "loss": 0.6702,
+      "step": 8264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.417595808085803e-06,
+      "loss": 0.6939,
+      "step": 8265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4169502501169936e-06,
+      "loss": 0.6719,
+      "step": 8266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.416304781037931e-06,
+      "loss": 1.0495,
+      "step": 8267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.415659400901591e-06,
+      "loss": 0.7036,
+      "step": 8268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4150141097609401e-06,
+      "loss": 0.8377,
+      "step": 8269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4143689076689427e-06,
+      "loss": 0.5879,
+      "step": 8270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.413723794678551e-06,
+      "loss": 0.7045,
+      "step": 8271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4130787708427124e-06,
+      "loss": 0.8324,
+      "step": 8272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4124338362143647e-06,
+      "loss": 0.7019,
+      "step": 8273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4117889908464426e-06,
+      "loss": 0.7047,
+      "step": 8274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4111442347918692e-06,
+      "loss": 0.705,
+      "step": 8275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4104995681035616e-06,
+      "loss": 0.6616,
+      "step": 8276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.409854990834431e-06,
+      "loss": 0.8445,
+      "step": 8277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.40921050303738e-06,
+      "loss": 0.7425,
+      "step": 8278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4085661047653025e-06,
+      "loss": 0.6906,
+      "step": 8279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.407921796071089e-06,
+      "loss": 0.766,
+      "step": 8280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.407277577007618e-06,
+      "loss": 0.6919,
+      "step": 8281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4066334476277646e-06,
+      "loss": 0.6104,
+      "step": 8282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4059894079843929e-06,
+      "loss": 0.6945,
+      "step": 8283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4053454581303637e-06,
+      "loss": 0.6877,
+      "step": 8284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4047015981185274e-06,
+      "loss": 0.7154,
+      "step": 8285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4040578280017275e-06,
+      "loss": 0.7053,
+      "step": 8286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4034141478327996e-06,
+      "loss": 0.7021,
+      "step": 8287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4027705576645744e-06,
+      "loss": 0.9674,
+      "step": 8288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4021270575498738e-06,
+      "loss": 0.6551,
+      "step": 8289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4014836475415102e-06,
+      "loss": 0.6806,
+      "step": 8290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4008403276922928e-06,
+      "loss": 0.6768,
+      "step": 8291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4001970980550203e-06,
+      "loss": 0.6543,
+      "step": 8292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3995539586824847e-06,
+      "loss": 0.7236,
+      "step": 8293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3989109096274694e-06,
+      "loss": 0.8481,
+      "step": 8294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3982679509427545e-06,
+      "loss": 0.8976,
+      "step": 8295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.397625082681108e-06,
+      "loss": 0.6558,
+      "step": 8296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3969823048952921e-06,
+      "loss": 0.7023,
+      "step": 8297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3963396176380637e-06,
+      "loss": 0.7172,
+      "step": 8298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.395697020962168e-06,
+      "loss": 0.7489,
+      "step": 8299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3950545149203478e-06,
+      "loss": 0.6288,
+      "step": 8300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3944120995653338e-06,
+      "loss": 0.7419,
+      "step": 8301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3937697749498527e-06,
+      "loss": 0.6191,
+      "step": 8302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.393127541126622e-06,
+      "loss": 0.7036,
+      "step": 8303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3924853981483504e-06,
+      "loss": 0.6928,
+      "step": 8304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3918433460677439e-06,
+      "loss": 0.7945,
+      "step": 8305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3912013849374963e-06,
+      "loss": 0.6438,
+      "step": 8306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3905595148102946e-06,
+      "loss": 0.6509,
+      "step": 8307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3899177357388218e-06,
+      "loss": 0.5981,
+      "step": 8308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3892760477757496e-06,
+      "loss": 0.8048,
+      "step": 8309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3886344509737437e-06,
+      "loss": 0.6886,
+      "step": 8310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3879929453854616e-06,
+      "loss": 0.8063,
+      "step": 8311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3873515310635555e-06,
+      "loss": 0.6813,
+      "step": 8312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3867102080606672e-06,
+      "loss": 0.6934,
+      "step": 8313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3860689764294322e-06,
+      "loss": 0.7772,
+      "step": 8314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3854278362224805e-06,
+      "loss": 0.7461,
+      "step": 8315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3847867874924312e-06,
+      "loss": 0.7942,
+      "step": 8316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.384145830291897e-06,
+      "loss": 0.6558,
+      "step": 8317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3835049646734852e-06,
+      "loss": 0.8074,
+      "step": 8318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3828641906897921e-06,
+      "loss": 0.763,
+      "step": 8319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3822235083934104e-06,
+      "loss": 0.6936,
+      "step": 8320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.381582917836921e-06,
+      "loss": 0.6607,
+      "step": 8321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3809424190729016e-06,
+      "loss": 0.6525,
+      "step": 8322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3803020121539188e-06,
+      "loss": 0.6596,
+      "step": 8323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3796616971325335e-06,
+      "loss": 0.737,
+      "step": 8324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3790214740612975e-06,
+      "loss": 0.6103,
+      "step": 8325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3783813429927585e-06,
+      "loss": 0.6898,
+      "step": 8326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3777413039794523e-06,
+      "loss": 0.6385,
+      "step": 8327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3771013570739094e-06,
+      "loss": 0.6027,
+      "step": 8328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3764615023286543e-06,
+      "loss": 0.6676,
+      "step": 8329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3758217397962004e-06,
+      "loss": 0.6957,
+      "step": 8330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3751820695290558e-06,
+      "loss": 0.5639,
+      "step": 8331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3745424915797197e-06,
+      "loss": 0.7432,
+      "step": 8332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3739030060006863e-06,
+      "loss": 0.7023,
+      "step": 8333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3732636128444393e-06,
+      "loss": 0.6789,
+      "step": 8334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3726243121634552e-06,
+      "loss": 0.7237,
+      "step": 8335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3719851040102058e-06,
+      "loss": 0.7579,
+      "step": 8336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.371345988437151e-06,
+      "loss": 0.7973,
+      "step": 8337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.370706965496747e-06,
+      "loss": 0.7963,
+      "step": 8338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3700680352414391e-06,
+      "loss": 0.672,
+      "step": 8339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3694291977236686e-06,
+      "loss": 0.6272,
+      "step": 8340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3687904529958662e-06,
+      "loss": 0.5998,
+      "step": 8341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3681518011104544e-06,
+      "loss": 0.7095,
+      "step": 8342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3675132421198522e-06,
+      "loss": 0.795,
+      "step": 8343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3668747760764671e-06,
+      "loss": 0.7822,
+      "step": 8344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3662364030326997e-06,
+      "loss": 0.7638,
+      "step": 8345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3655981230409453e-06,
+      "loss": 0.6788,
+      "step": 8346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3649599361535887e-06,
+      "loss": 0.8169,
+      "step": 8347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3643218424230079e-06,
+      "loss": 0.7322,
+      "step": 8348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.363683841901573e-06,
+      "loss": 0.7348,
+      "step": 8349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.363045934641649e-06,
+      "loss": 0.7346,
+      "step": 8350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3624081206955897e-06,
+      "loss": 0.7052,
+      "step": 8351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3617704001157426e-06,
+      "loss": 0.7116,
+      "step": 8352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3611327729544489e-06,
+      "loss": 0.8077,
+      "step": 8353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3604952392640402e-06,
+      "loss": 0.6601,
+      "step": 8354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3598577990968404e-06,
+      "loss": 0.7078,
+      "step": 8355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.359220452505168e-06,
+      "loss": 0.7251,
+      "step": 8356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3585831995413306e-06,
+      "loss": 0.6197,
+      "step": 8357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3579460402576322e-06,
+      "loss": 0.6937,
+      "step": 8358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.357308974706364e-06,
+      "loss": 0.794,
+      "step": 8359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3566720029398145e-06,
+      "loss": 0.6865,
+      "step": 8360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3560351250102617e-06,
+      "loss": 0.5995,
+      "step": 8361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3553983409699756e-06,
+      "loss": 0.6895,
+      "step": 8362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3547616508712186e-06,
+      "loss": 0.6843,
+      "step": 8363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3541250547662488e-06,
+      "loss": 0.7013,
+      "step": 8364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3534885527073121e-06,
+      "loss": 0.7044,
+      "step": 8365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3528521447466482e-06,
+      "loss": 0.7132,
+      "step": 8366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3522158309364902e-06,
+      "loss": 0.6464,
+      "step": 8367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.351579611329063e-06,
+      "loss": 0.7107,
+      "step": 8368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3509434859765823e-06,
+      "loss": 0.6932,
+      "step": 8369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3503074549312572e-06,
+      "loss": 0.6909,
+      "step": 8370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3496715182452901e-06,
+      "loss": 0.6477,
+      "step": 8371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3490356759708745e-06,
+      "loss": 0.8042,
+      "step": 8372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3483999281601943e-06,
+      "loss": 0.6923,
+      "step": 8373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3477642748654301e-06,
+      "loss": 0.5895,
+      "step": 8374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.34712871613875e-06,
+      "loss": 0.7141,
+      "step": 8375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3464932520323188e-06,
+      "loss": 0.7129,
+      "step": 8376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3458578825982892e-06,
+      "loss": 0.6475,
+      "step": 8377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3452226078888103e-06,
+      "loss": 0.7755,
+      "step": 8378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.34458742795602e-06,
+      "loss": 0.5788,
+      "step": 8379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3439523428520496e-06,
+      "loss": 0.792,
+      "step": 8380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3433173526290238e-06,
+      "loss": 0.7206,
+      "step": 8381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3426824573390583e-06,
+      "loss": 0.7893,
+      "step": 8382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3420476570342595e-06,
+      "loss": 0.6787,
+      "step": 8383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.34141295176673e-06,
+      "loss": 0.7703,
+      "step": 8384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.340778341588562e-06,
+      "loss": 0.7562,
+      "step": 8385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3401438265518393e-06,
+      "loss": 0.5984,
+      "step": 8386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.339509406708638e-06,
+      "loss": 0.7454,
+      "step": 8387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3388750821110297e-06,
+      "loss": 0.6419,
+      "step": 8388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.338240852811074e-06,
+      "loss": 0.6828,
+      "step": 8389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3376067188608238e-06,
+      "loss": 0.8136,
+      "step": 8390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3369726803123267e-06,
+      "loss": 0.6694,
+      "step": 8391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.336338737217619e-06,
+      "loss": 0.547,
+      "step": 8392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3357048896287306e-06,
+      "loss": 0.8566,
+      "step": 8393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3350711375976844e-06,
+      "loss": 0.7033,
+      "step": 8394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3344374811764937e-06,
+      "loss": 0.6838,
+      "step": 8395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3338039204171668e-06,
+      "loss": 0.777,
+      "step": 8396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3331704553717002e-06,
+      "loss": 0.7577,
+      "step": 8397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.332537086092086e-06,
+      "loss": 0.6516,
+      "step": 8398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3319038126303068e-06,
+      "loss": 0.7096,
+      "step": 8399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3312706350383375e-06,
+      "loss": 0.7791,
+      "step": 8400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.330637553368144e-06,
+      "loss": 0.6984,
+      "step": 8401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3300045676716877e-06,
+      "loss": 0.7733,
+      "step": 8402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3293716780009191e-06,
+      "loss": 0.6517,
+      "step": 8403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3287388844077806e-06,
+      "loss": 0.6655,
+      "step": 8404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3281061869442097e-06,
+      "loss": 0.7192,
+      "step": 8405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3274735856621335e-06,
+      "loss": 0.7326,
+      "step": 8406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3268410806134713e-06,
+      "loss": 0.7654,
+      "step": 8407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3262086718501343e-06,
+      "loss": 0.625,
+      "step": 8408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3255763594240287e-06,
+      "loss": 0.6024,
+      "step": 8409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3249441433870493e-06,
+      "loss": 0.604,
+      "step": 8410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.324312023791084e-06,
+      "loss": 0.646,
+      "step": 8411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.323680000688014e-06,
+      "loss": 0.6234,
+      "step": 8412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3230480741297108e-06,
+      "loss": 0.6501,
+      "step": 8413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.32241624416804e-06,
+      "loss": 0.6041,
+      "step": 8414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3217845108548567e-06,
+      "loss": 0.6371,
+      "step": 8415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.321152874242011e-06,
+      "loss": 0.7891,
+      "step": 8416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3205213343813433e-06,
+      "loss": 0.6575,
+      "step": 8417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3198898913246844e-06,
+      "loss": 0.5832,
+      "step": 8418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3192585451238621e-06,
+      "loss": 0.5957,
+      "step": 8419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3186272958306911e-06,
+      "loss": 0.7159,
+      "step": 8420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.31799614349698e-06,
+      "loss": 0.6337,
+      "step": 8421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.317365088174532e-06,
+      "loss": 0.6623,
+      "step": 8422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3167341299151385e-06,
+      "loss": 0.8085,
+      "step": 8423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3161032687705844e-06,
+      "loss": 0.5768,
+      "step": 8424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.315472504792646e-06,
+      "loss": 0.848,
+      "step": 8425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3148418380330942e-06,
+      "loss": 0.7204,
+      "step": 8426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3142112685436895e-06,
+      "loss": 0.7248,
+      "step": 8427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3135807963761832e-06,
+      "loss": 0.6765,
+      "step": 8428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.312950421582323e-06,
+      "loss": 0.7637,
+      "step": 8429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.312320144213845e-06,
+      "loss": 0.7675,
+      "step": 8430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.311689964322477e-06,
+      "loss": 0.7241,
+      "step": 8431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3110598819599423e-06,
+      "loss": 0.7248,
+      "step": 8432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3104298971779516e-06,
+      "loss": 0.5981,
+      "step": 8433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3098000100282127e-06,
+      "loss": 0.6687,
+      "step": 8434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3091702205624204e-06,
+      "loss": 0.8361,
+      "step": 8435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3085405288322657e-06,
+      "loss": 0.7615,
+      "step": 8436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.307910934889428e-06,
+      "loss": 0.694,
+      "step": 8437
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3072814387855815e-06,
+      "loss": 0.7052,
+      "step": 8438
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3066520405723893e-06,
+      "loss": 0.7433,
+      "step": 8439
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3060227403015107e-06,
+      "loss": 0.6037,
+      "step": 8440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3053935380245941e-06,
+      "loss": 0.6704,
+      "step": 8441
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3047644337932784e-06,
+      "loss": 0.7075,
+      "step": 8442
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.304135427659199e-06,
+      "loss": 0.6712,
+      "step": 8443
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3035065196739794e-06,
+      "loss": 0.6207,
+      "step": 8444
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3028777098892366e-06,
+      "loss": 0.7604,
+      "step": 8445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3022489983565779e-06,
+      "loss": 0.5916,
+      "step": 8446
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.301620385127606e-06,
+      "loss": 0.6738,
+      "step": 8447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3009918702539126e-06,
+      "loss": 0.788,
+      "step": 8448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3003634537870812e-06,
+      "loss": 0.711,
+      "step": 8449
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2997351357786902e-06,
+      "loss": 0.7993,
+      "step": 8450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2991069162803057e-06,
+      "loss": 0.7275,
+      "step": 8451
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2984787953434898e-06,
+      "loss": 0.5987,
+      "step": 8452
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2978507730197932e-06,
+      "loss": 0.7214,
+      "step": 8453
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2972228493607609e-06,
+      "loss": 0.8168,
+      "step": 8454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2965950244179288e-06,
+      "loss": 0.6326,
+      "step": 8455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2959672982428235e-06,
+      "loss": 0.6626,
+      "step": 8456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2953396708869661e-06,
+      "loss": 0.631,
+      "step": 8457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.294712142401868e-06,
+      "loss": 0.7017,
+      "step": 8458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2940847128390313e-06,
+      "loss": 0.6268,
+      "step": 8459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2934573822499535e-06,
+      "loss": 0.7533,
+      "step": 8460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2928301506861206e-06,
+      "loss": 0.7652,
+      "step": 8461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2922030181990123e-06,
+      "loss": 0.637,
+      "step": 8462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.291575984840098e-06,
+      "loss": 0.6442,
+      "step": 8463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2909490506608429e-06,
+      "loss": 0.6621,
+      "step": 8464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2903222157127005e-06,
+      "loss": 0.6672,
+      "step": 8465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2896954800471167e-06,
+      "loss": 0.6077,
+      "step": 8466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2890688437155315e-06,
+      "loss": 0.7071,
+      "step": 8467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2884423067693746e-06,
+      "loss": 0.7216,
+      "step": 8468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.2878158692600667e-06,
+      "loss": 0.6009,
+      "step": 8469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.287189531239024e-06,
+      "loss": 0.5788,
+      "step": 8470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2865632927576504e-06,
+      "loss": 0.7233,
+      "step": 8471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.285937153867345e-06,
+      "loss": 0.687,
+      "step": 8472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.285311114619496e-06,
+      "loss": 0.6379,
+      "step": 8473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2846851750654859e-06,
+      "loss": 0.7511,
+      "step": 8474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2840593352566877e-06,
+      "loss": 0.6849,
+      "step": 8475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2834335952444654e-06,
+      "loss": 0.7156,
+      "step": 8476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2828079550801752e-06,
+      "loss": 0.6417,
+      "step": 8477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2821824148151673e-06,
+      "loss": 0.5856,
+      "step": 8478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2815569745007815e-06,
+      "loss": 0.6488,
+      "step": 8479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2809316341883482e-06,
+      "loss": 0.7446,
+      "step": 8480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2803063939291943e-06,
+      "loss": 0.7326,
+      "step": 8481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2796812537746333e-06,
+      "loss": 0.6861,
+      "step": 8482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2790562137759737e-06,
+      "loss": 0.764,
+      "step": 8483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2784312739845133e-06,
+      "loss": 0.6122,
+      "step": 8484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2778064344515451e-06,
+      "loss": 0.6274,
+      "step": 8485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2771816952283512e-06,
+      "loss": 0.7987,
+      "step": 8486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2765570563662044e-06,
+      "loss": 0.6912,
+      "step": 8487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2759325179163739e-06,
+      "loss": 0.7498,
+      "step": 8488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2753080799301157e-06,
+      "loss": 0.6519,
+      "step": 8489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.274683742458681e-06,
+      "loss": 0.6841,
+      "step": 8490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2740595055533103e-06,
+      "loss": 0.6285,
+      "step": 8491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2734353692652382e-06,
+      "loss": 0.629,
+      "step": 8492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.272811333645689e-06,
+      "loss": 0.7127,
+      "step": 8493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2721873987458789e-06,
+      "loss": 0.6909,
+      "step": 8494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2715635646170178e-06,
+      "loss": 0.6514,
+      "step": 8495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2709398313103052e-06,
+      "loss": 0.682,
+      "step": 8496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2703161988769324e-06,
+      "loss": 0.5835,
+      "step": 8497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.269692667368085e-06,
+      "loss": 0.6115,
+      "step": 8498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2690692368349376e-06,
+      "loss": 0.8644,
+      "step": 8499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.268445907328657e-06,
+      "loss": 0.746,
+      "step": 8500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2678226789004012e-06,
+      "loss": 0.6184,
+      "step": 8501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.267199551601323e-06,
+      "loss": 0.6108,
+      "step": 8502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2665765254825633e-06,
+      "loss": 0.6851,
+      "step": 8503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2659536005952552e-06,
+      "loss": 0.666,
+      "step": 8504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2653307769905265e-06,
+      "loss": 0.8416,
+      "step": 8505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2647080547194934e-06,
+      "loss": 0.6359,
+      "step": 8506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2640854338332643e-06,
+      "loss": 0.6818,
+      "step": 8507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2634629143829413e-06,
+      "loss": 0.7518,
+      "step": 8508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2628404964196154e-06,
+      "loss": 0.6531,
+      "step": 8509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2622181799943722e-06,
+      "loss": 0.7194,
+      "step": 8510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2615959651582855e-06,
+      "loss": 0.6714,
+      "step": 8511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2609738519624249e-06,
+      "loss": 0.5555,
+      "step": 8512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2603518404578482e-06,
+      "loss": 0.6651,
+      "step": 8513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2597299306956062e-06,
+      "loss": 0.6346,
+      "step": 8514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2591081227267405e-06,
+      "loss": 0.6169,
+      "step": 8515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2584864166022863e-06,
+      "loss": 0.6355,
+      "step": 8516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2578648123732695e-06,
+      "loss": 0.6697,
+      "step": 8517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2572433100907054e-06,
+      "loss": 0.6015,
+      "step": 8518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.256621909805605e-06,
+      "loss": 0.666,
+      "step": 8519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.256000611568968e-06,
+      "loss": 0.6576,
+      "step": 8520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2553794154317867e-06,
+      "loss": 0.7796,
+      "step": 8521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2547583214450435e-06,
+      "loss": 0.7318,
+      "step": 8522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2541373296597158e-06,
+      "loss": 0.7722,
+      "step": 8523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2535164401267702e-06,
+      "loss": 0.7049,
+      "step": 8524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2528956528971642e-06,
+      "loss": 0.6658,
+      "step": 8525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2522749680218493e-06,
+      "loss": 0.7587,
+      "step": 8526
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.8022055625915527,
+      "eval_runtime": 436.8347,
+      "eval_samples_per_second": 6.865,
+      "eval_steps_per_second": 0.858,
+      "step": 8526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.251654385551766e-06,
+      "loss": 0.7142,
+      "step": 8527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2510339055378492e-06,
+      "loss": 0.6664,
+      "step": 8528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2504135280310227e-06,
+      "loss": 0.739,
+      "step": 8529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.249793253082204e-06,
+      "loss": 0.6507,
+      "step": 8530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2491730807423011e-06,
+      "loss": 0.6612,
+      "step": 8531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2485530110622124e-06,
+      "loss": 0.6402,
+      "step": 8532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2479330440928314e-06,
+      "loss": 0.6778,
+      "step": 8533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2473131798850397e-06,
+      "loss": 0.7179,
+      "step": 8534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2466934184897112e-06,
+      "loss": 0.7065,
+      "step": 8535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2460737599577135e-06,
+      "loss": 0.702,
+      "step": 8536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2454542043399034e-06,
+      "loss": 0.7247,
+      "step": 8537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2448347516871304e-06,
+      "loss": 0.5194,
+      "step": 8538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2442154020502333e-06,
+      "loss": 0.7834,
+      "step": 8539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2435961554800472e-06,
+      "loss": 0.6772,
+      "step": 8540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.242977012027394e-06,
+      "loss": 0.7978,
+      "step": 8541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.242357971743089e-06,
+      "loss": 0.7127,
+      "step": 8542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.24173903467794e-06,
+      "loss": 0.6944,
+      "step": 8543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2411202008827456e-06,
+      "loss": 0.5828,
+      "step": 8544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2405014704082934e-06,
+      "loss": 0.6105,
+      "step": 8545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2398828433053676e-06,
+      "loss": 0.7438,
+      "step": 8546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2392643196247389e-06,
+      "loss": 0.6755,
+      "step": 8547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2386458994171737e-06,
+      "loss": 0.6884,
+      "step": 8548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2380275827334263e-06,
+      "loss": 0.7681,
+      "step": 8549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2374093696242457e-06,
+      "loss": 0.7511,
+      "step": 8550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.23679126014037e-06,
+      "loss": 0.6726,
+      "step": 8551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2361732543325293e-06,
+      "loss": 0.688,
+      "step": 8552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2355553522514454e-06,
+      "loss": 0.685,
+      "step": 8553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.234937553947833e-06,
+      "loss": 0.6799,
+      "step": 8554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2343198594723963e-06,
+      "loss": 0.6708,
+      "step": 8555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2337022688758307e-06,
+      "loss": 0.6253,
+      "step": 8556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2330847822088258e-06,
+      "loss": 0.6445,
+      "step": 8557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2324673995220601e-06,
+      "loss": 0.7981,
+      "step": 8558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2318501208662042e-06,
+      "loss": 0.7019,
+      "step": 8559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2312329462919198e-06,
+      "loss": 0.6039,
+      "step": 8560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2306158758498623e-06,
+      "loss": 0.8079,
+      "step": 8561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.229998909590676e-06,
+      "loss": 0.7482,
+      "step": 8562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2293820475649962e-06,
+      "loss": 0.7375,
+      "step": 8563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2287652898234534e-06,
+      "loss": 0.7111,
+      "step": 8564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2281486364166648e-06,
+      "loss": 0.6526,
+      "step": 8565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2275320873952435e-06,
+      "loss": 0.6422,
+      "step": 8566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2269156428097898e-06,
+      "loss": 0.7005,
+      "step": 8567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2262993027108993e-06,
+      "loss": 0.6978,
+      "step": 8568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2256830671491564e-06,
+      "loss": 0.6977,
+      "step": 8569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2250669361751372e-06,
+      "loss": 0.9129,
+      "step": 8570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.224450909839411e-06,
+      "loss": 0.6353,
+      "step": 8571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2238349881925366e-06,
+      "loss": 0.7104,
+      "step": 8572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.223219171285064e-06,
+      "loss": 0.6066,
+      "step": 8573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2226034591675373e-06,
+      "loss": 0.6856,
+      "step": 8574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2219878518904893e-06,
+      "loss": 0.7734,
+      "step": 8575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2213723495044452e-06,
+      "loss": 0.5791,
+      "step": 8576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2207569520599199e-06,
+      "loss": 0.684,
+      "step": 8577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2201416596074243e-06,
+      "loss": 0.744,
+      "step": 8578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2195264721974553e-06,
+      "loss": 0.7629,
+      "step": 8579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2189113898805035e-06,
+      "loss": 0.6775,
+      "step": 8580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2182964127070529e-06,
+      "loss": 0.7264,
+      "step": 8581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2176815407275752e-06,
+      "loss": 0.6369,
+      "step": 8582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2170667739925348e-06,
+      "loss": 0.7295,
+      "step": 8583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2164521125523892e-06,
+      "loss": 0.6066,
+      "step": 8584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2158375564575845e-06,
+      "loss": 0.6583,
+      "step": 8585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2152231057585609e-06,
+      "loss": 0.7403,
+      "step": 8586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2146087605057469e-06,
+      "loss": 0.6333,
+      "step": 8587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2139945207495659e-06,
+      "loss": 0.8473,
+      "step": 8588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2133803865404298e-06,
+      "loss": 0.7054,
+      "step": 8589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2127663579287427e-06,
+      "loss": 0.7742,
+      "step": 8590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.2121524349648993e-06,
+      "loss": 0.5998,
+      "step": 8591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2115386176992883e-06,
+      "loss": 0.766,
+      "step": 8592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2109249061822866e-06,
+      "loss": 0.7113,
+      "step": 8593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.210311300464263e-06,
+      "loss": 0.7114,
+      "step": 8594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2096978005955803e-06,
+      "loss": 0.6636,
+      "step": 8595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2090844066265896e-06,
+      "loss": 0.7907,
+      "step": 8596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2084711186076345e-06,
+      "loss": 0.6374,
+      "step": 8597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2078579365890483e-06,
+      "loss": 0.6488,
+      "step": 8598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2072448606211593e-06,
+      "loss": 0.7239,
+      "step": 8599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2066318907542838e-06,
+      "loss": 0.7085,
+      "step": 8600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2060190270387292e-06,
+      "loss": 0.7387,
+      "step": 8601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2054062695247978e-06,
+      "loss": 0.6443,
+      "step": 8602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2047936182627787e-06,
+      "loss": 0.603,
+      "step": 8603
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.204181073302956e-06,
+      "loss": 0.6466,
+      "step": 8604
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2035686346956018e-06,
+      "loss": 0.7349,
+      "step": 8605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2029563024909831e-06,
+      "loss": 0.651,
+      "step": 8606
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.202344076739355e-06,
+      "loss": 0.6301,
+      "step": 8607
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2017319574909645e-06,
+      "loss": 0.7531,
+      "step": 8608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2011199447960517e-06,
+      "loss": 0.6081,
+      "step": 8609
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2005080387048459e-06,
+      "loss": 0.7587,
+      "step": 8610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.199896239267568e-06,
+      "loss": 0.6781,
+      "step": 8611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1992845465344316e-06,
+      "loss": 0.6406,
+      "step": 8612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1986729605556403e-06,
+      "loss": 0.7511,
+      "step": 8613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1980614813813884e-06,
+      "loss": 0.7315,
+      "step": 8614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1974501090618621e-06,
+      "loss": 0.6196,
+      "step": 8615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.19683884364724e-06,
+      "loss": 0.5857,
+      "step": 8616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1962276851876905e-06,
+      "loss": 0.7355,
+      "step": 8617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1956166337333723e-06,
+      "loss": 0.8427,
+      "step": 8618
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.195005689334438e-06,
+      "loss": 0.6682,
+      "step": 8619
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.19439485204103e-06,
+      "loss": 0.6371,
+      "step": 8620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1937841219032804e-06,
+      "loss": 0.6802,
+      "step": 8621
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1931734989713156e-06,
+      "loss": 0.7958,
+      "step": 8622
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1925629832952504e-06,
+      "loss": 0.606,
+      "step": 8623
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1919525749251934e-06,
+      "loss": 0.5238,
+      "step": 8624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1913422739112413e-06,
+      "loss": 0.6893,
+      "step": 8625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1907320803034855e-06,
+      "loss": 0.628,
+      "step": 8626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1901219941520056e-06,
+      "loss": 0.6359,
+      "step": 8627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.189512015506874e-06,
+      "loss": 0.7972,
+      "step": 8628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1889021444181525e-06,
+      "loss": 0.7548,
+      "step": 8629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1882923809358975e-06,
+      "loss": 0.7108,
+      "step": 8630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1876827251101532e-06,
+      "loss": 0.7837,
+      "step": 8631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1870731769909558e-06,
+      "loss": 0.6704,
+      "step": 8632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1864637366283347e-06,
+      "loss": 0.6303,
+      "step": 8633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1858544040723075e-06,
+      "loss": 0.5668,
+      "step": 8634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1852451793728849e-06,
+      "loss": 0.7642,
+      "step": 8635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1846360625800668e-06,
+      "loss": 0.7892,
+      "step": 8636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1840270537438476e-06,
+      "loss": 0.6333,
+      "step": 8637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.18341815291421e-06,
+      "loss": 0.6398,
+      "step": 8638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1828093601411275e-06,
+      "loss": 0.7706,
+      "step": 8639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1822006754745679e-06,
+      "loss": 0.8064,
+      "step": 8640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1815920989644863e-06,
+      "loss": 0.6341,
+      "step": 8641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1809836306608315e-06,
+      "loss": 0.6134,
+      "step": 8642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1803752706135437e-06,
+      "loss": 0.687,
+      "step": 8643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1797670188725525e-06,
+      "loss": 0.899,
+      "step": 8644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1791588754877788e-06,
+      "loss": 0.6007,
+      "step": 8645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1785508405091341e-06,
+      "loss": 0.6972,
+      "step": 8646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1779429139865246e-06,
+      "loss": 0.5983,
+      "step": 8647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1773350959698435e-06,
+      "loss": 0.6697,
+      "step": 8648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1767273865089755e-06,
+      "loss": 0.7005,
+      "step": 8649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1761197856537999e-06,
+      "loss": 0.6088,
+      "step": 8650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.175512293454183e-06,
+      "loss": 0.6183,
+      "step": 8651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1749049099599847e-06,
+      "loss": 0.7398,
+      "step": 8652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1742976352210533e-06,
+      "loss": 0.6187,
+      "step": 8653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1736904692872326e-06,
+      "loss": 0.7341,
+      "step": 8654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1730834122083536e-06,
+      "loss": 0.6431,
+      "step": 8655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1724764640342387e-06,
+      "loss": 0.7398,
+      "step": 8656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1718696248147043e-06,
+      "loss": 0.8061,
+      "step": 8657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1712628945995549e-06,
+      "loss": 0.6739,
+      "step": 8658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.170656273438586e-06,
+      "loss": 0.7113,
+      "step": 8659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1700497613815858e-06,
+      "loss": 0.6354,
+      "step": 8660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1694433584783342e-06,
+      "loss": 0.6387,
+      "step": 8661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1688370647786002e-06,
+      "loss": 0.6223,
+      "step": 8662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1682308803321429e-06,
+      "loss": 0.7156,
+      "step": 8663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1676248051887165e-06,
+      "loss": 0.611,
+      "step": 8664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1670188393980621e-06,
+      "loss": 0.6195,
+      "step": 8665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.166412983009914e-06,
+      "loss": 0.7149,
+      "step": 8666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.165807236073996e-06,
+      "loss": 0.6752,
+      "step": 8667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1652015986400257e-06,
+      "loss": 0.6341,
+      "step": 8668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1645960707577088e-06,
+      "loss": 0.78,
+      "step": 8669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1639906524767425e-06,
+      "loss": 0.6693,
+      "step": 8670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1633853438468174e-06,
+      "loss": 0.668,
+      "step": 8671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.162780144917612e-06,
+      "loss": 0.6603,
+      "step": 8672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.162175055738798e-06,
+      "loss": 0.6582,
+      "step": 8673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1615700763600355e-06,
+      "loss": 0.6485,
+      "step": 8674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1609652068309793e-06,
+      "loss": 0.7793,
+      "step": 8675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1603604472012725e-06,
+      "loss": 0.7711,
+      "step": 8676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.159755797520549e-06,
+      "loss": 0.7155,
+      "step": 8677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1591512578384362e-06,
+      "loss": 0.6907,
+      "step": 8678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1585468282045492e-06,
+      "loss": 0.6196,
+      "step": 8679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1579425086684962e-06,
+      "loss": 0.722,
+      "step": 8680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1573382992798773e-06,
+      "loss": 0.8224,
+      "step": 8681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1567342000882806e-06,
+      "loss": 0.7401,
+      "step": 8682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1561302111432873e-06,
+      "loss": 0.655,
+      "step": 8683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1555263324944673e-06,
+      "loss": 0.6604,
+      "step": 8684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1549225641913859e-06,
+      "loss": 0.6517,
+      "step": 8685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1543189062835943e-06,
+      "loss": 0.6904,
+      "step": 8686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1537153588206372e-06,
+      "loss": 0.5744,
+      "step": 8687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.153111921852051e-06,
+      "loss": 0.621,
+      "step": 8688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.152508595427361e-06,
+      "loss": 0.739,
+      "step": 8689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1519053795960845e-06,
+      "loss": 0.6327,
+      "step": 8690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1513022744077287e-06,
+      "loss": 0.6288,
+      "step": 8691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1506992799117944e-06,
+      "loss": 0.6252,
+      "step": 8692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1500963961577705e-06,
+      "loss": 0.6624,
+      "step": 8693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.149493623195137e-06,
+      "loss": 0.6402,
+      "step": 8694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1488909610733675e-06,
+      "loss": 0.6791,
+      "step": 8695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1482884098419237e-06,
+      "loss": 0.8193,
+      "step": 8696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1476859695502579e-06,
+      "loss": 0.6623,
+      "step": 8697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1470836402478155e-06,
+      "loss": 0.6403,
+      "step": 8698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1464814219840333e-06,
+      "loss": 0.6448,
+      "step": 8699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.145879314808336e-06,
+      "loss": 0.7128,
+      "step": 8700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1452773187701399e-06,
+      "loss": 0.6102,
+      "step": 8701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1446754339188547e-06,
+      "loss": 0.7074,
+      "step": 8702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1440736603038787e-06,
+      "loss": 0.8335,
+      "step": 8703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.143471997974601e-06,
+      "loss": 0.6435,
+      "step": 8704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.142870446980402e-06,
+      "loss": 0.7554,
+      "step": 8705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1422690073706544e-06,
+      "loss": 0.7436,
+      "step": 8706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1416676791947195e-06,
+      "loss": 0.7125,
+      "step": 8707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1410664625019499e-06,
+      "loss": 0.7004,
+      "step": 8708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1404653573416913e-06,
+      "loss": 0.7042,
+      "step": 8709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1398643637632776e-06,
+      "loss": 0.8081,
+      "step": 8710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1392634818160345e-06,
+      "loss": 0.7738,
+      "step": 8711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.1386627115492773e-06,
+      "loss": 0.6865,
+      "step": 8712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1380620530123157e-06,
+      "loss": 0.7503,
+      "step": 8713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.137461506254447e-06,
+      "loss": 0.7119,
+      "step": 8714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1368610713249587e-06,
+      "loss": 0.6662,
+      "step": 8715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1362607482731316e-06,
+      "loss": 0.7788,
+      "step": 8716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1356605371482378e-06,
+      "loss": 0.657,
+      "step": 8717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1350604379995369e-06,
+      "loss": 0.712,
+      "step": 8718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1344604508762825e-06,
+      "loss": 0.6687,
+      "step": 8719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.133860575827717e-06,
+      "loss": 0.7385,
+      "step": 8720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1332608129030745e-06,
+      "loss": 0.7858,
+      "step": 8721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1326611621515784e-06,
+      "loss": 0.637,
+      "step": 8722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1320616236224463e-06,
+      "loss": 0.6277,
+      "step": 8723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1314621973648834e-06,
+      "loss": 0.6443,
+      "step": 8724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.130862883428086e-06,
+      "loss": 0.65,
+      "step": 8725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1302636818612433e-06,
+      "loss": 0.7132,
+      "step": 8726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1296645927135336e-06,
+      "loss": 0.7423,
+      "step": 8727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1290656160341256e-06,
+      "loss": 0.7426,
+      "step": 8728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1284667518721793e-06,
+      "loss": 0.6302,
+      "step": 8729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.127868000276847e-06,
+      "loss": 0.6558,
+      "step": 8730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.127269361297269e-06,
+      "loss": 0.7563,
+      "step": 8731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1266708349825777e-06,
+      "loss": 0.647,
+      "step": 8732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1260724213818976e-06,
+      "loss": 0.704,
+      "step": 8733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1254741205443413e-06,
+      "loss": 0.7151,
+      "step": 8734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1248759325190147e-06,
+      "loss": 0.6801,
+      "step": 8735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1242778573550116e-06,
+      "loss": 0.5639,
+      "step": 8736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1236798951014203e-06,
+      "loss": 0.7536,
+      "step": 8737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.123082045807316e-06,
+      "loss": 0.8358,
+      "step": 8738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1224843095217664e-06,
+      "loss": 0.7078,
+      "step": 8739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1218866862938306e-06,
+      "loss": 0.7451,
+      "step": 8740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1212891761725578e-06,
+      "loss": 0.7616,
+      "step": 8741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1206917792069872e-06,
+      "loss": 0.7022,
+      "step": 8742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1200944954461485e-06,
+      "loss": 0.6637,
+      "step": 8743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1194973249390644e-06,
+      "loss": 0.5781,
+      "step": 8744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1189002677347467e-06,
+      "loss": 0.6735,
+      "step": 8745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1183033238821963e-06,
+      "loss": 0.8012,
+      "step": 8746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1177064934304088e-06,
+      "loss": 0.6242,
+      "step": 8747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.117109776428367e-06,
+      "loss": 0.6121,
+      "step": 8748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.116513172925046e-06,
+      "loss": 0.6965,
+      "step": 8749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1159166829694097e-06,
+      "loss": 0.7179,
+      "step": 8750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1153203066104163e-06,
+      "loss": 0.736,
+      "step": 8751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1147240438970106e-06,
+      "loss": 0.6451,
+      "step": 8752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1141278948781318e-06,
+      "loss": 0.6815,
+      "step": 8753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1135318596027061e-06,
+      "loss": 0.6284,
+      "step": 8754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1129359381196547e-06,
+      "loss": 0.7883,
+      "step": 8755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1123401304778844e-06,
+      "loss": 0.8185,
+      "step": 8756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1117444367262972e-06,
+      "loss": 0.7538,
+      "step": 8757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1111488569137832e-06,
+      "loss": 0.7449,
+      "step": 8758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1105533910892235e-06,
+      "loss": 0.6889,
+      "step": 8759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1099580393014892e-06,
+      "loss": 0.804,
+      "step": 8760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1093628015994452e-06,
+      "loss": 0.8643,
+      "step": 8761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.108767678031943e-06,
+      "loss": 0.773,
+      "step": 8762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1081726686478265e-06,
+      "loss": 0.5731,
+      "step": 8763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1075777734959314e-06,
+      "loss": 0.771,
+      "step": 8764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1069829926250822e-06,
+      "loss": 0.7896,
+      "step": 8765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1063883260840948e-06,
+      "loss": 0.7279,
+      "step": 8766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1057937739217745e-06,
+      "loss": 0.6551,
+      "step": 8767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1051993361869204e-06,
+      "loss": 0.6813,
+      "step": 8768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1046050129283186e-06,
+      "loss": 0.8529,
+      "step": 8769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1040108041947472e-06,
+      "loss": 0.671,
+      "step": 8770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1034167100349763e-06,
+      "loss": 0.6695,
+      "step": 8771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1028227304977637e-06,
+      "loss": 0.7583,
+      "step": 8772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1022288656318614e-06,
+      "loss": 0.6055,
+      "step": 8773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1016351154860078e-06,
+      "loss": 0.7239,
+      "step": 8774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.101041480108936e-06,
+      "loss": 0.6279,
+      "step": 8775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.1004479595493672e-06,
+      "loss": 0.7822,
+      "step": 8776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0998545538560124e-06,
+      "loss": 0.6723,
+      "step": 8777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0992612630775768e-06,
+      "loss": 0.6507,
+      "step": 8778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0986680872627525e-06,
+      "loss": 0.7342,
+      "step": 8779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0980750264602237e-06,
+      "loss": 0.7246,
+      "step": 8780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0974820807186643e-06,
+      "loss": 0.7599,
+      "step": 8781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0968892500867408e-06,
+      "loss": 0.6899,
+      "step": 8782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0962965346131088e-06,
+      "loss": 0.6608,
+      "step": 8783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.095703934346413e-06,
+      "loss": 0.6054,
+      "step": 8784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.095111449335292e-06,
+      "loss": 0.7371,
+      "step": 8785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0945190796283728e-06,
+      "loss": 0.6577,
+      "step": 8786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.093926825274273e-06,
+      "loss": 0.7116,
+      "step": 8787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0933346863215998e-06,
+      "loss": 0.7084,
+      "step": 8788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0927426628189543e-06,
+      "loss": 0.6308,
+      "step": 8789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0921507548149242e-06,
+      "loss": 0.7638,
+      "step": 8790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0915589623580913e-06,
+      "loss": 0.7379,
+      "step": 8791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0909672854970244e-06,
+      "loss": 0.6489,
+      "step": 8792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.090375724280286e-06,
+      "loss": 0.7181,
+      "step": 8793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0897842787564255e-06,
+      "loss": 0.6999,
+      "step": 8794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0891929489739876e-06,
+      "loss": 0.6492,
+      "step": 8795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0886017349815036e-06,
+      "loss": 0.714,
+      "step": 8796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0880106368274968e-06,
+      "loss": 0.6482,
+      "step": 8797
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.087419654560479e-06,
+      "loss": 0.6615,
+      "step": 8798
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0868287882289565e-06,
+      "loss": 0.586,
+      "step": 8799
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0862380378814232e-06,
+      "loss": 0.7244,
+      "step": 8800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.085647403566363e-06,
+      "loss": 0.7893,
+      "step": 8801
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0850568853322529e-06,
+      "loss": 0.5992,
+      "step": 8802
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0844664832275584e-06,
+      "loss": 0.7261,
+      "step": 8803
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0838761973007353e-06,
+      "loss": 0.6113,
+      "step": 8804
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0832860276002299e-06,
+      "loss": 0.8067,
+      "step": 8805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0826959741744816e-06,
+      "loss": 0.6925,
+      "step": 8806
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0821060370719167e-06,
+      "loss": 0.6325,
+      "step": 8807
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0815162163409529e-06,
+      "loss": 0.5688,
+      "step": 8808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0809265120300008e-06,
+      "loss": 0.6005,
+      "step": 8809
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0803369241874572e-06,
+      "loss": 0.6304,
+      "step": 8810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0797474528617141e-06,
+      "loss": 0.6305,
+      "step": 8811
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0791580981011493e-06,
+      "loss": 0.625,
+      "step": 8812
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0785688599541353e-06,
+      "loss": 0.7497,
+      "step": 8813
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0779797384690319e-06,
+      "loss": 0.7023,
+      "step": 8814
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0773907336941895e-06,
+      "loss": 0.7407,
+      "step": 8815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0768018456779516e-06,
+      "loss": 0.747,
+      "step": 8816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0762130744686497e-06,
+      "loss": 0.7007,
+      "step": 8817
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0756244201146057e-06,
+      "loss": 0.6539,
+      "step": 8818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0750358826641325e-06,
+      "loss": 0.6968,
+      "step": 8819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0744474621655348e-06,
+      "loss": 0.6878,
+      "step": 8820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0738591586671054e-06,
+      "loss": 0.6608,
+      "step": 8821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0732709722171277e-06,
+      "loss": 0.6384,
+      "step": 8822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.072682902863878e-06,
+      "loss": 0.691,
+      "step": 8823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0720949506556206e-06,
+      "loss": 0.8099,
+      "step": 8824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0715071156406103e-06,
+      "loss": 0.8672,
+      "step": 8825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0709193978670925e-06,
+      "loss": 0.6151,
+      "step": 8826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0703317973833046e-06,
+      "loss": 0.6659,
+      "step": 8827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0697443142374717e-06,
+      "loss": 0.6502,
+      "step": 8828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0691569484778123e-06,
+      "loss": 0.6743,
+      "step": 8829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0685697001525314e-06,
+      "loss": 0.6263,
+      "step": 8830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0679825693098287e-06,
+      "loss": 0.8244,
+      "step": 8831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0673955559978902e-06,
+      "loss": 0.7761,
+      "step": 8832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0668086602648962e-06,
+      "loss": 0.6805,
+      "step": 8833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.0662218821590142e-06,
+      "loss": 0.6536,
+      "step": 8834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0656352217284032e-06,
+      "loss": 0.6645,
+      "step": 8835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0650486790212117e-06,
+      "loss": 0.6584,
+      "step": 8836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0644622540855813e-06,
+      "loss": 0.7734,
+      "step": 8837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0638759469696406e-06,
+      "loss": 0.6546,
+      "step": 8838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0632897577215093e-06,
+      "loss": 0.6468,
+      "step": 8839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0627036863892998e-06,
+      "loss": 0.6548,
+      "step": 8840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0621177330211125e-06,
+      "loss": 0.6227,
+      "step": 8841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0615318976650379e-06,
+      "loss": 0.782,
+      "step": 8842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0609461803691574e-06,
+      "loss": 0.7369,
+      "step": 8843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0603605811815442e-06,
+      "loss": 0.8216,
+      "step": 8844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0597751001502604e-06,
+      "loss": 0.7968,
+      "step": 8845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0591897373233566e-06,
+      "loss": 0.7702,
+      "step": 8846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.058604492748878e-06,
+      "loss": 0.5756,
+      "step": 8847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0580193664748558e-06,
+      "loss": 0.6601,
+      "step": 8848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0574343585493152e-06,
+      "loss": 0.6054,
+      "step": 8849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.056849469020268e-06,
+      "loss": 0.6769,
+      "step": 8850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0562646979357202e-06,
+      "loss": 0.6451,
+      "step": 8851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0556800453436646e-06,
+      "loss": 0.6922,
+      "step": 8852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0550955112920857e-06,
+      "loss": 0.8006,
+      "step": 8853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0545110958289593e-06,
+      "loss": 0.7688,
+      "step": 8854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0539267990022498e-06,
+      "loss": 0.6455,
+      "step": 8855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0533426208599127e-06,
+      "loss": 0.7777,
+      "step": 8856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0527585614498923e-06,
+      "loss": 0.7395,
+      "step": 8857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0521746208201266e-06,
+      "loss": 0.7051,
+      "step": 8858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0515907990185407e-06,
+      "loss": 0.696,
+      "step": 8859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0510070960930498e-06,
+      "loss": 0.6503,
+      "step": 8860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0504235120915627e-06,
+      "loss": 0.668,
+      "step": 8861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0498400470619749e-06,
+      "loss": 0.7208,
+      "step": 8862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0492567010521736e-06,
+      "loss": 0.814,
+      "step": 8863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0486734741100354e-06,
+      "loss": 0.7198,
+      "step": 8864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0480903662834293e-06,
+      "loss": 0.6423,
+      "step": 8865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0475073776202114e-06,
+      "loss": 0.6905,
+      "step": 8866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0469245081682314e-06,
+      "loss": 0.6423,
+      "step": 8867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0463417579753254e-06,
+      "loss": 0.6326,
+      "step": 8868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0457591270893241e-06,
+      "loss": 0.7509,
+      "step": 8869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0451766155580443e-06,
+      "loss": 0.7222,
+      "step": 8870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.044594223429296e-06,
+      "loss": 0.7712,
+      "step": 8871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0440119507508776e-06,
+      "loss": 0.7151,
+      "step": 8872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0434297975705784e-06,
+      "loss": 0.6732,
+      "step": 8873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.042847763936177e-06,
+      "loss": 0.5978,
+      "step": 8874
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0422658498954443e-06,
+      "loss": 0.7825,
+      "step": 8875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0416840554961396e-06,
+      "loss": 0.7776,
+      "step": 8876
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0411023807860115e-06,
+      "loss": 0.7463,
+      "step": 8877
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0405208258128025e-06,
+      "loss": 0.6727,
+      "step": 8878
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0399393906242416e-06,
+      "loss": 0.6317,
+      "step": 8879
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0393580752680493e-06,
+      "loss": 0.5925,
+      "step": 8880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0387768797919354e-06,
+      "loss": 0.6005,
+      "step": 8881
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0381958042436021e-06,
+      "loss": 0.7214,
+      "step": 8882
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0376148486707399e-06,
+      "loss": 0.7882,
+      "step": 8883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0370340131210292e-06,
+      "loss": 0.6177,
+      "step": 8884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0364532976421423e-06,
+      "loss": 0.7279,
+      "step": 8885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0358727022817394e-06,
+      "loss": 0.8035,
+      "step": 8886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0352922270874734e-06,
+      "loss": 0.6461,
+      "step": 8887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0347118721069846e-06,
+      "loss": 0.6583,
+      "step": 8888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0341316373879062e-06,
+      "loss": 0.6439,
+      "step": 8889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0335515229778591e-06,
+      "loss": 0.6934,
+      "step": 8890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.032971528924455e-06,
+      "loss": 0.6251,
+      "step": 8891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.032391655275298e-06,
+      "loss": 0.687,
+      "step": 8892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0318119020779787e-06,
+      "loss": 0.7773,
+      "step": 8893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.03123226938008e-06,
+      "loss": 0.6031,
+      "step": 8894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0306527572291736e-06,
+      "loss": 0.7234,
+      "step": 8895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0300733656728234e-06,
+      "loss": 0.5777,
+      "step": 8896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0294940947585818e-06,
+      "loss": 0.7994,
+      "step": 8897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0289149445339908e-06,
+      "loss": 0.7341,
+      "step": 8898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0283359150465844e-06,
+      "loss": 0.8691,
+      "step": 8899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0277570063438852e-06,
+      "loss": 0.802,
+      "step": 8900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0271782184734064e-06,
+      "loss": 0.7916,
+      "step": 8901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0265995514826502e-06,
+      "loss": 0.7333,
+      "step": 8902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0260210054191113e-06,
+      "loss": 0.789,
+      "step": 8903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0254425803302717e-06,
+      "loss": 0.6997,
+      "step": 8904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0248642762636063e-06,
+      "loss": 0.6571,
+      "step": 8905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0242860932665769e-06,
+      "loss": 0.6408,
+      "step": 8906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.023708031386639e-06,
+      "loss": 0.7303,
+      "step": 8907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.023130090671234e-06,
+      "loss": 0.6521,
+      "step": 8908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0225522711677976e-06,
+      "loss": 0.6294,
+      "step": 8909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0219745729237525e-06,
+      "loss": 0.6353,
+      "step": 8910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0213969959865127e-06,
+      "loss": 0.6196,
+      "step": 8911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.020819540403481e-06,
+      "loss": 0.8193,
+      "step": 8912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0202422062220529e-06,
+      "loss": 0.6565,
+      "step": 8913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0196649934896114e-06,
+      "loss": 0.7326,
+      "step": 8914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0190879022535293e-06,
+      "loss": 0.7586,
+      "step": 8915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0185109325611728e-06,
+      "loss": 0.6873,
+      "step": 8916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0179340844598944e-06,
+      "loss": 0.7113,
+      "step": 8917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0173573579970389e-06,
+      "loss": 0.656,
+      "step": 8918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0167807532199386e-06,
+      "loss": 0.5456,
+      "step": 8919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0162042701759196e-06,
+      "loss": 0.7397,
+      "step": 8920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0156279089122948e-06,
+      "loss": 0.641,
+      "step": 8921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0150516694763677e-06,
+      "loss": 0.7023,
+      "step": 8922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.014475551915434e-06,
+      "loss": 0.6207,
+      "step": 8923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0138995562767759e-06,
+      "loss": 0.7344,
+      "step": 8924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0133236826076689e-06,
+      "loss": 0.7349,
+      "step": 8925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0127479309553755e-06,
+      "loss": 0.6907,
+      "step": 8926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0121723013671518e-06,
+      "loss": 0.6904,
+      "step": 8927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0115967938902402e-06,
+      "loss": 0.7378,
+      "step": 8928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0110214085718742e-06,
+      "loss": 0.7953,
+      "step": 8929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0104461454592794e-06,
+      "loss": 0.7806,
+      "step": 8930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0098710045996687e-06,
+      "loss": 0.6877,
+      "step": 8931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.009295986040246e-06,
+      "loss": 0.6427,
+      "step": 8932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0087210898282044e-06,
+      "loss": 0.6325,
+      "step": 8933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0081463160107292e-06,
+      "loss": 0.6397,
+      "step": 8934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0075716646349933e-06,
+      "loss": 0.7139,
+      "step": 8935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0069971357481598e-06,
+      "loss": 0.6456,
+      "step": 8936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0064227293973836e-06,
+      "loss": 0.5544,
+      "step": 8937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0058484456298078e-06,
+      "loss": 0.7141,
+      "step": 8938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0052742844925658e-06,
+      "loss": 0.749,
+      "step": 8939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.00470024603278e-06,
+      "loss": 0.6842,
+      "step": 8940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0041263302975657e-06,
+      "loss": 0.8894,
+      "step": 8941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0035525373340244e-06,
+      "loss": 0.7214,
+      "step": 8942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0029788671892512e-06,
+      "loss": 0.7304,
+      "step": 8943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0024053199103273e-06,
+      "loss": 0.7254,
+      "step": 8944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0018318955443276e-06,
+      "loss": 0.8124,
+      "step": 8945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0012585941383135e-06,
+      "loss": 0.6393,
+      "step": 8946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0006854157393394e-06,
+      "loss": 0.8157,
+      "step": 8947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.0001123603944473e-06,
+      "loss": 0.7411,
+      "step": 8948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.995394281506698e-07,
+      "loss": 0.752,
+      "step": 8949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.989666190550288e-07,
+      "loss": 0.7468,
+      "step": 8950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.983939331545384e-07,
+      "loss": 0.6171,
+      "step": 8951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.978213704962e-07,
+      "loss": 0.5879,
+      "step": 8952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.97248931127005e-07,
+      "loss": 0.6565,
+      "step": 8953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.966766150939378e-07,
+      "loss": 0.7672,
+      "step": 8954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.96104422443969e-07,
+      "loss": 0.6811,
+      "step": 8955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.955323532240605e-07,
+      "loss": 0.6526,
+      "step": 8956
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.94960407481163e-07,
+      "loss": 0.7101,
+      "step": 8957
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.943885852622206e-07,
+      "loss": 0.8761,
+      "step": 8958
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.938168866141634e-07,
+      "loss": 0.6794,
+      "step": 8959
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.932453115839117e-07,
+      "loss": 0.6911,
+      "step": 8960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.92673860218379e-07,
+      "loss": 0.7023,
+      "step": 8961
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.921025325644639e-07,
+      "loss": 0.7154,
+      "step": 8962
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.915313286690598e-07,
+      "loss": 0.6872,
+      "step": 8963
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.909602485790452e-07,
+      "loss": 0.7613,
+      "step": 8964
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.903892923412924e-07,
+      "loss": 0.7735,
+      "step": 8965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.89818460002661e-07,
+      "loss": 0.6489,
+      "step": 8966
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.892477516100008e-07,
+      "loss": 0.7822,
+      "step": 8967
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.886771672101531e-07,
+      "loss": 0.6966,
+      "step": 8968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.88106706849947e-07,
+      "loss": 0.6125,
+      "step": 8969
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.875363705762023e-07,
+      "loss": 0.6746,
+      "step": 8970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.869661584357276e-07,
+      "loss": 0.6662,
+      "step": 8971
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.863960704753241e-07,
+      "loss": 0.6908,
+      "step": 8972
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.858261067417802e-07,
+      "loss": 0.723,
+      "step": 8973
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852562672818733e-07,
+      "loss": 0.648,
+      "step": 8974
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.846865521423746e-07,
+      "loss": 0.5904,
+      "step": 8975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.841169613700416e-07,
+      "loss": 0.6983,
+      "step": 8976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.835474950116226e-07,
+      "loss": 0.7326,
+      "step": 8977
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.829781531138546e-07,
+      "loss": 0.6297,
+      "step": 8978
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.824089357234678e-07,
+      "loss": 0.7088,
+      "step": 8979
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.818398428871778e-07,
+      "loss": 0.7297,
+      "step": 8980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.812708746516939e-07,
+      "loss": 0.77,
+      "step": 8981
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.807020310637113e-07,
+      "loss": 0.7093,
+      "step": 8982
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.801333121699193e-07,
+      "loss": 0.6372,
+      "step": 8983
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.795647180169926e-07,
+      "loss": 0.8468,
+      "step": 8984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.789962486515998e-07,
+      "loss": 0.8455,
+      "step": 8985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.784279041203956e-07,
+      "loss": 0.6457,
+      "step": 8986
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.77859684470027e-07,
+      "loss": 0.6628,
+      "step": 8987
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.772915897471282e-07,
+      "loss": 0.6049,
+      "step": 8988
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.76723619998327e-07,
+      "loss": 0.5886,
+      "step": 8989
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.761557752702374e-07,
+      "loss": 0.7464,
+      "step": 8990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.755880556094641e-07,
+      "loss": 0.6166,
+      "step": 8991
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.75020461062603e-07,
+      "loss": 0.8325,
+      "step": 8992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.744529916762383e-07,
+      "loss": 0.7771,
+      "step": 8993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.738856474969435e-07,
+      "loss": 0.7661,
+      "step": 8994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.733184285712826e-07,
+      "loss": 0.6282,
+      "step": 8995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.727513349458104e-07,
+      "loss": 0.6945,
+      "step": 8996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.721843666670696e-07,
+      "loss": 0.6548,
+      "step": 8997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.716175237815923e-07,
+      "loss": 0.6776,
+      "step": 8998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.710508063359034e-07,
+      "loss": 0.7777,
+      "step": 8999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.704842143765133e-07,
+      "loss": 0.7288,
+      "step": 9000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.699177479499262e-07,
+      "loss": 0.5676,
+      "step": 9001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.69351407102632e-07,
+      "loss": 0.6451,
+      "step": 9002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.687851918811143e-07,
+      "loss": 0.7471,
+      "step": 9003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.682191023318435e-07,
+      "loss": 0.6533,
+      "step": 9004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.676531385012797e-07,
+      "loss": 0.6182,
+      "step": 9005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.670873004358752e-07,
+      "loss": 0.6222,
+      "step": 9006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.665215881820697e-07,
+      "loss": 0.7653,
+      "step": 9007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.65956001786293e-07,
+      "loss": 0.6502,
+      "step": 9008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.653905412949638e-07,
+      "loss": 0.7227,
+      "step": 9009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.648252067544933e-07,
+      "loss": 0.7977,
+      "step": 9010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.642599982112798e-07,
+      "loss": 0.7094,
+      "step": 9011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.636949157117115e-07,
+      "loss": 0.5751,
+      "step": 9012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.631299593021677e-07,
+      "loss": 0.6145,
+      "step": 9013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.625651290290155e-07,
+      "loss": 0.715,
+      "step": 9014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.62000424938613e-07,
+      "loss": 0.5998,
+      "step": 9015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.614358470773067e-07,
+      "loss": 0.9692,
+      "step": 9016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.608713954914348e-07,
+      "loss": 0.7607,
+      "step": 9017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.603070702273224e-07,
+      "loss": 0.8036,
+      "step": 9018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.597428713312874e-07,
+      "loss": 0.7416,
+      "step": 9019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.591787988496334e-07,
+      "loss": 0.6552,
+      "step": 9020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.586148528286584e-07,
+      "loss": 0.6945,
+      "step": 9021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.580510333146451e-07,
+      "loss": 0.6749,
+      "step": 9022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.5748734035387e-07,
+      "loss": 0.7824,
+      "step": 9023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.56923773992597e-07,
+      "loss": 0.6872,
+      "step": 9024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.563603342770795e-07,
+      "loss": 0.6518,
+      "step": 9025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.5579702125356e-07,
+      "loss": 0.7534,
+      "step": 9026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.552338349682737e-07,
+      "loss": 0.6146,
+      "step": 9027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.546707754674425e-07,
+      "loss": 0.6504,
+      "step": 9028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.541078427972776e-07,
+      "loss": 0.8342,
+      "step": 9029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.535450370039825e-07,
+      "loss": 0.683,
+      "step": 9030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.52982358133748e-07,
+      "loss": 0.603,
+      "step": 9031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.524198062327553e-07,
+      "loss": 0.6662,
+      "step": 9032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.518573813471741e-07,
+      "loss": 0.8032,
+      "step": 9033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.51295083523166e-07,
+      "loss": 0.6763,
+      "step": 9034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.507329128068801e-07,
+      "loss": 0.7165,
+      "step": 9035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.501708692444553e-07,
+      "loss": 0.6416,
+      "step": 9036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.496089528820218e-07,
+      "loss": 0.5752,
+      "step": 9037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.490471637656962e-07,
+      "loss": 0.8276,
+      "step": 9038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.484855019415886e-07,
+      "loss": 0.7109,
+      "step": 9039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.47923967455795e-07,
+      "loss": 0.6842,
+      "step": 9040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.47362560354404e-07,
+      "loss": 0.7735,
+      "step": 9041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.468012806834912e-07,
+      "loss": 0.6782,
+      "step": 9042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.462401284891224e-07,
+      "loss": 0.683,
+      "step": 9043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.456791038173546e-07,
+      "loss": 0.6799,
+      "step": 9044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.451182067142328e-07,
+      "loss": 0.7036,
+      "step": 9045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.445574372257916e-07,
+      "loss": 0.6874,
+      "step": 9046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.439967953980539e-07,
+      "loss": 0.7839,
+      "step": 9047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.434362812770361e-07,
+      "loss": 0.7752,
+      "step": 9048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.428758949087405e-07,
+      "loss": 0.6973,
+      "step": 9049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.423156363391592e-07,
+      "loss": 0.7145,
+      "step": 9050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.41755505614276e-07,
+      "loss": 0.6711,
+      "step": 9051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.41195502780062e-07,
+      "loss": 0.7203,
+      "step": 9052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.406356278824791e-07,
+      "loss": 0.7049,
+      "step": 9053
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.400758809674771e-07,
+      "loss": 0.649,
+      "step": 9054
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.395162620809978e-07,
+      "loss": 0.6178,
+      "step": 9055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.389567712689698e-07,
+      "loss": 0.7172,
+      "step": 9056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.383974085773143e-07,
+      "loss": 0.7554,
+      "step": 9057
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.378381740519382e-07,
+      "loss": 0.723,
+      "step": 9058
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.372790677387417e-07,
+      "loss": 0.6537,
+      "step": 9059
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.367200896836115e-07,
+      "loss": 0.7572,
+      "step": 9060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.361612399324246e-07,
+      "loss": 0.6859,
+      "step": 9061
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.356025185310491e-07,
+      "loss": 0.7327,
+      "step": 9062
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.350439255253405e-07,
+      "loss": 0.7766,
+      "step": 9063
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.344854609611436e-07,
+      "loss": 0.6453,
+      "step": 9064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.339271248842954e-07,
+      "loss": 0.7759,
+      "step": 9065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.333689173406196e-07,
+      "loss": 0.7865,
+      "step": 9066
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.328108383759297e-07,
+      "loss": 0.7313,
+      "step": 9067
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.322528880360304e-07,
+      "loss": 0.7351,
+      "step": 9068
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.31695066366714e-07,
+      "loss": 0.5772,
+      "step": 9069
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.311373734137632e-07,
+      "loss": 0.5851,
+      "step": 9070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.30579809222949e-07,
+      "loss": 0.8634,
+      "step": 9071
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.300223738400337e-07,
+      "loss": 0.6199,
+      "step": 9072
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.29465067310768e-07,
+      "loss": 0.6483,
+      "step": 9073
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.289078896808909e-07,
+      "loss": 0.6528,
+      "step": 9074
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.283508409961334e-07,
+      "loss": 0.7155,
+      "step": 9075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.277939213022128e-07,
+      "loss": 0.749,
+      "step": 9076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.272371306448394e-07,
+      "loss": 0.7614,
+      "step": 9077
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.266804690697093e-07,
+      "loss": 0.8316,
+      "step": 9078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.261239366225114e-07,
+      "loss": 0.7692,
+      "step": 9079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.255675333489211e-07,
+      "loss": 0.702,
+      "step": 9080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.25011259294604e-07,
+      "loss": 1.0311,
+      "step": 9081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.24455114505217e-07,
+      "loss": 0.7693,
+      "step": 9082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.238990990264038e-07,
+      "loss": 0.8332,
+      "step": 9083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.233432129037986e-07,
+      "loss": 0.7293,
+      "step": 9084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.227874561830247e-07,
+      "loss": 0.8142,
+      "step": 9085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.222318289096962e-07,
+      "loss": 0.7644,
+      "step": 9086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.216763311294147e-07,
+      "loss": 0.8617,
+      "step": 9087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.21120962887771e-07,
+      "loss": 0.7592,
+      "step": 9088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.205657242303478e-07,
+      "loss": 0.6458,
+      "step": 9089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.200106152027146e-07,
+      "loss": 0.7324,
+      "step": 9090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.194556358504314e-07,
+      "loss": 0.6867,
+      "step": 9091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.189007862190463e-07,
+      "loss": 0.6924,
+      "step": 9092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.183460663540997e-07,
+      "loss": 0.6517,
+      "step": 9093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.177914763011175e-07,
+      "loss": 0.6318,
+      "step": 9094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.172370161056188e-07,
+      "loss": 0.7846,
+      "step": 9095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.166826858131081e-07,
+      "loss": 0.7183,
+      "step": 9096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.161284854690832e-07,
+      "loss": 0.6833,
+      "step": 9097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.155744151190285e-07,
+      "loss": 0.7255,
+      "step": 9098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.150204748084174e-07,
+      "loss": 0.7073,
+      "step": 9099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.144666645827154e-07,
+      "loss": 0.7791,
+      "step": 9100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.139129844873754e-07,
+      "loss": 0.6583,
+      "step": 9101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.133594345678387e-07,
+      "loss": 0.6828,
+      "step": 9102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.128060148695386e-07,
+      "loss": 0.6018,
+      "step": 9103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.122527254378958e-07,
+      "loss": 0.6817,
+      "step": 9104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.116995663183192e-07,
+      "loss": 0.7196,
+      "step": 9105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.111465375562112e-07,
+      "loss": 0.6118,
+      "step": 9106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.105936391969591e-07,
+      "loss": 0.6318,
+      "step": 9107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.100408712859418e-07,
+      "loss": 0.7744,
+      "step": 9108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.094882338685257e-07,
+      "loss": 0.8758,
+      "step": 9109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.089357269900698e-07,
+      "loss": 0.8922,
+      "step": 9110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.083833506959191e-07,
+      "loss": 0.6481,
+      "step": 9111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.078311050314087e-07,
+      "loss": 0.625,
+      "step": 9112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.072789900418646e-07,
+      "loss": 0.7427,
+      "step": 9113
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.067270057725991e-07,
+      "loss": 0.7538,
+      "step": 9114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.061751522689177e-07,
+      "loss": 0.6231,
+      "step": 9115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.05623429576111e-07,
+      "loss": 0.7198,
+      "step": 9116
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.050718377394623e-07,
+      "loss": 0.8318,
+      "step": 9117
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.045203768042424e-07,
+      "loss": 0.7979,
+      "step": 9118
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.039690468157103e-07,
+      "loss": 0.6009,
+      "step": 9119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.034178478191174e-07,
+      "loss": 0.7083,
+      "step": 9120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.028667798597021e-07,
+      "loss": 0.7299,
+      "step": 9121
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.023158429826923e-07,
+      "loss": 0.6594,
+      "step": 9122
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.017650372333042e-07,
+      "loss": 0.731,
+      "step": 9123
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.012143626567465e-07,
+      "loss": 0.8334,
+      "step": 9124
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.00663819298214e-07,
+      "loss": 0.6414,
+      "step": 9125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.001134072028909e-07,
+      "loss": 0.6302,
+      "step": 9126
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.99563126415953e-07,
+      "loss": 0.7188,
+      "step": 9127
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.990129769825634e-07,
+      "loss": 0.6172,
+      "step": 9128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.984629589478744e-07,
+      "loss": 0.6786,
+      "step": 9129
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.979130723570272e-07,
+      "loss": 0.6219,
+      "step": 9130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.973633172551546e-07,
+      "loss": 0.8392,
+      "step": 9131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.968136936873755e-07,
+      "loss": 0.7268,
+      "step": 9132
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.962642016988013e-07,
+      "loss": 0.6695,
+      "step": 9133
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.957148413345285e-07,
+      "loss": 0.6018,
+      "step": 9134
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.951656126396469e-07,
+      "loss": 0.9217,
+      "step": 9135
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.801607072353363,
+      "eval_runtime": 437.7357,
+      "eval_samples_per_second": 6.851,
+      "eval_steps_per_second": 0.857,
+      "step": 9135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.946165156592332e-07,
+      "loss": 0.7252,
+      "step": 9136
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.940675504383522e-07,
+      "loss": 0.7103,
+      "step": 9137
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.935187170220622e-07,
+      "loss": 0.7934,
+      "step": 9138
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.929700154554058e-07,
+      "loss": 0.7233,
+      "step": 9139
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.924214457834171e-07,
+      "loss": 0.8479,
+      "step": 9140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.918730080511201e-07,
+      "loss": 0.6754,
+      "step": 9141
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.913247023035268e-07,
+      "loss": 0.7477,
+      "step": 9142
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.907765285856373e-07,
+      "loss": 0.6802,
+      "step": 9143
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.902284869424441e-07,
+      "loss": 0.6959,
+      "step": 9144
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.89680577418926e-07,
+      "loss": 0.7693,
+      "step": 9145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.891328000600518e-07,
+      "loss": 0.7403,
+      "step": 9146
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.885851549107788e-07,
+      "loss": 0.6053,
+      "step": 9147
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.880376420160561e-07,
+      "loss": 0.8073,
+      "step": 9148
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.874902614208186e-07,
+      "loss": 0.6801,
+      "step": 9149
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.869430131699914e-07,
+      "loss": 0.6337,
+      "step": 9150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.863958973084907e-07,
+      "loss": 0.709,
+      "step": 9151
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.858489138812185e-07,
+      "loss": 0.6053,
+      "step": 9152
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.853020629330695e-07,
+      "loss": 0.7206,
+      "step": 9153
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.847553445089239e-07,
+      "loss": 0.8573,
+      "step": 9154
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.842087586536543e-07,
+      "loss": 0.6413,
+      "step": 9155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.836623054121204e-07,
+      "loss": 0.7507,
+      "step": 9156
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.83115984829171e-07,
+      "loss": 0.6964,
+      "step": 9157
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.825697969496455e-07,
+      "loss": 0.7213,
+      "step": 9158
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.820237418183711e-07,
+      "loss": 0.6143,
+      "step": 9159
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.814778194801646e-07,
+      "loss": 0.7801,
+      "step": 9160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.809320299798305e-07,
+      "loss": 0.7771,
+      "step": 9161
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.80386373362166e-07,
+      "loss": 0.723,
+      "step": 9162
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.798408496719534e-07,
+      "loss": 0.7529,
+      "step": 9163
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.792954589539659e-07,
+      "loss": 0.6582,
+      "step": 9164
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.787502012529666e-07,
+      "loss": 0.7292,
+      "step": 9165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.782050766137063e-07,
+      "loss": 0.8342,
+      "step": 9166
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.77660085080925e-07,
+      "loss": 0.6519,
+      "step": 9167
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.771152266993516e-07,
+      "loss": 0.6768,
+      "step": 9168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.765705015137057e-07,
+      "loss": 0.8366,
+      "step": 9169
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.76025909568694e-07,
+      "loss": 0.6746,
+      "step": 9170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.754814509090132e-07,
+      "loss": 0.6472,
+      "step": 9171
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.749371255793501e-07,
+      "loss": 0.7176,
+      "step": 9172
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.743929336243786e-07,
+      "loss": 0.6286,
+      "step": 9173
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.738488750887622e-07,
+      "loss": 0.6351,
+      "step": 9174
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.733049500171534e-07,
+      "loss": 0.8049,
+      "step": 9175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.727611584541951e-07,
+      "loss": 0.6114,
+      "step": 9176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.72217500444518e-07,
+      "loss": 0.665,
+      "step": 9177
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.716739760327409e-07,
+      "loss": 0.69,
+      "step": 9178
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.711305852634741e-07,
+      "loss": 0.7419,
+      "step": 9179
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.705873281813154e-07,
+      "loss": 0.5967,
+      "step": 9180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.700442048308508e-07,
+      "loss": 0.7677,
+      "step": 9181
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.695012152566578e-07,
+      "loss": 0.6899,
+      "step": 9182
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.68958359503301e-07,
+      "loss": 0.6595,
+      "step": 9183
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.684156376153344e-07,
+      "loss": 0.7345,
+      "step": 9184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.678730496373e-07,
+      "loss": 0.7056,
+      "step": 9185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.673305956137321e-07,
+      "loss": 0.6846,
+      "step": 9186
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.667882755891504e-07,
+      "loss": 0.6405,
+      "step": 9187
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.662460896080646e-07,
+      "loss": 0.6452,
+      "step": 9188
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.657040377149747e-07,
+      "loss": 0.7248,
+      "step": 9189
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.651621199543694e-07,
+      "loss": 0.6671,
+      "step": 9190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.646203363707251e-07,
+      "loss": 0.6808,
+      "step": 9191
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.640786870085072e-07,
+      "loss": 0.6586,
+      "step": 9192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.635371719121725e-07,
+      "loss": 0.7774,
+      "step": 9193
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.629957911261639e-07,
+      "loss": 0.6665,
+      "step": 9194
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.624545446949137e-07,
+      "loss": 0.742,
+      "step": 9195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.619134326628459e-07,
+      "loss": 0.7337,
+      "step": 9196
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.613724550743705e-07,
+      "loss": 0.6058,
+      "step": 9197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.608316119738871e-07,
+      "loss": 0.6222,
+      "step": 9198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.602909034057843e-07,
+      "loss": 0.9873,
+      "step": 9199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.597503294144413e-07,
+      "loss": 0.7653,
+      "step": 9200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.592098900442245e-07,
+      "loss": 0.8689,
+      "step": 9201
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.586695853394883e-07,
+      "loss": 0.6375,
+      "step": 9202
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.581294153445793e-07,
+      "loss": 0.6979,
+      "step": 9203
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.575893801038306e-07,
+      "loss": 0.6197,
+      "step": 9204
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.570494796615647e-07,
+      "loss": 0.8259,
+      "step": 9205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.56509714062092e-07,
+      "loss": 0.8268,
+      "step": 9206
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.559700833497145e-07,
+      "loss": 0.6418,
+      "step": 9207
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.554305875687216e-07,
+      "loss": 0.6701,
+      "step": 9208
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.548912267633905e-07,
+      "loss": 0.6517,
+      "step": 9209
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.543520009779898e-07,
+      "loss": 0.7711,
+      "step": 9210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.538129102567755e-07,
+      "loss": 0.7409,
+      "step": 9211
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.53273954643992e-07,
+      "loss": 0.7025,
+      "step": 9212
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.527351341838724e-07,
+      "loss": 0.685,
+      "step": 9213
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.521964489206419e-07,
+      "loss": 0.7284,
+      "step": 9214
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.51657898898511e-07,
+      "loss": 0.6919,
+      "step": 9215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.511194841616802e-07,
+      "loss": 0.6764,
+      "step": 9216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.505812047543399e-07,
+      "loss": 0.6014,
+      "step": 9217
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.500430607206684e-07,
+      "loss": 0.6661,
+      "step": 9218
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.495050521048321e-07,
+      "loss": 0.6573,
+      "step": 9219
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.489671789509888e-07,
+      "loss": 0.7527,
+      "step": 9220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.484294413032829e-07,
+      "loss": 0.7978,
+      "step": 9221
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.478918392058486e-07,
+      "loss": 0.7392,
+      "step": 9222
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.473543727028077e-07,
+      "loss": 0.7677,
+      "step": 9223
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.46817041838274e-07,
+      "loss": 0.7215,
+      "step": 9224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.462798466563471e-07,
+      "loss": 0.6877,
+      "step": 9225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.457427872011159e-07,
+      "loss": 0.8174,
+      "step": 9226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.452058635166594e-07,
+      "loss": 0.6792,
+      "step": 9227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.446690756470455e-07,
+      "loss": 0.641,
+      "step": 9228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.441324236363299e-07,
+      "loss": 0.6185,
+      "step": 9229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.435959075285563e-07,
+      "loss": 0.8055,
+      "step": 9230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.430595273677605e-07,
+      "loss": 0.7516,
+      "step": 9231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.42523283197964e-07,
+      "loss": 0.6398,
+      "step": 9232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.419871750631775e-07,
+      "loss": 0.6597,
+      "step": 9233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.414512030074032e-07,
+      "loss": 0.6137,
+      "step": 9234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.409153670746292e-07,
+      "loss": 0.7916,
+      "step": 9235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.403796673088335e-07,
+      "loss": 0.6678,
+      "step": 9236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.39844103753982e-07,
+      "loss": 0.8624,
+      "step": 9237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.39308676454032e-07,
+      "loss": 0.729,
+      "step": 9238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.387733854529273e-07,
+      "loss": 0.7521,
+      "step": 9239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.382382307946e-07,
+      "loss": 0.7833,
+      "step": 9240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.37703212522974e-07,
+      "loss": 0.6476,
+      "step": 9241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.371683306819589e-07,
+      "loss": 0.7915,
+      "step": 9242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.36633585315455e-07,
+      "loss": 0.7995,
+      "step": 9243
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.360989764673497e-07,
+      "loss": 0.6306,
+      "step": 9244
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.355645041815208e-07,
+      "loss": 0.7511,
+      "step": 9245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.350301685018356e-07,
+      "loss": 0.7982,
+      "step": 9246
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.344959694721469e-07,
+      "loss": 0.6838,
+      "step": 9247
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.339619071363001e-07,
+      "loss": 0.6808,
+      "step": 9248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.334279815381269e-07,
+      "loss": 0.5869,
+      "step": 9249
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.328941927214481e-07,
+      "loss": 0.6922,
+      "step": 9250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.323605407300733e-07,
+      "loss": 0.7803,
+      "step": 9251
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.318270256078026e-07,
+      "loss": 0.6441,
+      "step": 9252
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.312936473984229e-07,
+      "loss": 0.6934,
+      "step": 9253
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.307604061457091e-07,
+      "loss": 0.7619,
+      "step": 9254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.302273018934284e-07,
+      "loss": 0.6976,
+      "step": 9255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.296943346853334e-07,
+      "loss": 0.7145,
+      "step": 9256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.291615045651661e-07,
+      "loss": 0.8377,
+      "step": 9257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.286288115766592e-07,
+      "loss": 0.6826,
+      "step": 9258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.280962557635319e-07,
+      "loss": 0.8219,
+      "step": 9259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.275638371694927e-07,
+      "loss": 0.8088,
+      "step": 9260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.270315558382388e-07,
+      "loss": 0.6131,
+      "step": 9261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.264994118134579e-07,
+      "loss": 0.6656,
+      "step": 9262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.25967405138823e-07,
+      "loss": 0.7035,
+      "step": 9263
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.254355358579996e-07,
+      "loss": 0.6946,
+      "step": 9264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.249038040146387e-07,
+      "loss": 0.7535,
+      "step": 9265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.243722096523826e-07,
+      "loss": 0.7058,
+      "step": 9266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.238407528148609e-07,
+      "loss": 0.8286,
+      "step": 9267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.233094335456906e-07,
+      "loss": 0.6715,
+      "step": 9268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.227782518884814e-07,
+      "loss": 0.7741,
+      "step": 9269
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.222472078868279e-07,
+      "loss": 0.8318,
+      "step": 9270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.217163015843141e-07,
+      "loss": 0.7341,
+      "step": 9271
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.211855330245149e-07,
+      "loss": 0.6062,
+      "step": 9272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.206549022509922e-07,
+      "loss": 0.6445,
+      "step": 9273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.201244093072957e-07,
+      "loss": 0.6669,
+      "step": 9274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.195940542369645e-07,
+      "loss": 0.7734,
+      "step": 9275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.190638370835288e-07,
+      "loss": 0.7456,
+      "step": 9276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.18533757890504e-07,
+      "loss": 0.724,
+      "step": 9277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.180038167013949e-07,
+      "loss": 0.7346,
+      "step": 9278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.174740135596975e-07,
+      "loss": 0.7319,
+      "step": 9279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.169443485088935e-07,
+      "loss": 0.8769,
+      "step": 9280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.164148215924539e-07,
+      "loss": 0.5919,
+      "step": 9281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.158854328538404e-07,
+      "loss": 0.6649,
+      "step": 9282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.153561823365e-07,
+      "loss": 0.7025,
+      "step": 9283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.14827070083872e-07,
+      "loss": 0.5986,
+      "step": 9284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.142980961393807e-07,
+      "loss": 0.824,
+      "step": 9285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.137692605464428e-07,
+      "loss": 0.6098,
+      "step": 9286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.132405633484605e-07,
+      "loss": 0.7549,
+      "step": 9287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.127120045888265e-07,
+      "loss": 0.8203,
+      "step": 9288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.1218358431092e-07,
+      "loss": 0.6669,
+      "step": 9289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.116553025581123e-07,
+      "loss": 0.6993,
+      "step": 9290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.111271593737602e-07,
+      "loss": 0.6626,
+      "step": 9291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.105991548012102e-07,
+      "loss": 0.7294,
+      "step": 9292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.100712888837983e-07,
+      "loss": 0.8726,
+      "step": 9293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.095435616648484e-07,
+      "loss": 0.7262,
+      "step": 9294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.090159731876712e-07,
+      "loss": 0.6815,
+      "step": 9295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.084885234955702e-07,
+      "loss": 0.7157,
+      "step": 9296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.079612126318337e-07,
+      "loss": 0.6904,
+      "step": 9297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.074340406397402e-07,
+      "loss": 0.7796,
+      "step": 9298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.069070075625559e-07,
+      "loss": 0.7625,
+      "step": 9299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.063801134435376e-07,
+      "loss": 0.7772,
+      "step": 9300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.058533583259281e-07,
+      "loss": 0.8248,
+      "step": 9301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.053267422529615e-07,
+      "loss": 0.8062,
+      "step": 9302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.048002652678571e-07,
+      "loss": 0.6272,
+      "step": 9303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.04273927413827e-07,
+      "loss": 0.5609,
+      "step": 9304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.037477287340683e-07,
+      "loss": 0.7181,
+      "step": 9305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.032216692717676e-07,
+      "loss": 0.6633,
+      "step": 9306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.026957490701018e-07,
+      "loss": 0.7018,
+      "step": 9307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.021699681722344e-07,
+      "loss": 0.781,
+      "step": 9308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.016443266213172e-07,
+      "loss": 0.7677,
+      "step": 9309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.011188244604931e-07,
+      "loss": 0.7624,
+      "step": 9310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.005934617328911e-07,
+      "loss": 0.7223,
+      "step": 9311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.000682384816294e-07,
+      "loss": 0.8904,
+      "step": 9312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.995431547498147e-07,
+      "loss": 0.6611,
+      "step": 9313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.990182105805436e-07,
+      "loss": 0.6702,
+      "step": 9314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.984934060168994e-07,
+      "loss": 0.7328,
+      "step": 9315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.97968741101954e-07,
+      "loss": 0.8518,
+      "step": 9316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.974442158787702e-07,
+      "loss": 0.7044,
+      "step": 9317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.969198303903966e-07,
+      "loss": 0.7452,
+      "step": 9318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.96395584679871e-07,
+      "loss": 0.8681,
+      "step": 9319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.958714787902211e-07,
+      "loss": 0.7088,
+      "step": 9320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.953475127644608e-07,
+      "loss": 0.5968,
+      "step": 9321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.948236866455958e-07,
+      "loss": 0.614,
+      "step": 9322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.943000004766163e-07,
+      "loss": 0.659,
+      "step": 9323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.93776454300505e-07,
+      "loss": 0.7179,
+      "step": 9324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.932530481602302e-07,
+      "loss": 0.5749,
+      "step": 9325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.927297820987498e-07,
+      "loss": 0.7328,
+      "step": 9326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.922066561590092e-07,
+      "loss": 0.629,
+      "step": 9327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.916836703839448e-07,
+      "loss": 0.7201,
+      "step": 9328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.911608248164792e-07,
+      "loss": 0.7285,
+      "step": 9329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.906381194995238e-07,
+      "loss": 0.6323,
+      "step": 9330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.901155544759797e-07,
+      "loss": 0.7035,
+      "step": 9331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.895931297887352e-07,
+      "loss": 0.7326,
+      "step": 9332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.89070845480667e-07,
+      "loss": 0.7174,
+      "step": 9333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.885487015946422e-07,
+      "loss": 0.7251,
+      "step": 9334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.880266981735144e-07,
+      "loss": 0.6375,
+      "step": 9335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.875048352601259e-07,
+      "loss": 0.6785,
+      "step": 9336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.869831128973071e-07,
+      "loss": 0.619,
+      "step": 9337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.864615311278794e-07,
+      "loss": 0.7139,
+      "step": 9338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.859400899946495e-07,
+      "loss": 0.6372,
+      "step": 9339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.85418789540415e-07,
+      "loss": 0.709,
+      "step": 9340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.848976298079597e-07,
+      "loss": 0.8067,
+      "step": 9341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.843766108400586e-07,
+      "loss": 0.6791,
+      "step": 9342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.838557326794724e-07,
+      "loss": 0.7192,
+      "step": 9343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.833349953689512e-07,
+      "loss": 0.7089,
+      "step": 9344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.828143989512349e-07,
+      "loss": 0.6869,
+      "step": 9345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.8229394346905e-07,
+      "loss": 0.6611,
+      "step": 9346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.817736289651113e-07,
+      "loss": 0.6783,
+      "step": 9347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.812534554821247e-07,
+      "loss": 0.6574,
+      "step": 9348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.807334230627817e-07,
+      "loss": 0.674,
+      "step": 9349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.802135317497633e-07,
+      "loss": 0.7173,
+      "step": 9350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.796937815857378e-07,
+      "loss": 0.6741,
+      "step": 9351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.791741726133647e-07,
+      "loss": 0.7885,
+      "step": 9352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.786547048752896e-07,
+      "loss": 0.6305,
+      "step": 9353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.78135378414146e-07,
+      "loss": 0.7993,
+      "step": 9354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.776161932725584e-07,
+      "loss": 0.7695,
+      "step": 9355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.770971494931376e-07,
+      "loss": 0.6419,
+      "step": 9356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.765782471184824e-07,
+      "loss": 0.6234,
+      "step": 9357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.760594861911827e-07,
+      "loss": 0.6685,
+      "step": 9358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.755408667538133e-07,
+      "loss": 0.6508,
+      "step": 9359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.750223888489408e-07,
+      "loss": 0.6659,
+      "step": 9360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.745040525191172e-07,
+      "loss": 0.6561,
+      "step": 9361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.739858578068851e-07,
+      "loss": 0.7235,
+      "step": 9362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.734678047547744e-07,
+      "loss": 0.7998,
+      "step": 9363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.729498934053037e-07,
+      "loss": 0.7315,
+      "step": 9364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.724321238009782e-07,
+      "loss": 0.6615,
+      "step": 9365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.719144959842956e-07,
+      "loss": 0.5813,
+      "step": 9366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.713970099977383e-07,
+      "loss": 0.7044,
+      "step": 9367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.708796658837773e-07,
+      "loss": 0.6938,
+      "step": 9368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.703624636848742e-07,
+      "loss": 0.5559,
+      "step": 9369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.698454034434777e-07,
+      "loss": 0.8097,
+      "step": 9370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.693284852020231e-07,
+      "loss": 0.5943,
+      "step": 9371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.688117090029376e-07,
+      "loss": 0.6464,
+      "step": 9372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.682950748886345e-07,
+      "loss": 0.6814,
+      "step": 9373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.677785829015153e-07,
+      "loss": 0.6537,
+      "step": 9374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.672622330839696e-07,
+      "loss": 0.8288,
+      "step": 9375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.667460254783776e-07,
+      "loss": 0.55,
+      "step": 9376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.66229960127105e-07,
+      "loss": 0.6238,
+      "step": 9377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.657140370725086e-07,
+      "loss": 0.7083,
+      "step": 9378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.651982563569302e-07,
+      "loss": 0.6486,
+      "step": 9379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.646826180227038e-07,
+      "loss": 0.6573,
+      "step": 9380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.641671221121485e-07,
+      "loss": 0.878,
+      "step": 9381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.63651768667572e-07,
+      "loss": 0.7517,
+      "step": 9382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.631365577312733e-07,
+      "loss": 0.5937,
+      "step": 9383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.626214893455363e-07,
+      "loss": 0.8562,
+      "step": 9384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.62106563552634e-07,
+      "loss": 0.6967,
+      "step": 9385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.615917803948298e-07,
+      "loss": 0.808,
+      "step": 9386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.610771399143727e-07,
+      "loss": 0.6913,
+      "step": 9387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.605626421535012e-07,
+      "loss": 0.672,
+      "step": 9388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.600482871544413e-07,
+      "loss": 0.693,
+      "step": 9389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.595340749594094e-07,
+      "loss": 0.7218,
+      "step": 9390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.590200056106078e-07,
+      "loss": 0.7247,
+      "step": 9391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.585060791502275e-07,
+      "loss": 0.7426,
+      "step": 9392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.579922956204497e-07,
+      "loss": 0.5862,
+      "step": 9393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.574786550634416e-07,
+      "loss": 0.6834,
+      "step": 9394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.569651575213591e-07,
+      "loss": 0.7172,
+      "step": 9395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.564518030363477e-07,
+      "loss": 0.6923,
+      "step": 9396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.559385916505394e-07,
+      "loss": 0.6109,
+      "step": 9397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.554255234060562e-07,
+      "loss": 0.759,
+      "step": 9398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.549125983450059e-07,
+      "loss": 0.7972,
+      "step": 9399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.543998165094882e-07,
+      "loss": 0.771,
+      "step": 9400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.538871779415876e-07,
+      "loss": 0.6219,
+      "step": 9401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.533746826833788e-07,
+      "loss": 0.7831,
+      "step": 9402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.528623307769225e-07,
+      "loss": 0.6764,
+      "step": 9403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.523501222642715e-07,
+      "loss": 0.6503,
+      "step": 9404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.518380571874634e-07,
+      "loss": 0.7796,
+      "step": 9405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.513261355885246e-07,
+      "loss": 0.6026,
+      "step": 9406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.508143575094717e-07,
+      "loss": 0.7589,
+      "step": 9407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.503027229923076e-07,
+      "loss": 0.7401,
+      "step": 9408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.497912320790232e-07,
+      "loss": 0.6677,
+      "step": 9409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.492798848115995e-07,
+      "loss": 0.7273,
+      "step": 9410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.487686812320044e-07,
+      "loss": 0.6304,
+      "step": 9411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.482576213821938e-07,
+      "loss": 0.7399,
+      "step": 9412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.477467053041118e-07,
+      "loss": 0.7316,
+      "step": 9413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.472359330396922e-07,
+      "loss": 0.681,
+      "step": 9414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.467253046308546e-07,
+      "loss": 0.6198,
+      "step": 9415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.462148201195094e-07,
+      "loss": 0.6881,
+      "step": 9416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.457044795475527e-07,
+      "loss": 0.761,
+      "step": 9417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.451942829568712e-07,
+      "loss": 0.6923,
+      "step": 9418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.446842303893381e-07,
+      "loss": 0.7851,
+      "step": 9419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.441743218868142e-07,
+      "loss": 0.6545,
+      "step": 9420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.43664557491151e-07,
+      "loss": 0.6267,
+      "step": 9421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.431549372441862e-07,
+      "loss": 0.7065,
+      "step": 9422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.426454611877451e-07,
+      "loss": 0.7123,
+      "step": 9423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.421361293636439e-07,
+      "loss": 0.6785,
+      "step": 9424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.416269418136845e-07,
+      "loss": 0.7167,
+      "step": 9425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.411178985796577e-07,
+      "loss": 0.747,
+      "step": 9426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.406089997033416e-07,
+      "loss": 0.7432,
+      "step": 9427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.401002452265049e-07,
+      "loss": 0.6718,
+      "step": 9428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.395916351909024e-07,
+      "loss": 0.8733,
+      "step": 9429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.390831696382767e-07,
+      "loss": 0.6018,
+      "step": 9430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.385748486103605e-07,
+      "loss": 0.7632,
+      "step": 9431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.380666721488733e-07,
+      "loss": 0.6911,
+      "step": 9432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.375586402955218e-07,
+      "loss": 0.7658,
+      "step": 9433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.370507530920034e-07,
+      "loss": 0.5916,
+      "step": 9434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.365430105800012e-07,
+      "loss": 0.6795,
+      "step": 9435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.360354128011885e-07,
+      "loss": 0.6033,
+      "step": 9436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.355279597972245e-07,
+      "loss": 0.6977,
+      "step": 9437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.350206516097588e-07,
+      "loss": 0.6649,
+      "step": 9438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.345134882804278e-07,
+      "loss": 0.7623,
+      "step": 9439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.340064698508553e-07,
+      "loss": 0.6983,
+      "step": 9440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.334995963626543e-07,
+      "loss": 0.7866,
+      "step": 9441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.329928678574266e-07,
+      "loss": 0.6739,
+      "step": 9442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.324862843767608e-07,
+      "loss": 0.6058,
+      "step": 9443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.319798459622329e-07,
+      "loss": 0.6113,
+      "step": 9444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.3147355265541e-07,
+      "loss": 0.7342,
+      "step": 9445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.309674044978446e-07,
+      "loss": 0.6022,
+      "step": 9446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.304614015310768e-07,
+      "loss": 0.8171,
+      "step": 9447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.299555437966383e-07,
+      "loss": 0.685,
+      "step": 9448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.294498313360454e-07,
+      "loss": 0.7754,
+      "step": 9449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.289442641908037e-07,
+      "loss": 0.7782,
+      "step": 9450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.284388424024067e-07,
+      "loss": 0.736,
+      "step": 9451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.279335660123369e-07,
+      "loss": 0.691,
+      "step": 9452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.274284350620633e-07,
+      "loss": 0.6161,
+      "step": 9453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.269234495930446e-07,
+      "loss": 0.7375,
+      "step": 9454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.26418609646726e-07,
+      "loss": 0.6751,
+      "step": 9455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.259139152645428e-07,
+      "loss": 0.731,
+      "step": 9456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.254093664879158e-07,
+      "loss": 0.6492,
+      "step": 9457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.24904963358255e-07,
+      "loss": 0.707,
+      "step": 9458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.244007059169597e-07,
+      "loss": 0.783,
+      "step": 9459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.238965942054154e-07,
+      "loss": 0.7346,
+      "step": 9460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.233926282649958e-07,
+      "loss": 0.6575,
+      "step": 9461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.228888081370647e-07,
+      "loss": 0.6229,
+      "step": 9462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.223851338629712e-07,
+      "loss": 0.6815,
+      "step": 9463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.218816054840538e-07,
+      "loss": 0.7838,
+      "step": 9464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.213782230416385e-07,
+      "loss": 0.6974,
+      "step": 9465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.208749865770409e-07,
+      "loss": 0.7489,
+      "step": 9466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.203718961315626e-07,
+      "loss": 0.629,
+      "step": 9467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.198689517464935e-07,
+      "loss": 0.6991,
+      "step": 9468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.193661534631132e-07,
+      "loss": 0.7347,
+      "step": 9469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.188635013226878e-07,
+      "loss": 0.5653,
+      "step": 9470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.183609953664705e-07,
+      "loss": 0.6302,
+      "step": 9471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.178586356357057e-07,
+      "loss": 0.6179,
+      "step": 9472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.173564221716222e-07,
+      "loss": 0.8445,
+      "step": 9473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.1685435501544e-07,
+      "loss": 0.6851,
+      "step": 9474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.163524342083638e-07,
+      "loss": 0.7427,
+      "step": 9475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.158506597915898e-07,
+      "loss": 0.7582,
+      "step": 9476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.153490318062995e-07,
+      "loss": 0.6379,
+      "step": 9477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.148475502936636e-07,
+      "loss": 0.6716,
+      "step": 9478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.143462152948391e-07,
+      "loss": 0.7573,
+      "step": 9479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.138450268509744e-07,
+      "loss": 0.588,
+      "step": 9480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.133439850032029e-07,
+      "loss": 0.704,
+      "step": 9481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.128430897926461e-07,
+      "loss": 0.6573,
+      "step": 9482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.123423412604158e-07,
+      "loss": 0.7004,
+      "step": 9483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.118417394476096e-07,
+      "loss": 0.6466,
+      "step": 9484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.113412843953124e-07,
+      "loss": 0.7135,
+      "step": 9485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.108409761446006e-07,
+      "loss": 0.5904,
+      "step": 9486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.103408147365348e-07,
+      "loss": 0.6346,
+      "step": 9487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.098408002121654e-07,
+      "loss": 0.5442,
+      "step": 9488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.093409326125297e-07,
+      "loss": 0.6786,
+      "step": 9489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.08841211978655e-07,
+      "loss": 0.6212,
+      "step": 9490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.083416383515534e-07,
+      "loss": 0.9039,
+      "step": 9491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.078422117722287e-07,
+      "loss": 0.7691,
+      "step": 9492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.073429322816686e-07,
+      "loss": 0.7069,
+      "step": 9493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.068437999208524e-07,
+      "loss": 0.73,
+      "step": 9494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.063448147307453e-07,
+      "loss": 0.7114,
+      "step": 9495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.058459767522996e-07,
+      "loss": 0.7861,
+      "step": 9496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.053472860264582e-07,
+      "loss": 0.7515,
+      "step": 9497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.0484874259415e-07,
+      "loss": 0.7926,
+      "step": 9498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.043503464962912e-07,
+      "loss": 0.797,
+      "step": 9499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.038520977737887e-07,
+      "loss": 0.6152,
+      "step": 9500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.033539964675342e-07,
+      "loss": 0.7505,
+      "step": 9501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.028560426184094e-07,
+      "loss": 0.8564,
+      "step": 9502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.023582362672824e-07,
+      "loss": 0.7756,
+      "step": 9503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.018605774550105e-07,
+      "loss": 0.6757,
+      "step": 9504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.013630662224385e-07,
+      "loss": 0.8593,
+      "step": 9505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.008657026103979e-07,
+      "loss": 0.7242,
+      "step": 9506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.003684866597105e-07,
+      "loss": 0.6367,
+      "step": 9507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.998714184111838e-07,
+      "loss": 0.66,
+      "step": 9508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.993744979056133e-07,
+      "loss": 0.7189,
+      "step": 9509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.988777251837844e-07,
+      "loss": 0.7402,
+      "step": 9510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.983811002864679e-07,
+      "loss": 0.7533,
+      "step": 9511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.978846232544248e-07,
+      "loss": 0.6823,
+      "step": 9512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.973882941284011e-07,
+      "loss": 0.5854,
+      "step": 9513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.96892112949134e-07,
+      "loss": 0.7027,
+      "step": 9514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.96396079757346e-07,
+      "loss": 0.7472,
+      "step": 9515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.959001945937485e-07,
+      "loss": 0.7277,
+      "step": 9516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.954044574990395e-07,
+      "loss": 0.638,
+      "step": 9517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.949088685139077e-07,
+      "loss": 0.6899,
+      "step": 9518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.944134276790271e-07,
+      "loss": 0.6491,
+      "step": 9519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.939181350350591e-07,
+      "loss": 0.7284,
+      "step": 9520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.934229906226561e-07,
+      "loss": 0.7106,
+      "step": 9521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.929279944824554e-07,
+      "loss": 0.7294,
+      "step": 9522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.924331466550827e-07,
+      "loss": 0.7139,
+      "step": 9523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.919384471811527e-07,
+      "loss": 0.7711,
+      "step": 9524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.914438961012671e-07,
+      "loss": 0.6309,
+      "step": 9525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.909494934560151e-07,
+      "loss": 0.7623,
+      "step": 9526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.904552392859734e-07,
+      "loss": 0.6023,
+      "step": 9527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.899611336317086e-07,
+      "loss": 0.7097,
+      "step": 9528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.894671765337724e-07,
+      "loss": 0.616,
+      "step": 9529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.889733680327071e-07,
+      "loss": 0.7589,
+      "step": 9530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.884797081690398e-07,
+      "loss": 0.6875,
+      "step": 9531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.879861969832885e-07,
+      "loss": 0.741,
+      "step": 9532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.874928345159562e-07,
+      "loss": 0.6882,
+      "step": 9533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.86999620807535e-07,
+      "loss": 0.8051,
+      "step": 9534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.865065558985054e-07,
+      "loss": 0.7703,
+      "step": 9535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.860136398293346e-07,
+      "loss": 0.6655,
+      "step": 9536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.855208726404771e-07,
+      "loss": 0.6756,
+      "step": 9537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.850282543723777e-07,
+      "loss": 0.6326,
+      "step": 9538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.845357850654666e-07,
+      "loss": 0.7734,
+      "step": 9539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.840434647601624e-07,
+      "loss": 0.7841,
+      "step": 9540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.835512934968707e-07,
+      "loss": 0.7308,
+      "step": 9541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.830592713159878e-07,
+      "loss": 0.7007,
+      "step": 9542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.825673982578943e-07,
+      "loss": 0.716,
+      "step": 9543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.820756743629595e-07,
+      "loss": 0.6543,
+      "step": 9544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.815840996715425e-07,
+      "loss": 0.742,
+      "step": 9545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.810926742239879e-07,
+      "loss": 0.6667,
+      "step": 9546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.806013980606274e-07,
+      "loss": 0.6476,
+      "step": 9547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.801102712217841e-07,
+      "loss": 0.5871,
+      "step": 9548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.796192937477647e-07,
+      "loss": 0.6323,
+      "step": 9549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.791284656788666e-07,
+      "loss": 0.6707,
+      "step": 9550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.786377870553729e-07,
+      "loss": 0.7488,
+      "step": 9551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.781472579175563e-07,
+      "loss": 0.705,
+      "step": 9552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.776568783056761e-07,
+      "loss": 0.6991,
+      "step": 9553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.77166648259979e-07,
+      "loss": 0.7899,
+      "step": 9554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.766765678206994e-07,
+      "loss": 0.6392,
+      "step": 9555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.761866370280615e-07,
+      "loss": 0.705,
+      "step": 9556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.756968559222745e-07,
+      "loss": 0.7216,
+      "step": 9557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.752072245435364e-07,
+      "loss": 0.6975,
+      "step": 9558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.747177429320339e-07,
+      "loss": 0.6863,
+      "step": 9559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.742284111279402e-07,
+      "loss": 0.7202,
+      "step": 9560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.737392291714153e-07,
+      "loss": 0.6865,
+      "step": 9561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.7325019710261e-07,
+      "loss": 0.8129,
+      "step": 9562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.727613149616602e-07,
+      "loss": 0.5441,
+      "step": 9563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.722725827886897e-07,
+      "loss": 0.8773,
+      "step": 9564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.7178400062381e-07,
+      "loss": 0.6569,
+      "step": 9565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.712955685071224e-07,
+      "loss": 0.8081,
+      "step": 9566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.708072864787127e-07,
+      "loss": 0.7708,
+      "step": 9567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.703191545786572e-07,
+      "loss": 0.5996,
+      "step": 9568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.698311728470175e-07,
+      "loss": 0.6996,
+      "step": 9569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.693433413238453e-07,
+      "loss": 0.6853,
+      "step": 9570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.68855660049178e-07,
+      "loss": 0.6567,
+      "step": 9571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.683681290630406e-07,
+      "loss": 0.7021,
+      "step": 9572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.678807484054478e-07,
+      "loss": 0.6851,
+      "step": 9573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.673935181163998e-07,
+      "loss": 0.6922,
+      "step": 9574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.669064382358853e-07,
+      "loss": 0.6505,
+      "step": 9575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.664195088038813e-07,
+      "loss": 0.6324,
+      "step": 9576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.659327298603519e-07,
+      "loss": 0.6633,
+      "step": 9577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.654461014452479e-07,
+      "loss": 0.6501,
+      "step": 9578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.649596235985087e-07,
+      "loss": 0.6904,
+      "step": 9579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.644732963600622e-07,
+      "loss": 0.6941,
+      "step": 9580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.639871197698225e-07,
+      "loss": 0.5834,
+      "step": 9581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.635010938676912e-07,
+      "loss": 0.6805,
+      "step": 9582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.630152186935592e-07,
+      "loss": 0.645,
+      "step": 9583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.62529494287304e-07,
+      "loss": 0.6777,
+      "step": 9584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.620439206887893e-07,
+      "loss": 0.751,
+      "step": 9585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.615584979378695e-07,
+      "loss": 0.7433,
+      "step": 9586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.610732260743838e-07,
+      "loss": 0.6447,
+      "step": 9587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.605881051381616e-07,
+      "loss": 0.6641,
+      "step": 9588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.601031351690165e-07,
+      "loss": 0.6264,
+      "step": 9589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.596183162067538e-07,
+      "loss": 0.6528,
+      "step": 9590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.591336482911634e-07,
+      "loss": 0.7191,
+      "step": 9591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.586491314620236e-07,
+      "loss": 0.8771,
+      "step": 9592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.581647657590998e-07,
+      "loss": 0.7467,
+      "step": 9593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.57680551222147e-07,
+      "loss": 0.71,
+      "step": 9594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.571964878909057e-07,
+      "loss": 0.5528,
+      "step": 9595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.567125758051041e-07,
+      "loss": 0.6659,
+      "step": 9596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.562288150044599e-07,
+      "loss": 0.6415,
+      "step": 9597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.557452055286761e-07,
+      "loss": 0.8872,
+      "step": 9598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.552617474174441e-07,
+      "loss": 0.7597,
+      "step": 9599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.54778440710444e-07,
+      "loss": 0.6862,
+      "step": 9600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.542952854473422e-07,
+      "loss": 0.71,
+      "step": 9601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.538122816677925e-07,
+      "loss": 0.7149,
+      "step": 9602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.533294294114365e-07,
+      "loss": 0.6691,
+      "step": 9603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.528467287179047e-07,
+      "loss": 0.6394,
+      "step": 9604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.523641796268124e-07,
+      "loss": 0.6614,
+      "step": 9605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.518817821777662e-07,
+      "loss": 0.7468,
+      "step": 9606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.513995364103562e-07,
+      "loss": 0.6891,
+      "step": 9607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.50917442364164e-07,
+      "loss": 0.6666,
+      "step": 9608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.504355000787554e-07,
+      "loss": 0.6887,
+      "step": 9609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.499537095936848e-07,
+      "loss": 0.7651,
+      "step": 9610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.494720709484958e-07,
+      "loss": 0.6689,
+      "step": 9611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.489905841827177e-07,
+      "loss": 0.744,
+      "step": 9612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.485092493358666e-07,
+      "loss": 0.7358,
+      "step": 9613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.480280664474492e-07,
+      "loss": 0.6658,
+      "step": 9614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.475470355569572e-07,
+      "loss": 0.7557,
+      "step": 9615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.470661567038705e-07,
+      "loss": 0.647,
+      "step": 9616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.465854299276555e-07,
+      "loss": 0.6566,
+      "step": 9617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.46104855267769e-07,
+      "loss": 0.7159,
+      "step": 9618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.456244327636524e-07,
+      "loss": 0.6103,
+      "step": 9619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.451441624547352e-07,
+      "loss": 0.697,
+      "step": 9620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.446640443804361e-07,
+      "loss": 0.7315,
+      "step": 9621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.441840785801598e-07,
+      "loss": 0.6478,
+      "step": 9622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.437042650932979e-07,
+      "loss": 0.7865,
+      "step": 9623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.432246039592316e-07,
+      "loss": 0.688,
+      "step": 9624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.42745095217327e-07,
+      "loss": 0.6965,
+      "step": 9625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.422657389069409e-07,
+      "loss": 0.5249,
+      "step": 9626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.417865350674138e-07,
+      "loss": 0.6237,
+      "step": 9627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.413074837380776e-07,
+      "loss": 0.6854,
+      "step": 9628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.408285849582488e-07,
+      "loss": 0.7214,
+      "step": 9629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.403498387672322e-07,
+      "loss": 0.7015,
+      "step": 9630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.398712452043199e-07,
+      "loss": 0.7188,
+      "step": 9631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.393928043087927e-07,
+      "loss": 0.6027,
+      "step": 9632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.389145161199176e-07,
+      "loss": 0.7842,
+      "step": 9633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.384363806769487e-07,
+      "loss": 0.7406,
+      "step": 9634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.379583980191292e-07,
+      "loss": 0.7708,
+      "step": 9635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.374805681856888e-07,
+      "loss": 0.6937,
+      "step": 9636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.370028912158435e-07,
+      "loss": 0.6506,
+      "step": 9637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.365253671487997e-07,
+      "loss": 0.7076,
+      "step": 9638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.360479960237484e-07,
+      "loss": 0.5871,
+      "step": 9639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.355707778798697e-07,
+      "loss": 0.7666,
+      "step": 9640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.350937127563295e-07,
+      "loss": 0.6782,
+      "step": 9641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.346168006922834e-07,
+      "loss": 0.6688,
+      "step": 9642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.341400417268725e-07,
+      "loss": 0.6572,
+      "step": 9643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.336634358992264e-07,
+      "loss": 0.7127,
+      "step": 9644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.331869832484628e-07,
+      "loss": 0.6231,
+      "step": 9645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.327106838136845e-07,
+      "loss": 0.6316,
+      "step": 9646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.322345376339841e-07,
+      "loss": 0.6364,
+      "step": 9647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.317585447484393e-07,
+      "loss": 0.6122,
+      "step": 9648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.312827051961176e-07,
+      "loss": 0.6313,
+      "step": 9649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.308070190160729e-07,
+      "loss": 0.7556,
+      "step": 9650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.303314862473453e-07,
+      "loss": 0.6519,
+      "step": 9651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.298561069289652e-07,
+      "loss": 0.6299,
+      "step": 9652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.293808810999474e-07,
+      "loss": 0.6109,
+      "step": 9653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.289058087992961e-07,
+      "loss": 0.8888,
+      "step": 9654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.284308900660007e-07,
+      "loss": 0.9068,
+      "step": 9655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.279561249390417e-07,
+      "loss": 0.7218,
+      "step": 9656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.274815134573834e-07,
+      "loss": 0.8079,
+      "step": 9657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.270070556599783e-07,
+      "loss": 0.7986,
+      "step": 9658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.265327515857686e-07,
+      "loss": 0.6632,
+      "step": 9659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.260586012736808e-07,
+      "loss": 0.6672,
+      "step": 9660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.255846047626302e-07,
+      "loss": 0.7068,
+      "step": 9661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.251107620915192e-07,
+      "loss": 0.6016,
+      "step": 9662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.246370732992391e-07,
+      "loss": 0.731,
+      "step": 9663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.241635384246664e-07,
+      "loss": 0.8287,
+      "step": 9664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.236901575066648e-07,
+      "loss": 0.6827,
+      "step": 9665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.232169305840879e-07,
+      "loss": 0.7568,
+      "step": 9666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.227438576957748e-07,
+      "loss": 0.7506,
+      "step": 9667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.222709388805517e-07,
+      "loss": 0.753,
+      "step": 9668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.217981741772322e-07,
+      "loss": 0.6925,
+      "step": 9669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.213255636246193e-07,
+      "loss": 0.6831,
+      "step": 9670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.208531072615012e-07,
+      "loss": 0.7287,
+      "step": 9671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.203808051266533e-07,
+      "loss": 0.6618,
+      "step": 9672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.199086572588406e-07,
+      "loss": 0.6031,
+      "step": 9673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.19436663696813e-07,
+      "loss": 0.6463,
+      "step": 9674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.189648244793081e-07,
+      "loss": 0.6231,
+      "step": 9675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.184931396450527e-07,
+      "loss": 0.7008,
+      "step": 9676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.180216092327592e-07,
+      "loss": 0.7755,
+      "step": 9677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.175502332811278e-07,
+      "loss": 0.8221,
+      "step": 9678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.170790118288453e-07,
+      "loss": 0.6793,
+      "step": 9679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.166079449145876e-07,
+      "loss": 0.8368,
+      "step": 9680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.161370325770156e-07,
+      "loss": 0.8385,
+      "step": 9681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.156662748547795e-07,
+      "loss": 0.7713,
+      "step": 9682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.151956717865168e-07,
+      "loss": 0.6932,
+      "step": 9683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.147252234108508e-07,
+      "loss": 0.715,
+      "step": 9684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.14254929766393e-07,
+      "loss": 0.581,
+      "step": 9685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.137847908917413e-07,
+      "loss": 0.6651,
+      "step": 9686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.13314806825483e-07,
+      "loss": 0.6805,
+      "step": 9687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.128449776061906e-07,
+      "loss": 0.6282,
+      "step": 9688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.123753032724239e-07,
+      "loss": 0.7068,
+      "step": 9689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.119057838627326e-07,
+      "loss": 0.619,
+      "step": 9690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.114364194156508e-07,
+      "loss": 0.7234,
+      "step": 9691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.109672099697009e-07,
+      "loss": 0.6284,
+      "step": 9692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.104981555633918e-07,
+      "loss": 0.6176,
+      "step": 9693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.100292562352222e-07,
+      "loss": 0.7172,
+      "step": 9694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.095605120236753e-07,
+      "loss": 0.7104,
+      "step": 9695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.090919229672221e-07,
+      "loss": 0.7082,
+      "step": 9696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.086234891043224e-07,
+      "loss": 0.6547,
+      "step": 9697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.081552104734223e-07,
+      "loss": 0.7007,
+      "step": 9698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.076870871129536e-07,
+      "loss": 0.7129,
+      "step": 9699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.07219119061338e-07,
+      "loss": 0.7174,
+      "step": 9700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.067513063569839e-07,
+      "loss": 0.6289,
+      "step": 9701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.062836490382857e-07,
+      "loss": 0.9154,
+      "step": 9702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.058161471436247e-07,
+      "loss": 0.619,
+      "step": 9703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.053488007113723e-07,
+      "loss": 0.7573,
+      "step": 9704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.048816097798843e-07,
+      "loss": 0.7133,
+      "step": 9705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.044145743875046e-07,
+      "loss": 0.6353,
+      "step": 9706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.039476945725642e-07,
+      "loss": 0.6474,
+      "step": 9707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.034809703733826e-07,
+      "loss": 0.6488,
+      "step": 9708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.030144018282652e-07,
+      "loss": 0.8207,
+      "step": 9709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.025479889755037e-07,
+      "loss": 0.6925,
+      "step": 9710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.0208173185338e-07,
+      "loss": 0.7062,
+      "step": 9711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.016156305001608e-07,
+      "loss": 0.6332,
+      "step": 9712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.011496849540999e-07,
+      "loss": 0.6907,
+      "step": 9713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.006838952534405e-07,
+      "loss": 0.6493,
+      "step": 9714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.002182614364108e-07,
+      "loss": 0.6156,
+      "step": 9715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.997527835412273e-07,
+      "loss": 0.7719,
+      "step": 9716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.992874616060926e-07,
+      "loss": 0.5692,
+      "step": 9717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.98822295669198e-07,
+      "loss": 0.6575,
+      "step": 9718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.983572857687222e-07,
+      "loss": 0.6776,
+      "step": 9719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.978924319428287e-07,
+      "loss": 0.6907,
+      "step": 9720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.974277342296708e-07,
+      "loss": 0.8283,
+      "step": 9721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.969631926673877e-07,
+      "loss": 0.9108,
+      "step": 9722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.964988072941058e-07,
+      "loss": 0.7612,
+      "step": 9723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.96034578147938e-07,
+      "loss": 0.6082,
+      "step": 9724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.955705052669866e-07,
+      "loss": 0.6538,
+      "step": 9725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.951065886893392e-07,
+      "loss": 0.6562,
+      "step": 9726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.946428284530705e-07,
+      "loss": 0.6519,
+      "step": 9727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.941792245962442e-07,
+      "loss": 0.841,
+      "step": 9728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.937157771569091e-07,
+      "loss": 0.6127,
+      "step": 9729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.932524861731021e-07,
+      "loss": 0.7673,
+      "step": 9730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.927893516828465e-07,
+      "loss": 0.676,
+      "step": 9731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.923263737241547e-07,
+      "loss": 0.5955,
+      "step": 9732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.918635523350241e-07,
+      "loss": 0.7268,
+      "step": 9733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.914008875534399e-07,
+      "loss": 0.8719,
+      "step": 9734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.909383794173754e-07,
+      "loss": 0.6827,
+      "step": 9735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.904760279647894e-07,
+      "loss": 0.6779,
+      "step": 9736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9001383323363e-07,
+      "loss": 0.6529,
+      "step": 9737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.895517952618296e-07,
+      "loss": 0.6729,
+      "step": 9738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.890899140873107e-07,
+      "loss": 0.7019,
+      "step": 9739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.886281897479812e-07,
+      "loss": 0.6576,
+      "step": 9740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.88166622281735e-07,
+      "loss": 0.6413,
+      "step": 9741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.87705211726457e-07,
+      "loss": 0.7367,
+      "step": 9742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.872439581200154e-07,
+      "loss": 0.6496,
+      "step": 9743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.867828615002671e-07,
+      "loss": 0.6938,
+      "step": 9744
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.8011443018913269,
+      "eval_runtime": 437.6599,
+      "eval_samples_per_second": 6.852,
+      "eval_steps_per_second": 0.857,
+      "step": 9744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.86321921905055e-07,
+      "loss": 0.6983,
+      "step": 9745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.858611393722121e-07,
+      "loss": 0.7414,
+      "step": 9746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.854005139395552e-07,
+      "loss": 0.6806,
+      "step": 9747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.84940045644889e-07,
+      "loss": 0.7395,
+      "step": 9748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.84479734526007e-07,
+      "loss": 0.7202,
+      "step": 9749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.84019580620688e-07,
+      "loss": 0.6847,
+      "step": 9750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.835595839666977e-07,
+      "loss": 0.7167,
+      "step": 9751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.830997446017911e-07,
+      "loss": 0.6506,
+      "step": 9752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.826400625637083e-07,
+      "loss": 0.682,
+      "step": 9753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.82180537890176e-07,
+      "loss": 0.8239,
+      "step": 9754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.817211706189105e-07,
+      "loss": 0.9322,
+      "step": 9755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.812619607876128e-07,
+      "loss": 0.6602,
+      "step": 9756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.808029084339725e-07,
+      "loss": 0.7759,
+      "step": 9757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.803440135956651e-07,
+      "loss": 0.8251,
+      "step": 9758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.798852763103544e-07,
+      "loss": 0.8338,
+      "step": 9759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.794266966156903e-07,
+      "loss": 0.6148,
+      "step": 9760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.7896827454931e-07,
+      "loss": 0.5921,
+      "step": 9761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.78510010148837e-07,
+      "loss": 0.7038,
+      "step": 9762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.780519034518845e-07,
+      "loss": 0.7469,
+      "step": 9763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.775939544960496e-07,
+      "loss": 0.6715,
+      "step": 9764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.771361633189174e-07,
+      "loss": 0.8197,
+      "step": 9765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.766785299580621e-07,
+      "loss": 0.6383,
+      "step": 9766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.762210544510421e-07,
+      "loss": 0.7764,
+      "step": 9767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.757637368354047e-07,
+      "loss": 0.7766,
+      "step": 9768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.753065771486824e-07,
+      "loss": 0.6682,
+      "step": 9769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.748495754283972e-07,
+      "loss": 0.6262,
+      "step": 9770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.743927317120565e-07,
+      "loss": 0.5947,
+      "step": 9771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.739360460371543e-07,
+      "loss": 0.6696,
+      "step": 9772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.73479518441174e-07,
+      "loss": 0.7508,
+      "step": 9773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.730231489615826e-07,
+      "loss": 0.6191,
+      "step": 9774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.725669376358375e-07,
+      "loss": 0.7642,
+      "step": 9775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.721108845013806e-07,
+      "loss": 0.7627,
+      "step": 9776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.716549895956427e-07,
+      "loss": 0.7162,
+      "step": 9777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.711992529560403e-07,
+      "loss": 0.5657,
+      "step": 9778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.707436746199766e-07,
+      "loss": 0.8574,
+      "step": 9779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.702882546248436e-07,
+      "loss": 0.6424,
+      "step": 9780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.69832993008019e-07,
+      "loss": 0.6773,
+      "step": 9781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.693778898068674e-07,
+      "loss": 0.6751,
+      "step": 9782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.689229450587405e-07,
+      "loss": 0.6926,
+      "step": 9783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.684681588009782e-07,
+      "loss": 0.6548,
+      "step": 9784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.680135310709059e-07,
+      "loss": 0.7333,
+      "step": 9785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.675590619058358e-07,
+      "loss": 0.6249,
+      "step": 9786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.671047513430689e-07,
+      "loss": 0.6758,
+      "step": 9787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.666505994198918e-07,
+      "loss": 0.6688,
+      "step": 9788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.661966061735776e-07,
+      "loss": 0.7423,
+      "step": 9789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.657427716413883e-07,
+      "loss": 0.7942,
+      "step": 9790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.652890958605714e-07,
+      "loss": 0.7454,
+      "step": 9791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.648355788683605e-07,
+      "loss": 0.6547,
+      "step": 9792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.643822207019792e-07,
+      "loss": 0.6217,
+      "step": 9793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.639290213986345e-07,
+      "loss": 0.6361,
+      "step": 9794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.634759809955237e-07,
+      "loss": 0.7325,
+      "step": 9795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.63023099529828e-07,
+      "loss": 0.6829,
+      "step": 9796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.625703770387183e-07,
+      "loss": 0.597,
+      "step": 9797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.621178135593505e-07,
+      "loss": 0.7417,
+      "step": 9798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.616654091288679e-07,
+      "loss": 0.784,
+      "step": 9799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.612131637844007e-07,
+      "loss": 0.6617,
+      "step": 9800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.607610775630673e-07,
+      "loss": 0.6498,
+      "step": 9801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.603091505019711e-07,
+      "loss": 0.6904,
+      "step": 9802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.598573826382033e-07,
+      "loss": 0.7023,
+      "step": 9803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.594057740088432e-07,
+      "loss": 0.6782,
+      "step": 9804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.589543246509555e-07,
+      "loss": 0.6545,
+      "step": 9805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.585030346015915e-07,
+      "loss": 0.6588,
+      "step": 9806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.580519038977903e-07,
+      "loss": 0.8058,
+      "step": 9807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.576009325765786e-07,
+      "loss": 0.7286,
+      "step": 9808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.571501206749688e-07,
+      "loss": 0.692,
+      "step": 9809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.566994682299598e-07,
+      "loss": 0.7929,
+      "step": 9810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.5624897527854e-07,
+      "loss": 0.8229,
+      "step": 9811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.557986418576811e-07,
+      "loss": 0.6812,
+      "step": 9812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.553484680043452e-07,
+      "loss": 0.6824,
+      "step": 9813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.548984537554786e-07,
+      "loss": 0.6634,
+      "step": 9814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.544485991480164e-07,
+      "loss": 0.6311,
+      "step": 9815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.539989042188793e-07,
+      "loss": 0.8792,
+      "step": 9816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.535493690049745e-07,
+      "loss": 0.7044,
+      "step": 9817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.530999935431985e-07,
+      "loss": 0.6785,
+      "step": 9818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.526507778704326e-07,
+      "loss": 0.6694,
+      "step": 9819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.522017220235454e-07,
+      "loss": 0.6849,
+      "step": 9820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.517528260393917e-07,
+      "loss": 0.9346,
+      "step": 9821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.513040899548156e-07,
+      "loss": 0.6992,
+      "step": 9822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.508555138066458e-07,
+      "loss": 0.7017,
+      "step": 9823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.504070976316972e-07,
+      "loss": 0.6854,
+      "step": 9824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.499588414667753e-07,
+      "loss": 0.6663,
+      "step": 9825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.495107453486686e-07,
+      "loss": 0.8137,
+      "step": 9826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.490628093141534e-07,
+      "loss": 0.8019,
+      "step": 9827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.486150333999951e-07,
+      "loss": 0.6324,
+      "step": 9828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.481674176429436e-07,
+      "loss": 0.7699,
+      "step": 9829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.47719962079735e-07,
+      "loss": 0.6925,
+      "step": 9830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.472726667470956e-07,
+      "loss": 0.6694,
+      "step": 9831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.468255316817348e-07,
+      "loss": 0.672,
+      "step": 9832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.463785569203522e-07,
+      "loss": 0.6209,
+      "step": 9833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.45931742499631e-07,
+      "loss": 0.6266,
+      "step": 9834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.454850884562444e-07,
+      "loss": 0.6123,
+      "step": 9835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4503859482685e-07,
+      "loss": 0.6705,
+      "step": 9836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.445922616480931e-07,
+      "loss": 0.7958,
+      "step": 9837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.441460889566052e-07,
+      "loss": 0.6646,
+      "step": 9838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.437000767890072e-07,
+      "loss": 0.6685,
+      "step": 9839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.432542251819034e-07,
+      "loss": 0.8051,
+      "step": 9840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.428085341718861e-07,
+      "loss": 0.7326,
+      "step": 9841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.423630037955363e-07,
+      "loss": 0.6917,
+      "step": 9842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.419176340894194e-07,
+      "loss": 0.6728,
+      "step": 9843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.414724250900885e-07,
+      "loss": 0.7865,
+      "step": 9844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.410273768340826e-07,
+      "loss": 0.7618,
+      "step": 9845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4058248935793e-07,
+      "loss": 0.6798,
+      "step": 9846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.401377626981435e-07,
+      "loss": 0.6281,
+      "step": 9847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.396931968912227e-07,
+      "loss": 0.7168,
+      "step": 9848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.392487919736558e-07,
+      "loss": 0.764,
+      "step": 9849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.38804547981916e-07,
+      "loss": 0.6794,
+      "step": 9850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.383604649524646e-07,
+      "loss": 0.6882,
+      "step": 9851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.379165429217481e-07,
+      "loss": 0.7915,
+      "step": 9852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.374727819262018e-07,
+      "loss": 0.7312,
+      "step": 9853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.370291820022466e-07,
+      "loss": 0.6782,
+      "step": 9854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.365857431862892e-07,
+      "loss": 0.6991,
+      "step": 9855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.36142465514726e-07,
+      "loss": 0.5944,
+      "step": 9856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.35699349023937e-07,
+      "loss": 0.6961,
+      "step": 9857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.352563937502911e-07,
+      "loss": 0.5833,
+      "step": 9858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.348135997301418e-07,
+      "loss": 0.7909,
+      "step": 9859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.343709669998329e-07,
+      "loss": 0.631,
+      "step": 9860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.339284955956917e-07,
+      "loss": 0.6602,
+      "step": 9861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.334861855540327e-07,
+      "loss": 0.694,
+      "step": 9862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.330440369111595e-07,
+      "loss": 0.6926,
+      "step": 9863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.326020497033597e-07,
+      "loss": 0.5878,
+      "step": 9864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.321602239669091e-07,
+      "loss": 0.7208,
+      "step": 9865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.317185597380695e-07,
+      "loss": 0.5665,
+      "step": 9866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.312770570530903e-07,
+      "loss": 0.7631,
+      "step": 9867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.308357159482069e-07,
+      "loss": 0.687,
+      "step": 9868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.303945364596424e-07,
+      "loss": 0.7352,
+      "step": 9869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.299535186236049e-07,
+      "loss": 0.7535,
+      "step": 9870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.295126624762914e-07,
+      "loss": 0.7159,
+      "step": 9871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.290719680538836e-07,
+      "loss": 0.6302,
+      "step": 9872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.286314353925517e-07,
+      "loss": 0.6995,
+      "step": 9873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.281910645284516e-07,
+      "loss": 0.7291,
+      "step": 9874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.277508554977256e-07,
+      "loss": 0.7753,
+      "step": 9875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.27310808336503e-07,
+      "loss": 0.6943,
+      "step": 9876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.268709230809014e-07,
+      "loss": 0.6508,
+      "step": 9877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.264311997670227e-07,
+      "loss": 0.7311,
+      "step": 9878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.25991638430956e-07,
+      "loss": 0.6028,
+      "step": 9879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.255522391087789e-07,
+      "loss": 0.5766,
+      "step": 9880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.251130018365544e-07,
+      "loss": 0.6703,
+      "step": 9881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.246739266503315e-07,
+      "loss": 0.6856,
+      "step": 9882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.242350135861462e-07,
+      "loss": 0.7722,
+      "step": 9883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.237962626800236e-07,
+      "loss": 0.6198,
+      "step": 9884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.233576739679721e-07,
+      "loss": 0.6526,
+      "step": 9885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.229192474859878e-07,
+      "loss": 0.9169,
+      "step": 9886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.224809832700556e-07,
+      "loss": 0.7248,
+      "step": 9887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.220428813561434e-07,
+      "loss": 0.5942,
+      "step": 9888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.216049417802096e-07,
+      "loss": 0.6458,
+      "step": 9889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.211671645781963e-07,
+      "loss": 0.6034,
+      "step": 9890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.207295497860343e-07,
+      "loss": 0.5974,
+      "step": 9891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.2029209743964e-07,
+      "loss": 0.6411,
+      "step": 9892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.198548075749154e-07,
+      "loss": 0.7144,
+      "step": 9893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.194176802277523e-07,
+      "loss": 0.7788,
+      "step": 9894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.189807154340262e-07,
+      "loss": 0.7392,
+      "step": 9895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.185439132296008e-07,
+      "loss": 0.706,
+      "step": 9896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.181072736503251e-07,
+      "loss": 0.6316,
+      "step": 9897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.176707967320371e-07,
+      "loss": 0.6791,
+      "step": 9898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.172344825105591e-07,
+      "loss": 0.7501,
+      "step": 9899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.167983310217006e-07,
+      "loss": 0.7045,
+      "step": 9900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.163623423012593e-07,
+      "loss": 0.664,
+      "step": 9901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.159265163850177e-07,
+      "loss": 0.688,
+      "step": 9902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.154908533087455e-07,
+      "loss": 0.5949,
+      "step": 9903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.150553531081984e-07,
+      "loss": 0.611,
+      "step": 9904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.146200158191211e-07,
+      "loss": 0.6991,
+      "step": 9905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.141848414772418e-07,
+      "loss": 0.6748,
+      "step": 9906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.137498301182781e-07,
+      "loss": 0.6824,
+      "step": 9907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.133149817779315e-07,
+      "loss": 0.6942,
+      "step": 9908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.12880296491893e-07,
+      "loss": 0.8801,
+      "step": 9909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.124457742958375e-07,
+      "loss": 0.6827,
+      "step": 9910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.120114152254288e-07,
+      "loss": 0.6718,
+      "step": 9911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.11577219316316e-07,
+      "loss": 0.7142,
+      "step": 9912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.111431866041349e-07,
+      "loss": 0.6319,
+      "step": 9913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.107093171245076e-07,
+      "loss": 0.707,
+      "step": 9914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.102756109130444e-07,
+      "loss": 0.66,
+      "step": 9915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.098420680053407e-07,
+      "loss": 0.7751,
+      "step": 9916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.094086884369781e-07,
+      "loss": 0.6825,
+      "step": 9917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.089754722435269e-07,
+      "loss": 0.7541,
+      "step": 9918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.085424194605423e-07,
+      "loss": 0.748,
+      "step": 9919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.081095301235661e-07,
+      "loss": 0.7791,
+      "step": 9920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.076768042681265e-07,
+      "loss": 0.637,
+      "step": 9921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.072442419297405e-07,
+      "loss": 0.7245,
+      "step": 9922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.068118431439089e-07,
+      "loss": 0.794,
+      "step": 9923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0637960794612e-07,
+      "loss": 0.7054,
+      "step": 9924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.059475363718496e-07,
+      "loss": 0.7044,
+      "step": 9925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.055156284565588e-07,
+      "loss": 0.7319,
+      "step": 9926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.050838842356967e-07,
+      "loss": 0.7535,
+      "step": 9927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.046523037446965e-07,
+      "loss": 0.6734,
+      "step": 9928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.042208870189816e-07,
+      "loss": 0.7685,
+      "step": 9929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.037896340939585e-07,
+      "loss": 0.7169,
+      "step": 9930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.033585450050215e-07,
+      "loss": 0.7185,
+      "step": 9931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.029276197875527e-07,
+      "loss": 0.6304,
+      "step": 9932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.024968584769191e-07,
+      "loss": 0.6731,
+      "step": 9933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.020662611084745e-07,
+      "loss": 0.6774,
+      "step": 9934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.016358277175596e-07,
+      "loss": 0.6554,
+      "step": 9935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.012055583395022e-07,
+      "loss": 0.6426,
+      "step": 9936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.007754530096157e-07,
+      "loss": 0.6681,
+      "step": 9937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.003455117631997e-07,
+      "loss": 0.6199,
+      "step": 9938
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.999157346355423e-07,
+      "loss": 0.6839,
+      "step": 9939
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.994861216619162e-07,
+      "loss": 0.8535,
+      "step": 9940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.990566728775814e-07,
+      "loss": 0.6074,
+      "step": 9941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.986273883177834e-07,
+      "loss": 0.6397,
+      "step": 9942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.981982680177566e-07,
+      "loss": 0.6982,
+      "step": 9943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.977693120127189e-07,
+      "loss": 0.6735,
+      "step": 9944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.973405203378779e-07,
+      "loss": 0.6762,
+      "step": 9945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.969118930284242e-07,
+      "loss": 0.7923,
+      "step": 9946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.964834301195388e-07,
+      "loss": 0.6162,
+      "step": 9947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.960551316463849e-07,
+      "loss": 0.64,
+      "step": 9948
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.95626997644117e-07,
+      "loss": 0.6765,
+      "step": 9949
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.95199028147872e-07,
+      "loss": 0.6757,
+      "step": 9950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.947712231927751e-07,
+      "loss": 0.6985,
+      "step": 9951
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.943435828139373e-07,
+      "loss": 0.697,
+      "step": 9952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.939161070464577e-07,
+      "loss": 0.6106,
+      "step": 9953
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.934887959254203e-07,
+      "loss": 0.6016,
+      "step": 9954
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.93061649485895e-07,
+      "loss": 0.7416,
+      "step": 9955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.926346677629412e-07,
+      "loss": 0.758,
+      "step": 9956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.922078507916014e-07,
+      "loss": 0.7904,
+      "step": 9957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.917811986069065e-07,
+      "loss": 0.7036,
+      "step": 9958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.913547112438726e-07,
+      "loss": 0.7108,
+      "step": 9959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.909283887375044e-07,
+      "loss": 0.7035,
+      "step": 9960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.905022311227905e-07,
+      "loss": 0.6485,
+      "step": 9961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.900762384347074e-07,
+      "loss": 0.6717,
+      "step": 9962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.896504107082184e-07,
+      "loss": 0.7143,
+      "step": 9963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.892247479782719e-07,
+      "loss": 0.6459,
+      "step": 9964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.887992502798047e-07,
+      "loss": 0.7519,
+      "step": 9965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.883739176477373e-07,
+      "loss": 0.6207,
+      "step": 9966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8794875011698e-07,
+      "loss": 0.7818,
+      "step": 9967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.87523747722427e-07,
+      "loss": 0.6343,
+      "step": 9968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.87098910498959e-07,
+      "loss": 0.6119,
+      "step": 9969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.866742384814455e-07,
+      "loss": 0.6951,
+      "step": 9970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.862497317047399e-07,
+      "loss": 0.6499,
+      "step": 9971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.858253902036828e-07,
+      "loss": 0.629,
+      "step": 9972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.854012140131012e-07,
+      "loss": 0.6328,
+      "step": 9973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.849772031678102e-07,
+      "loss": 0.8736,
+      "step": 9974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.845533577026085e-07,
+      "loss": 0.6143,
+      "step": 9975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.841296776522824e-07,
+      "loss": 0.742,
+      "step": 9976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.837061630516063e-07,
+      "loss": 0.6258,
+      "step": 9977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.832828139353388e-07,
+      "loss": 0.5958,
+      "step": 9978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.828596303382252e-07,
+      "loss": 0.7389,
+      "step": 9979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.824366122949978e-07,
+      "loss": 0.6205,
+      "step": 9980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.820137598403757e-07,
+      "loss": 0.624,
+      "step": 9981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.815910730090633e-07,
+      "loss": 0.7142,
+      "step": 9982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.811685518357528e-07,
+      "loss": 0.7714,
+      "step": 9983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.807461963551211e-07,
+      "loss": 0.6702,
+      "step": 9984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.803240066018336e-07,
+      "loss": 0.8067,
+      "step": 9985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.799019826105395e-07,
+      "loss": 0.6112,
+      "step": 9986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.79480124415877e-07,
+      "loss": 0.6652,
+      "step": 9987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.790584320524691e-07,
+      "loss": 0.7143,
+      "step": 9988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.786369055549256e-07,
+      "loss": 0.6487,
+      "step": 9989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.782155449578419e-07,
+      "loss": 0.7404,
+      "step": 9990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.777943502958018e-07,
+      "loss": 0.7204,
+      "step": 9991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.77373321603374e-07,
+      "loss": 0.7285,
+      "step": 9992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769524589151125e-07,
+      "loss": 0.666,
+      "step": 9993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.76531762265561e-07,
+      "loss": 0.5342,
+      "step": 9994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.761112316892466e-07,
+      "loss": 0.6517,
+      "step": 9995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7569086722068365e-07,
+      "loss": 0.6806,
+      "step": 9996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.752706688943723e-07,
+      "loss": 0.7173,
+      "step": 9997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7485063674480107e-07,
+      "loss": 0.7349,
+      "step": 9998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.74430770806443e-07,
+      "loss": 0.6846,
+      "step": 9999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7401107111375744e-07,
+      "loss": 0.7106,
+      "step": 10000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7359153770119125e-07,
+      "loss": 0.7257,
+      "step": 10001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7317217060317664e-07,
+      "loss": 0.7223,
+      "step": 10002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7275296985413306e-07,
+      "loss": 0.6183,
+      "step": 10003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7233393548846514e-07,
+      "loss": 0.6668,
+      "step": 10004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7191506754056537e-07,
+      "loss": 0.7684,
+      "step": 10005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.714963660448113e-07,
+      "loss": 0.7057,
+      "step": 10006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.710778310355665e-07,
+      "loss": 0.8576,
+      "step": 10007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.706594625471833e-07,
+      "loss": 0.7127,
+      "step": 10008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.702412606139975e-07,
+      "loss": 0.6347,
+      "step": 10009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.698232252703325e-07,
+      "loss": 0.7509,
+      "step": 10010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6940535655049747e-07,
+      "loss": 0.7031,
+      "step": 10011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6898765448878977e-07,
+      "loss": 0.7372,
+      "step": 10012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6857011911949107e-07,
+      "loss": 0.6041,
+      "step": 10013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6815275047686906e-07,
+      "loss": 0.6875,
+      "step": 10014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.677355485951801e-07,
+      "loss": 0.7275,
+      "step": 10015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.673185135086647e-07,
+      "loss": 0.7322,
+      "step": 10016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.669016452515507e-07,
+      "loss": 0.7343,
+      "step": 10017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.66484943858051e-07,
+      "loss": 0.7688,
+      "step": 10018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6606840936236714e-07,
+      "loss": 0.5999,
+      "step": 10019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.656520417986843e-07,
+      "loss": 0.7877,
+      "step": 10020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.652358412011768e-07,
+      "loss": 0.675,
+      "step": 10021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.648198076040017e-07,
+      "loss": 0.6403,
+      "step": 10022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6440394104130647e-07,
+      "loss": 0.6522,
+      "step": 10023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6398824154722107e-07,
+      "loss": 0.6594,
+      "step": 10024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6357270915586507e-07,
+      "loss": 0.6345,
+      "step": 10025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6315734390134123e-07,
+      "loss": 0.7029,
+      "step": 10026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.627421458177408e-07,
+      "loss": 0.7472,
+      "step": 10027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.623271149391398e-07,
+      "loss": 0.6849,
+      "step": 10028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6191225129960243e-07,
+      "loss": 0.606,
+      "step": 10029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6149755493317764e-07,
+      "loss": 0.7855,
+      "step": 10030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.610830258738999e-07,
+      "loss": 0.7634,
+      "step": 10031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.606686641557928e-07,
+      "loss": 0.8249,
+      "step": 10032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.602544698128636e-07,
+      "loss": 0.7274,
+      "step": 10033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5984044287910673e-07,
+      "loss": 0.7647,
+      "step": 10034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5942658338850254e-07,
+      "loss": 0.7049,
+      "step": 10035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.590128913750186e-07,
+      "loss": 0.6053,
+      "step": 10036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.585993668726083e-07,
+      "loss": 0.7994,
+      "step": 10037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.581860099152094e-07,
+      "loss": 0.7,
+      "step": 10038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5777282053674986e-07,
+      "loss": 0.6711,
+      "step": 10039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.573597987711395e-07,
+      "loss": 0.736,
+      "step": 10040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.569469446522784e-07,
+      "loss": 0.5938,
+      "step": 10041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5653425821404916e-07,
+      "loss": 0.791,
+      "step": 10042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.561217394903242e-07,
+      "loss": 0.6872,
+      "step": 10043
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.557093885149594e-07,
+      "loss": 0.6402,
+      "step": 10044
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5529720532179743e-07,
+      "loss": 0.7828,
+      "step": 10045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.548851899446688e-07,
+      "loss": 0.6008,
+      "step": 10046
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5447334241738863e-07,
+      "loss": 0.5615,
+      "step": 10047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5406166277375816e-07,
+      "loss": 0.6661,
+      "step": 10048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.536501510475655e-07,
+      "loss": 0.6773,
+      "step": 10049
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5323880727258575e-07,
+      "loss": 0.6652,
+      "step": 10050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.528276314825786e-07,
+      "loss": 0.6696,
+      "step": 10051
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5241662371129036e-07,
+      "loss": 0.6675,
+      "step": 10052
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5200578399245494e-07,
+      "loss": 0.6988,
+      "step": 10053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5159511235979116e-07,
+      "loss": 0.6647,
+      "step": 10054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5118460884700397e-07,
+      "loss": 0.7115,
+      "step": 10055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.507742734877843e-07,
+      "loss": 0.6483,
+      "step": 10056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.50364106315811e-07,
+      "loss": 0.62,
+      "step": 10057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.499541073647468e-07,
+      "loss": 0.7229,
+      "step": 10058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.495442766682431e-07,
+      "loss": 0.6453,
+      "step": 10059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4913461425993483e-07,
+      "loss": 0.7372,
+      "step": 10060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.487251201734455e-07,
+      "loss": 0.6889,
+      "step": 10061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.48315794442383e-07,
+      "loss": 0.817,
+      "step": 10062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4790663710034293e-07,
+      "loss": 0.669,
+      "step": 10063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.474976481809057e-07,
+      "loss": 0.6423,
+      "step": 10064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4708882771763887e-07,
+      "loss": 0.5953,
+      "step": 10065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.466801757440947e-07,
+      "loss": 0.6855,
+      "step": 10066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.46271692293814e-07,
+      "loss": 0.618,
+      "step": 10067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4586337740032235e-07,
+      "loss": 0.6761,
+      "step": 10068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.454552310971305e-07,
+      "loss": 0.7325,
+      "step": 10069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4504725341773777e-07,
+      "loss": 0.8472,
+      "step": 10070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4463944439562785e-07,
+      "loss": 0.7098,
+      "step": 10071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.44231804064271e-07,
+      "loss": 0.7056,
+      "step": 10072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4382433245712315e-07,
+      "loss": 0.6795,
+      "step": 10073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4341702960762814e-07,
+      "loss": 0.6451,
+      "step": 10074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4300989554921403e-07,
+      "loss": 0.6608,
+      "step": 10075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.426029303152954e-07,
+      "loss": 0.6966,
+      "step": 10076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.421961339392744e-07,
+      "loss": 0.7863,
+      "step": 10077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.417895064545369e-07,
+      "loss": 0.685,
+      "step": 10078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.413830478944578e-07,
+      "loss": 1.0784,
+      "step": 10079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4097675829239515e-07,
+      "loss": 0.7671,
+      "step": 10080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4057063768169594e-07,
+      "loss": 0.6332,
+      "step": 10081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.401646860956912e-07,
+      "loss": 0.693,
+      "step": 10082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3975890356769827e-07,
+      "loss": 0.7169,
+      "step": 10083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.393532901310224e-07,
+      "loss": 0.5853,
+      "step": 10084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.389478458189531e-07,
+      "loss": 0.6506,
+      "step": 10085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3854257066476666e-07,
+      "loss": 0.6336,
+      "step": 10086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3813746470172475e-07,
+      "loss": 0.7153,
+      "step": 10087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.377325279630773e-07,
+      "loss": 0.6094,
+      "step": 10088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3732776048205803e-07,
+      "loss": 0.6901,
+      "step": 10089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.369231622918871e-07,
+      "loss": 0.6644,
+      "step": 10090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.365187334257728e-07,
+      "loss": 0.6636,
+      "step": 10091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.361144739169071e-07,
+      "loss": 0.7251,
+      "step": 10092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3571038379846915e-07,
+      "loss": 0.8365,
+      "step": 10093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.353064631036233e-07,
+      "loss": 0.7769,
+      "step": 10094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.349027118655222e-07,
+      "loss": 0.6495,
+      "step": 10095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3449913011730183e-07,
+      "loss": 0.7005,
+      "step": 10096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.340957178920871e-07,
+      "loss": 0.7053,
+      "step": 10097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.336924752229857e-07,
+      "loss": 0.6804,
+      "step": 10098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.332894021430947e-07,
+      "loss": 0.7946,
+      "step": 10099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.328864986854947e-07,
+      "loss": 0.7671,
+      "step": 10100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.324837648832544e-07,
+      "loss": 0.7415,
+      "step": 10101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3208120076942694e-07,
+      "loss": 0.6519,
+      "step": 10102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3167880637705245e-07,
+      "loss": 0.7016,
+      "step": 10103
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.312765817391559e-07,
+      "loss": 0.6429,
+      "step": 10104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3087452688875096e-07,
+      "loss": 0.7249,
+      "step": 10105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3047264185883483e-07,
+      "loss": 0.7356,
+      "step": 10106
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3007092668239103e-07,
+      "loss": 0.6951,
+      "step": 10107
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.296693813923908e-07,
+      "loss": 0.7354,
+      "step": 10108
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.292680060217902e-07,
+      "loss": 0.7458,
+      "step": 10109
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2886680060353124e-07,
+      "loss": 0.5673,
+      "step": 10110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.284657651705418e-07,
+      "loss": 0.7774,
+      "step": 10111
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2806489975573737e-07,
+      "loss": 0.6807,
+      "step": 10112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2766420439201787e-07,
+      "loss": 0.6771,
+      "step": 10113
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2726367911226943e-07,
+      "loss": 0.6763,
+      "step": 10114
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2686332394936526e-07,
+      "loss": 0.6639,
+      "step": 10115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.264631389361631e-07,
+      "loss": 0.6699,
+      "step": 10116
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.260631241055088e-07,
+      "loss": 0.6672,
+      "step": 10117
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2566327949023154e-07,
+      "loss": 0.5744,
+      "step": 10118
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.252636051231496e-07,
+      "loss": 0.6623,
+      "step": 10119
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2486410103706437e-07,
+      "loss": 0.7265,
+      "step": 10120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.244647672647648e-07,
+      "loss": 0.6612,
+      "step": 10121
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2406560383902615e-07,
+      "loss": 0.7152,
+      "step": 10122
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.236666107926091e-07,
+      "loss": 0.7362,
+      "step": 10123
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2326778815826003e-07,
+      "loss": 0.8908,
+      "step": 10124
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2286913596871147e-07,
+      "loss": 0.7331,
+      "step": 10125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.224706542566831e-07,
+      "loss": 0.7302,
+      "step": 10126
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.220723430548795e-07,
+      "loss": 0.6573,
+      "step": 10127
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.216742023959905e-07,
+      "loss": 0.8023,
+      "step": 10128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2127623231269424e-07,
+      "loss": 0.6899,
+      "step": 10129
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2087843283765335e-07,
+      "loss": 0.6966,
+      "step": 10130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2048080400351606e-07,
+      "loss": 0.6899,
+      "step": 10131
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2008334584291684e-07,
+      "loss": 0.6394,
+      "step": 10132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1968605838847765e-07,
+      "loss": 0.6347,
+      "step": 10133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.192889416728041e-07,
+      "loss": 0.7668,
+      "step": 10134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.188919957284901e-07,
+      "loss": 0.6132,
+      "step": 10135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1849522058811347e-07,
+      "loss": 0.7192,
+      "step": 10136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.180986162842399e-07,
+      "loss": 0.6578,
+      "step": 10137
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1770218284941905e-07,
+      "loss": 0.7491,
+      "step": 10138
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.173059203161886e-07,
+      "loss": 0.7894,
+      "step": 10139
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1690982871707115e-07,
+      "loss": 0.6787,
+      "step": 10140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.165139080845748e-07,
+      "loss": 0.769,
+      "step": 10141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1611815845119363e-07,
+      "loss": 0.7399,
+      "step": 10142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1572257984940967e-07,
+      "loss": 0.765,
+      "step": 10143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.153271723116889e-07,
+      "loss": 0.738,
+      "step": 10144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1493193587048284e-07,
+      "loss": 0.7066,
+      "step": 10145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1453687055823143e-07,
+      "loss": 0.6901,
+      "step": 10146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.141419764073587e-07,
+      "loss": 0.7714,
+      "step": 10147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1374725345027456e-07,
+      "loss": 0.6802,
+      "step": 10148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1335270171937474e-07,
+      "loss": 0.8152,
+      "step": 10149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.129583212470431e-07,
+      "loss": 0.6083,
+      "step": 10150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1256411206564704e-07,
+      "loss": 0.6738,
+      "step": 10151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1217007420753993e-07,
+      "loss": 0.6071,
+      "step": 10152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.117762077050633e-07,
+      "loss": 0.7021,
+      "step": 10153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1138251259054213e-07,
+      "loss": 0.6524,
+      "step": 10154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1098898889628896e-07,
+      "loss": 0.7103,
+      "step": 10155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.105956366546013e-07,
+      "loss": 0.6618,
+      "step": 10156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.102024558977635e-07,
+      "loss": 0.7947,
+      "step": 10157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.098094466580449e-07,
+      "loss": 0.577,
+      "step": 10158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.094166089677007e-07,
+      "loss": 0.6218,
+      "step": 10159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0902394285897373e-07,
+      "loss": 0.6743,
+      "step": 10160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.086314483640905e-07,
+      "loss": 0.7358,
+      "step": 10161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0823912551526487e-07,
+      "loss": 0.7975,
+      "step": 10162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.078469743446956e-07,
+      "loss": 0.5826,
+      "step": 10163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0745499488456855e-07,
+      "loss": 0.8409,
+      "step": 10164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0706318716705466e-07,
+      "loss": 0.6627,
+      "step": 10165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.066715512243105e-07,
+      "loss": 0.7721,
+      "step": 10166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0628008708848033e-07,
+      "loss": 0.6815,
+      "step": 10167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0588879479169174e-07,
+      "loss": 0.6383,
+      "step": 10168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0549767436606017e-07,
+      "loss": 0.6048,
+      "step": 10169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0510672584368525e-07,
+      "loss": 0.6396,
+      "step": 10170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.047159492566552e-07,
+      "loss": 0.5531,
+      "step": 10171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.043253446370404e-07,
+      "loss": 0.7001,
+      "step": 10172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.039349120169006e-07,
+      "loss": 0.8177,
+      "step": 10173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0354465142828036e-07,
+      "loss": 0.7999,
+      "step": 10174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.031545629032088e-07,
+      "loss": 0.794,
+      "step": 10175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.027646464737017e-07,
+      "loss": 0.668,
+      "step": 10176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.023749021717621e-07,
+      "loss": 0.6948,
+      "step": 10177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.019853300293766e-07,
+      "loss": 0.562,
+      "step": 10178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.015959300785194e-07,
+      "loss": 0.5755,
+      "step": 10179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.012067023511487e-07,
+      "loss": 0.6786,
+      "step": 10180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.008176468792116e-07,
+      "loss": 0.7184,
+      "step": 10181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.004287636946383e-07,
+      "loss": 0.6842,
+      "step": 10182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.000400528293455e-07,
+      "loss": 0.6696,
+      "step": 10183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9965151431523686e-07,
+      "loss": 0.6976,
+      "step": 10184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9926314818420105e-07,
+      "loss": 0.7046,
+      "step": 10185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.988749544681123e-07,
+      "loss": 0.8832,
+      "step": 10186
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9848693319883056e-07,
+      "loss": 0.7155,
+      "step": 10187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9809908440820303e-07,
+      "loss": 0.6058,
+      "step": 10188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.977114081280617e-07,
+      "loss": 0.5853,
+      "step": 10189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9732390439022383e-07,
+      "loss": 0.7774,
+      "step": 10190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9693657322649377e-07,
+      "loss": 0.7003,
+      "step": 10191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.965494146686616e-07,
+      "loss": 0.6449,
+      "step": 10192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9616242874850246e-07,
+      "loss": 0.8224,
+      "step": 10193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9577561549777685e-07,
+      "loss": 0.8699,
+      "step": 10194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9538897494823314e-07,
+      "loss": 0.6329,
+      "step": 10195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.950025071316038e-07,
+      "loss": 0.6327,
+      "step": 10196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.946162120796071e-07,
+      "loss": 0.714,
+      "step": 10197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.942300898239487e-07,
+      "loss": 0.75,
+      "step": 10198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9384414039631837e-07,
+      "loss": 0.6372,
+      "step": 10199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.934583638283926e-07,
+      "loss": 0.7904,
+      "step": 10200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.930727601518325e-07,
+      "loss": 0.7705,
+      "step": 10201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9268732939828744e-07,
+      "loss": 0.6327,
+      "step": 10202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9230207159939025e-07,
+      "loss": 0.6743,
+      "step": 10203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9191698678676005e-07,
+      "loss": 0.6099,
+      "step": 10204
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.915320749920032e-07,
+      "loss": 0.6577,
+      "step": 10205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9114733624671023e-07,
+      "loss": 0.5575,
+      "step": 10206
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9076277058245795e-07,
+      "loss": 0.7753,
+      "step": 10207
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.903783780308082e-07,
+      "loss": 0.7277,
+      "step": 10208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8999415862331054e-07,
+      "loss": 0.6592,
+      "step": 10209
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.896101123914994e-07,
+      "loss": 0.7395,
+      "step": 10210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8922623936689414e-07,
+      "loss": 0.6155,
+      "step": 10211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.888425395810011e-07,
+      "loss": 0.6552,
+      "step": 10212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.884590130653118e-07,
+      "loss": 0.7206,
+      "step": 10213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8807565985130273e-07,
+      "loss": 0.6665,
+      "step": 10214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8769247997043844e-07,
+      "loss": 0.6467,
+      "step": 10215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.873094734541674e-07,
+      "loss": 0.6669,
+      "step": 10216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.869266403339239e-07,
+      "loss": 0.7713,
+      "step": 10217
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.865439806411281e-07,
+      "loss": 0.7862,
+      "step": 10218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.861614944071873e-07,
+      "loss": 0.7782,
+      "step": 10219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8577918166349313e-07,
+      "loss": 0.8835,
+      "step": 10220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8539704244142245e-07,
+      "loss": 0.6821,
+      "step": 10221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.850150767723404e-07,
+      "loss": 0.6114,
+      "step": 10222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8463328468759527e-07,
+      "loss": 0.7807,
+      "step": 10223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8425166621852224e-07,
+      "loss": 0.7492,
+      "step": 10224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8387022139644124e-07,
+      "loss": 0.8845,
+      "step": 10225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.834889502526606e-07,
+      "loss": 0.763,
+      "step": 10226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8310785281847146e-07,
+      "loss": 0.665,
+      "step": 10227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8272692912515173e-07,
+      "loss": 0.6556,
+      "step": 10228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.823461792039651e-07,
+      "loss": 0.7275,
+      "step": 10229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8196560308616237e-07,
+      "loss": 0.7761,
+      "step": 10230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8158520080297773e-07,
+      "loss": 0.6483,
+      "step": 10231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.812049723856315e-07,
+      "loss": 0.6303,
+      "step": 10232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.8082491786533197e-07,
+      "loss": 0.6279,
+      "step": 10233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.804450372732707e-07,
+      "loss": 0.6983,
+      "step": 10234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.800653306406252e-07,
+      "loss": 0.7317,
+      "step": 10235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7968579799856077e-07,
+      "loss": 0.683,
+      "step": 10236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.793064393782261e-07,
+      "loss": 0.6737,
+      "step": 10237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.789272548107567e-07,
+      "loss": 0.6648,
+      "step": 10238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.785482443272731e-07,
+      "loss": 0.7028,
+      "step": 10239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.78169407958883e-07,
+      "loss": 0.6453,
+      "step": 10240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7779074573667864e-07,
+      "loss": 0.7833,
+      "step": 10241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.774122576917369e-07,
+      "loss": 0.8453,
+      "step": 10242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.770339438551235e-07,
+      "loss": 0.7875,
+      "step": 10243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7665580425788687e-07,
+      "loss": 0.7031,
+      "step": 10244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.762778389310628e-07,
+      "loss": 0.5904,
+      "step": 10245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.759000479056713e-07,
+      "loss": 0.7809,
+      "step": 10246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.755224312127195e-07,
+      "loss": 0.6406,
+      "step": 10247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7514498888320093e-07,
+      "loss": 0.6433,
+      "step": 10248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.747677209480918e-07,
+      "loss": 0.6349,
+      "step": 10249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.743906274383574e-07,
+      "loss": 0.7807,
+      "step": 10250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7401370838494625e-07,
+      "loss": 0.7661,
+      "step": 10251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7363696381879313e-07,
+      "loss": 0.6697,
+      "step": 10252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.732603937708198e-07,
+      "loss": 0.6431,
+      "step": 10253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7288399827193234e-07,
+      "loss": 0.8089,
+      "step": 10254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.725077773530225e-07,
+      "loss": 0.6749,
+      "step": 10255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7213173104496794e-07,
+      "loss": 0.839,
+      "step": 10256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7175585937863273e-07,
+      "loss": 0.7145,
+      "step": 10257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.713801623848659e-07,
+      "loss": 0.6146,
+      "step": 10258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7100464009450137e-07,
+      "loss": 0.6817,
+      "step": 10259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.706292925383612e-07,
+      "loss": 0.6538,
+      "step": 10260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7025411974725033e-07,
+      "loss": 0.7632,
+      "step": 10261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6987912175196066e-07,
+      "loss": 0.6057,
+      "step": 10262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.695042985832695e-07,
+      "loss": 0.8314,
+      "step": 10263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.691296502719405e-07,
+      "loss": 0.8002,
+      "step": 10264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.687551768487213e-07,
+      "loss": 0.5954,
+      "step": 10265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6838087834434793e-07,
+      "loss": 0.7454,
+      "step": 10266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.680067547895386e-07,
+      "loss": 0.6663,
+      "step": 10267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.676328062150006e-07,
+      "loss": 0.6201,
+      "step": 10268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6725903265142414e-07,
+      "loss": 0.6272,
+      "step": 10269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.668854341294861e-07,
+      "loss": 0.8597,
+      "step": 10270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6651201067985003e-07,
+      "loss": 0.6875,
+      "step": 10271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.661387623331633e-07,
+      "loss": 0.7104,
+      "step": 10272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.657656891200595e-07,
+      "loss": 0.8799,
+      "step": 10273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.653927910711591e-07,
+      "loss": 0.7607,
+      "step": 10274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6502006821706654e-07,
+      "loss": 0.797,
+      "step": 10275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.646475205883723e-07,
+      "loss": 0.6758,
+      "step": 10276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.642751482156526e-07,
+      "loss": 0.7493,
+      "step": 10277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6390295112947e-07,
+      "loss": 0.6879,
+      "step": 10278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.63530929360372e-07,
+      "loss": 0.7779,
+      "step": 10279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.631590829388909e-07,
+      "loss": 0.7248,
+      "step": 10280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.627874118955463e-07,
+      "loss": 0.7149,
+      "step": 10281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.624159162608426e-07,
+      "loss": 0.5917,
+      "step": 10282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6204459606526884e-07,
+      "loss": 0.8279,
+      "step": 10283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6167345133930166e-07,
+      "loss": 0.6886,
+      "step": 10284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.613024821134012e-07,
+      "loss": 0.6585,
+      "step": 10285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6093168841801555e-07,
+      "loss": 0.7458,
+      "step": 10286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6056107028357547e-07,
+      "loss": 0.6904,
+      "step": 10287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6019062774050066e-07,
+      "loss": 0.6775,
+      "step": 10288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598203608191936e-07,
+      "loss": 0.6062,
+      "step": 10289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594502695500429e-07,
+      "loss": 0.7078,
+      "step": 10290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5908035396342466e-07,
+      "loss": 0.5959,
+      "step": 10291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.587106140896984e-07,
+      "loss": 0.772,
+      "step": 10292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.583410499592102e-07,
+      "loss": 0.7262,
+      "step": 10293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5797166160229047e-07,
+      "loss": 0.7683,
+      "step": 10294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.57602449049258e-07,
+      "loss": 0.6756,
+      "step": 10295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.572334123304141e-07,
+      "loss": 0.693,
+      "step": 10296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5686455147604715e-07,
+      "loss": 0.7264,
+      "step": 10297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5649586651643115e-07,
+      "loss": 0.639,
+      "step": 10298
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5612735748182553e-07,
+      "loss": 0.7277,
+      "step": 10299
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.557590244024747e-07,
+      "loss": 0.8641,
+      "step": 10300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.55390867308609e-07,
+      "loss": 0.6907,
+      "step": 10301
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.55022886230445e-07,
+      "loss": 0.6482,
+      "step": 10302
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.54655081198183e-07,
+      "loss": 0.6122,
+      "step": 10303
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.542874522420117e-07,
+      "loss": 0.6447,
+      "step": 10304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.539199993921025e-07,
+      "loss": 0.7076,
+      "step": 10305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.535527226786145e-07,
+      "loss": 0.735,
+      "step": 10306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5318562213169065e-07,
+      "loss": 0.5664,
+      "step": 10307
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5281869778146015e-07,
+      "loss": 0.7624,
+      "step": 10308
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5245194965803865e-07,
+      "loss": 0.6708,
+      "step": 10309
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5208537779152564e-07,
+      "loss": 0.6298,
+      "step": 10310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.517189822120068e-07,
+      "loss": 0.7314,
+      "step": 10311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5135276294955474e-07,
+      "loss": 0.7112,
+      "step": 10312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.509867200342254e-07,
+      "loss": 0.6646,
+      "step": 10313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5062085349606116e-07,
+      "loss": 0.6601,
+      "step": 10314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5025516336509e-07,
+      "loss": 0.7087,
+      "step": 10315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.498896496713264e-07,
+      "loss": 0.623,
+      "step": 10316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4952431244476837e-07,
+      "loss": 0.6624,
+      "step": 10317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.491591517154e-07,
+      "loss": 0.7174,
+      "step": 10318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.487941675131928e-07,
+      "loss": 0.64,
+      "step": 10319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.484293598681016e-07,
+      "loss": 0.7513,
+      "step": 10320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4806472881006664e-07,
+      "loss": 0.7819,
+      "step": 10321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4770027436901584e-07,
+      "loss": 0.5973,
+      "step": 10322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4733599657486016e-07,
+      "loss": 0.6384,
+      "step": 10323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.469718954574985e-07,
+      "loss": 0.6782,
+      "step": 10324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4660797104681234e-07,
+      "loss": 0.7481,
+      "step": 10325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.462442233726715e-07,
+      "loss": 0.7787,
+      "step": 10326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4588065246493e-07,
+      "loss": 0.6193,
+      "step": 10327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.455172583534261e-07,
+      "loss": 0.6411,
+      "step": 10328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.451540410679863e-07,
+      "loss": 0.6781,
+      "step": 10329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4479100063842057e-07,
+      "loss": 0.6143,
+      "step": 10330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4442813709452516e-07,
+      "loss": 0.812,
+      "step": 10331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.440654504660806e-07,
+      "loss": 0.6382,
+      "step": 10332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.437029407828552e-07,
+      "loss": 0.7147,
+      "step": 10333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.433406080746007e-07,
+      "loss": 0.6893,
+      "step": 10334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4297845237105497e-07,
+      "loss": 0.7297,
+      "step": 10335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.426164737019419e-07,
+      "loss": 0.6841,
+      "step": 10336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4225467209697023e-07,
+      "loss": 0.6647,
+      "step": 10337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4189304758583397e-07,
+      "loss": 0.6256,
+      "step": 10338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4153160019821275e-07,
+      "loss": 0.6572,
+      "step": 10339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.411703299637725e-07,
+      "loss": 0.6808,
+      "step": 10340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.408092369121635e-07,
+      "loss": 0.6305,
+      "step": 10341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.404483210730225e-07,
+      "loss": 0.8015,
+      "step": 10342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.4008758247597003e-07,
+      "loss": 0.7593,
+      "step": 10343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3972702115061456e-07,
+      "loss": 0.5879,
+      "step": 10344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.39366637126548e-07,
+      "loss": 0.724,
+      "step": 10345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.390064304333479e-07,
+      "loss": 0.6596,
+      "step": 10346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3864640110057886e-07,
+      "loss": 0.6073,
+      "step": 10347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.382865491577886e-07,
+      "loss": 0.7368,
+      "step": 10348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.379268746345116e-07,
+      "loss": 0.7563,
+      "step": 10349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3756737756026866e-07,
+      "loss": 0.6809,
+      "step": 10350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3720805796456394e-07,
+      "loss": 0.8249,
+      "step": 10351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3684891587688867e-07,
+      "loss": 0.7179,
+      "step": 10352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.364899513267178e-07,
+      "loss": 0.6962,
+      "step": 10353
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.8011298179626465,
+      "eval_runtime": 438.0015,
+      "eval_samples_per_second": 6.847,
+      "eval_steps_per_second": 0.856,
+      "step": 10353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3613116434351444e-07,
+      "loss": 0.7417,
+      "step": 10354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3577255495672477e-07,
+      "loss": 0.5881,
+      "step": 10355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3541412319578027e-07,
+      "loss": 0.7006,
+      "step": 10356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.350558690901004e-07,
+      "loss": 0.872,
+      "step": 10357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.346977926690875e-07,
+      "loss": 0.7079,
+      "step": 10358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3433989396212953e-07,
+      "loss": 0.7485,
+      "step": 10359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3398217299860173e-07,
+      "loss": 0.7249,
+      "step": 10360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3362462980786244e-07,
+      "loss": 0.6014,
+      "step": 10361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.332672644192575e-07,
+      "loss": 0.8339,
+      "step": 10362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.32910076862116e-07,
+      "loss": 0.669,
+      "step": 10363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.325530671657551e-07,
+      "loss": 0.7069,
+      "step": 10364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.321962353594749e-07,
+      "loss": 0.7275,
+      "step": 10365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.318395814725614e-07,
+      "loss": 0.7391,
+      "step": 10366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3148310553428794e-07,
+      "loss": 0.8516,
+      "step": 10367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.311268075739105e-07,
+      "loss": 0.6461,
+      "step": 10368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3077068762067207e-07,
+      "loss": 0.7154,
+      "step": 10369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.304147457038004e-07,
+      "loss": 0.9167,
+      "step": 10370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.3005898185250964e-07,
+      "loss": 0.5987,
+      "step": 10371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2970339609599803e-07,
+      "loss": 0.6429,
+      "step": 10372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2934798846344955e-07,
+      "loss": 0.6548,
+      "step": 10373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.289927589840344e-07,
+      "loss": 0.7388,
+      "step": 10374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2863770768690767e-07,
+      "loss": 0.671,
+      "step": 10375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2828283460120904e-07,
+      "loss": 0.5985,
+      "step": 10376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.279281397560638e-07,
+      "loss": 0.6789,
+      "step": 10377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.27573623180584e-07,
+      "loss": 0.6195,
+      "step": 10378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.272192849038655e-07,
+      "loss": 0.6791,
+      "step": 10379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2686512495499064e-07,
+      "loss": 0.6634,
+      "step": 10380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.265111433630258e-07,
+      "loss": 0.6712,
+      "step": 10381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2615734015702476e-07,
+      "loss": 0.5457,
+      "step": 10382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2580371536602427e-07,
+      "loss": 0.6544,
+      "step": 10383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.254502690190475e-07,
+      "loss": 0.8118,
+      "step": 10384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2509700114510404e-07,
+      "loss": 0.6691,
+      "step": 10385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.247439117731874e-07,
+      "loss": 0.6161,
+      "step": 10386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.243910009322762e-07,
+      "loss": 0.6436,
+      "step": 10387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2403826865133604e-07,
+      "loss": 0.6794,
+      "step": 10388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2368571495931645e-07,
+      "loss": 0.6333,
+      "step": 10389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2333333988515294e-07,
+      "loss": 0.6152,
+      "step": 10390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2298114345776563e-07,
+      "loss": 0.6549,
+      "step": 10391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2262912570606155e-07,
+      "loss": 0.711,
+      "step": 10392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2227728665893133e-07,
+      "loss": 0.7303,
+      "step": 10393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.219256263452511e-07,
+      "loss": 0.6457,
+      "step": 10394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.215741447938844e-07,
+      "loss": 0.7433,
+      "step": 10395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.212228420336777e-07,
+      "loss": 0.6745,
+      "step": 10396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.208717180934631e-07,
+      "loss": 0.8277,
+      "step": 10397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2052077300205984e-07,
+      "loss": 0.5274,
+      "step": 10398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2017000678827015e-07,
+      "loss": 0.5987,
+      "step": 10399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1981941948088345e-07,
+      "loss": 0.6706,
+      "step": 10400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.194690111086729e-07,
+      "loss": 0.7269,
+      "step": 10401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1911878170039877e-07,
+      "loss": 0.6397,
+      "step": 10402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.187687312848051e-07,
+      "loss": 0.8522,
+      "step": 10403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.18418859890621e-07,
+      "loss": 0.6594,
+      "step": 10404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1806916754656303e-07,
+      "loss": 0.7168,
+      "step": 10405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.177196542813313e-07,
+      "loss": 0.6613,
+      "step": 10406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1737032012361116e-07,
+      "loss": 0.7127,
+      "step": 10407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1702116510207325e-07,
+      "loss": 0.7692,
+      "step": 10408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.166721892453756e-07,
+      "loss": 0.682,
+      "step": 10409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.163233925821585e-07,
+      "loss": 0.6533,
+      "step": 10410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1597477514104914e-07,
+      "loss": 0.8075,
+      "step": 10411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.156263369506604e-07,
+      "loss": 0.6848,
+      "step": 10412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.152780780395895e-07,
+      "loss": 0.7128,
+      "step": 10413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1492999843641903e-07,
+      "loss": 0.7277,
+      "step": 10414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.14582098169717e-07,
+      "loss": 0.7674,
+      "step": 10415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1423437726803783e-07,
+      "loss": 0.6772,
+      "step": 10416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.138868357599187e-07,
+      "loss": 0.6724,
+      "step": 10417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.135394736738853e-07,
+      "loss": 0.6716,
+      "step": 10418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1319229103844525e-07,
+      "loss": 0.7326,
+      "step": 10419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1284528788209424e-07,
+      "loss": 0.6988,
+      "step": 10420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1249846423331174e-07,
+      "loss": 0.5966,
+      "step": 10421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1215182012056226e-07,
+      "loss": 0.7859,
+      "step": 10422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.1180535557229693e-07,
+      "loss": 0.6515,
+      "step": 10423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.11459070616951e-07,
+      "loss": 0.8049,
+      "step": 10424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.111129652829448e-07,
+      "loss": 0.745,
+      "step": 10425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.107670395986853e-07,
+      "loss": 0.7108,
+      "step": 10426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.104212935925632e-07,
+      "loss": 0.8499,
+      "step": 10427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.100757272929558e-07,
+      "loss": 0.7056,
+      "step": 10428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0973034072822374e-07,
+      "loss": 0.7415,
+      "step": 10429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.093851339267151e-07,
+      "loss": 0.6126,
+      "step": 10430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0904010691676227e-07,
+      "loss": 0.7178,
+      "step": 10431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0869525972668207e-07,
+      "loss": 0.8177,
+      "step": 10432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.083505923847782e-07,
+      "loss": 0.6945,
+      "step": 10433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.080061049193386e-07,
+      "loss": 0.6683,
+      "step": 10434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.076617973586357e-07,
+      "loss": 0.7134,
+      "step": 10435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0731766973092934e-07,
+      "loss": 0.7174,
+      "step": 10436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0697372206446196e-07,
+      "loss": 0.6441,
+      "step": 10437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0662995438746403e-07,
+      "loss": 0.5883,
+      "step": 10438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.062863667281482e-07,
+      "loss": 0.6466,
+      "step": 10439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0594295911471535e-07,
+      "loss": 0.6274,
+      "step": 10440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.055997315753498e-07,
+      "loss": 0.6143,
+      "step": 10441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.052566841382204e-07,
+      "loss": 0.7577,
+      "step": 10442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.049138168314839e-07,
+      "loss": 0.7356,
+      "step": 10443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0457112968327956e-07,
+      "loss": 0.7372,
+      "step": 10444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0422862272173337e-07,
+      "loss": 0.5997,
+      "step": 10445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0388629597495526e-07,
+      "loss": 0.6467,
+      "step": 10446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0354414947104253e-07,
+      "loss": 0.6486,
+      "step": 10447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0320218323807576e-07,
+      "loss": 0.5555,
+      "step": 10448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0286039730412084e-07,
+      "loss": 0.6636,
+      "step": 10449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0251879169723024e-07,
+      "loss": 0.6078,
+      "step": 10450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.021773664454403e-07,
+      "loss": 0.6295,
+      "step": 10451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0183612157677286e-07,
+      "loss": 0.6823,
+      "step": 10452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.014950571192349e-07,
+      "loss": 0.7535,
+      "step": 10453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.011541731008197e-07,
+      "loss": 0.5922,
+      "step": 10454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.008134695495035e-07,
+      "loss": 0.7891,
+      "step": 10455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.004729464932507e-07,
+      "loss": 0.6326,
+      "step": 10456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.001326039600075e-07,
+      "loss": 0.8374,
+      "step": 10457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.997924419777087e-07,
+      "loss": 0.7316,
+      "step": 10458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.994524605742716e-07,
+      "loss": 0.7779,
+      "step": 10459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9911265977759953e-07,
+      "loss": 0.6437,
+      "step": 10460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.987730396155819e-07,
+      "loss": 0.6731,
+      "step": 10461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.984336001160923e-07,
+      "loss": 0.6893,
+      "step": 10462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.980943413069887e-07,
+      "loss": 0.7234,
+      "step": 10463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.977552632161171e-07,
+      "loss": 0.6653,
+      "step": 10464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9741636587130554e-07,
+      "loss": 0.718,
+      "step": 10465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.970776493003691e-07,
+      "loss": 0.6907,
+      "step": 10466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.967391135311068e-07,
+      "loss": 0.7311,
+      "step": 10467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9640075859130425e-07,
+      "loss": 0.6277,
+      "step": 10468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9606258450873113e-07,
+      "loss": 0.6623,
+      "step": 10469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.95724591311142e-07,
+      "loss": 0.7864,
+      "step": 10470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9538677902627847e-07,
+      "loss": 0.8117,
+      "step": 10471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.950491476818651e-07,
+      "loss": 0.6942,
+      "step": 10472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9471169730561217e-07,
+      "loss": 0.656,
+      "step": 10473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.943744279252167e-07,
+      "loss": 0.9172,
+      "step": 10474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9403733956835793e-07,
+      "loss": 0.7226,
+      "step": 10475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9370043226270353e-07,
+      "loss": 0.6861,
+      "step": 10476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.933637060359032e-07,
+      "loss": 0.7677,
+      "step": 10477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9302716091559464e-07,
+      "loss": 0.6236,
+      "step": 10478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.926907969293988e-07,
+      "loss": 0.6352,
+      "step": 10479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9235461410492134e-07,
+      "loss": 0.5734,
+      "step": 10480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.920186124697555e-07,
+      "loss": 0.6782,
+      "step": 10481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9168279205147725e-07,
+      "loss": 0.7178,
+      "step": 10482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9134715287764876e-07,
+      "loss": 0.7806,
+      "step": 10483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9101169497581654e-07,
+      "loss": 0.7316,
+      "step": 10484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.906764183735139e-07,
+      "loss": 0.9012,
+      "step": 10485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.903413230982577e-07,
+      "loss": 0.6618,
+      "step": 10486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.900064091775498e-07,
+      "loss": 0.6314,
+      "step": 10487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.896716766388788e-07,
+      "loss": 0.7151,
+      "step": 10488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8933712550971687e-07,
+      "loss": 0.7699,
+      "step": 10489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.890027558175218e-07,
+      "loss": 0.7304,
+      "step": 10490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.886685675897363e-07,
+      "loss": 0.6883,
+      "step": 10491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.883345608537888e-07,
+      "loss": 0.6892,
+      "step": 10492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.88000735637092e-07,
+      "loss": 0.8338,
+      "step": 10493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.876670919670449e-07,
+      "loss": 0.6777,
+      "step": 10494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.873336298710297e-07,
+      "loss": 0.7899,
+      "step": 10495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.870003493764159e-07,
+      "loss": 0.7198,
+      "step": 10496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8666725051055667e-07,
+      "loss": 0.6971,
+      "step": 10497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.863343333007901e-07,
+      "loss": 0.6661,
+      "step": 10498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8600159777444044e-07,
+      "loss": 0.6933,
+      "step": 10499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.856690439588164e-07,
+      "loss": 0.6863,
+      "step": 10500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8533667188121145e-07,
+      "loss": 0.6622,
+      "step": 10501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.850044815689054e-07,
+      "loss": 0.7745,
+      "step": 10502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.846724730491615e-07,
+      "loss": 0.7478,
+      "step": 10503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8434064634922933e-07,
+      "loss": 0.7241,
+      "step": 10504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.840090014963423e-07,
+      "loss": 0.7625,
+      "step": 10505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8367753851772096e-07,
+      "loss": 0.7302,
+      "step": 10506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8334625744056876e-07,
+      "loss": 0.7857,
+      "step": 10507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.830151582920748e-07,
+      "loss": 0.7493,
+      "step": 10508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8268424109941464e-07,
+      "loss": 0.7459,
+      "step": 10509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8235350588974754e-07,
+      "loss": 0.6283,
+      "step": 10510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8202295269021713e-07,
+      "loss": 0.5638,
+      "step": 10511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.816925815279542e-07,
+      "loss": 0.676,
+      "step": 10512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.813623924300729e-07,
+      "loss": 0.8233,
+      "step": 10513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.810323854236735e-07,
+      "loss": 0.6588,
+      "step": 10514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.807025605358402e-07,
+      "loss": 0.6958,
+      "step": 10515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8037291779364373e-07,
+      "loss": 0.6819,
+      "step": 10516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8004345722413837e-07,
+      "loss": 0.7414,
+      "step": 10517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.797141788543642e-07,
+      "loss": 0.6516,
+      "step": 10518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7938508271134676e-07,
+      "loss": 0.6423,
+      "step": 10519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.790561688220958e-07,
+      "loss": 0.6929,
+      "step": 10520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7872743721360644e-07,
+      "loss": 0.5751,
+      "step": 10521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7839888791285825e-07,
+      "loss": 0.8164,
+      "step": 10522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7807052094681765e-07,
+      "loss": 0.6974,
+      "step": 10523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.777423363424342e-07,
+      "loss": 0.6596,
+      "step": 10524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.77414334126643e-07,
+      "loss": 0.7815,
+      "step": 10525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.770865143263649e-07,
+      "loss": 0.6999,
+      "step": 10526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7675887696850504e-07,
+      "loss": 0.599,
+      "step": 10527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.764314220799536e-07,
+      "loss": 0.8922,
+      "step": 10528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7610414968758546e-07,
+      "loss": 0.6919,
+      "step": 10529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7577705981826244e-07,
+      "loss": 0.8404,
+      "step": 10530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7545015249882847e-07,
+      "loss": 0.6657,
+      "step": 10531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7512342775611543e-07,
+      "loss": 0.6874,
+      "step": 10532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.747968856169372e-07,
+      "loss": 0.7153,
+      "step": 10533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.744705261080957e-07,
+      "loss": 0.6868,
+      "step": 10534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.741443492563761e-07,
+      "loss": 0.7693,
+      "step": 10535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.738183550885479e-07,
+      "loss": 0.7158,
+      "step": 10536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.734925436313676e-07,
+      "loss": 0.7207,
+      "step": 10537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7316691491157567e-07,
+      "loss": 0.6243,
+      "step": 10538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.72841468955897e-07,
+      "loss": 0.6833,
+      "step": 10539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.725162057910427e-07,
+      "loss": 0.6334,
+      "step": 10540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.721911254437079e-07,
+      "loss": 0.5376,
+      "step": 10541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.718662279405734e-07,
+      "loss": 0.5987,
+      "step": 10542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7154151330830437e-07,
+      "loss": 0.6572,
+      "step": 10543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.712169815735516e-07,
+      "loss": 0.6574,
+      "step": 10544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7089263276295045e-07,
+      "loss": 0.6784,
+      "step": 10545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7056846690312084e-07,
+      "loss": 0.6571,
+      "step": 10546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7024448402066917e-07,
+      "loss": 0.6445,
+      "step": 10547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6992068414218556e-07,
+      "loss": 0.6166,
+      "step": 10548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6959706729424467e-07,
+      "loss": 0.7097,
+      "step": 10549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.692736335034082e-07,
+      "loss": 0.7724,
+      "step": 10550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6895038279622007e-07,
+      "loss": 0.6831,
+      "step": 10551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.686273151992119e-07,
+      "loss": 0.781,
+      "step": 10552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6830443073889825e-07,
+      "loss": 0.7741,
+      "step": 10553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.679817294417797e-07,
+      "loss": 0.7736,
+      "step": 10554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6765921133434153e-07,
+      "loss": 0.7671,
+      "step": 10555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.673368764430534e-07,
+      "loss": 0.6808,
+      "step": 10556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6701472479437136e-07,
+      "loss": 0.8219,
+      "step": 10557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.666927564147351e-07,
+      "loss": 0.824,
+      "step": 10558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6637097133056953e-07,
+      "loss": 0.6947,
+      "step": 10559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.660493695682845e-07,
+      "loss": 0.6755,
+      "step": 10560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.657279511542757e-07,
+      "loss": 0.753,
+      "step": 10561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6540671611492293e-07,
+      "loss": 0.6898,
+      "step": 10562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6508566447659033e-07,
+      "loss": 0.6842,
+      "step": 10563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.647647962656288e-07,
+      "loss": 0.7343,
+      "step": 10564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6444411150837265e-07,
+      "loss": 0.7246,
+      "step": 10565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6412361023114177e-07,
+      "loss": 0.6218,
+      "step": 10566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6380329246024017e-07,
+      "loss": 0.6677,
+      "step": 10567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6348315822195826e-07,
+      "loss": 0.6982,
+      "step": 10568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6316320754257026e-07,
+      "loss": 0.645,
+      "step": 10569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6284344044833573e-07,
+      "loss": 0.5645,
+      "step": 10570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6252385696549893e-07,
+      "loss": 0.7492,
+      "step": 10571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6220445712028977e-07,
+      "loss": 0.6703,
+      "step": 10572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.618852409389222e-07,
+      "loss": 0.7956,
+      "step": 10573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6156620844759507e-07,
+      "loss": 0.6964,
+      "step": 10574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.612473596724929e-07,
+      "loss": 0.707,
+      "step": 10575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6092869463978486e-07,
+      "loss": 0.7225,
+      "step": 10576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6061021337562415e-07,
+      "loss": 0.7046,
+      "step": 10577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6029191590615095e-07,
+      "loss": 0.7822,
+      "step": 10578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5997380225748827e-07,
+      "loss": 0.7006,
+      "step": 10579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5965587245574464e-07,
+      "loss": 0.7561,
+      "step": 10580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5933812652701365e-07,
+      "loss": 0.7059,
+      "step": 10581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5902056449737476e-07,
+      "loss": 0.689,
+      "step": 10582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.587031863928907e-07,
+      "loss": 0.8702,
+      "step": 10583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5838599223960925e-07,
+      "loss": 0.6404,
+      "step": 10584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.580689820635651e-07,
+      "loss": 0.6563,
+      "step": 10585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5775215589077555e-07,
+      "loss": 0.7779,
+      "step": 10586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5743551374724313e-07,
+      "loss": 0.7951,
+      "step": 10587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.57119055658957e-07,
+      "loss": 0.6748,
+      "step": 10588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5680278165188894e-07,
+      "loss": 0.7089,
+      "step": 10589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.564866917519976e-07,
+      "loss": 0.6398,
+      "step": 10590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5617078598522445e-07,
+      "loss": 0.6993,
+      "step": 10591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5585506437749845e-07,
+      "loss": 0.7108,
+      "step": 10592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5553952695473117e-07,
+      "loss": 0.7253,
+      "step": 10593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5522417374281934e-07,
+      "loss": 0.7976,
+      "step": 10594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.549090047676464e-07,
+      "loss": 0.6692,
+      "step": 10595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5459402005507866e-07,
+      "loss": 0.8989,
+      "step": 10596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.542792196309679e-07,
+      "loss": 0.9098,
+      "step": 10597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.53964603521151e-07,
+      "loss": 0.5937,
+      "step": 10598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5365017175145003e-07,
+      "loss": 0.7113,
+      "step": 10599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5333592434767103e-07,
+      "loss": 0.6436,
+      "step": 10600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.530218613356053e-07,
+      "loss": 0.8116,
+      "step": 10601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5270798274103003e-07,
+      "loss": 0.7108,
+      "step": 10602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.523942885897057e-07,
+      "loss": 0.7043,
+      "step": 10603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5208077890737815e-07,
+      "loss": 0.7407,
+      "step": 10604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.517674537197781e-07,
+      "loss": 0.6926,
+      "step": 10605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5145431305262205e-07,
+      "loss": 0.6712,
+      "step": 10606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5114135693160956e-07,
+      "loss": 0.6356,
+      "step": 10607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.508285853824269e-07,
+      "loss": 0.6876,
+      "step": 10608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.505159984307437e-07,
+      "loss": 0.6613,
+      "step": 10609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5020359610221615e-07,
+      "loss": 0.7804,
+      "step": 10610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4989137842248324e-07,
+      "loss": 0.648,
+      "step": 10611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4957934541716965e-07,
+      "loss": 0.6262,
+      "step": 10612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.492674971118858e-07,
+      "loss": 0.7745,
+      "step": 10613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.489558335322259e-07,
+      "loss": 0.811,
+      "step": 10614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.486443547037687e-07,
+      "loss": 0.6984,
+      "step": 10615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.483330606520795e-07,
+      "loss": 0.7033,
+      "step": 10616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4802195140270657e-07,
+      "loss": 0.7508,
+      "step": 10617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4771102698118386e-07,
+      "loss": 0.853,
+      "step": 10618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.474002874130296e-07,
+      "loss": 0.7822,
+      "step": 10619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4708973272374815e-07,
+      "loss": 0.6745,
+      "step": 10620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4677936293882766e-07,
+      "loss": 1.0072,
+      "step": 10621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4646917808374026e-07,
+      "loss": 0.7029,
+      "step": 10622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4615917818394534e-07,
+      "loss": 0.7525,
+      "step": 10623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.458493632648853e-07,
+      "loss": 0.6328,
+      "step": 10624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.455397333519868e-07,
+      "loss": 0.7934,
+      "step": 10625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4523028847066367e-07,
+      "loss": 0.693,
+      "step": 10626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.449210286463119e-07,
+      "loss": 0.5281,
+      "step": 10627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4461195390431456e-07,
+      "loss": 0.6553,
+      "step": 10628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.443030642700378e-07,
+      "loss": 0.7438,
+      "step": 10629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.439943597688338e-07,
+      "loss": 0.7593,
+      "step": 10630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.436858404260389e-07,
+      "loss": 0.6551,
+      "step": 10631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4337750626697436e-07,
+      "loss": 0.5947,
+      "step": 10632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.430693573169457e-07,
+      "loss": 0.7078,
+      "step": 10633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4276139360124464e-07,
+      "loss": 0.6553,
+      "step": 10634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4245361514514676e-07,
+      "loss": 0.6897,
+      "step": 10635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.421460219739116e-07,
+      "loss": 0.5524,
+      "step": 10636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.418386141127854e-07,
+      "loss": 0.6698,
+      "step": 10637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.415313915869982e-07,
+      "loss": 0.7467,
+      "step": 10638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.4122435442176407e-07,
+      "loss": 0.7242,
+      "step": 10639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.409175026422836e-07,
+      "loss": 0.7096,
+      "step": 10640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.406108362737408e-07,
+      "loss": 0.6856,
+      "step": 10641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.403043553413048e-07,
+      "loss": 0.7537,
+      "step": 10642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.39998059870129e-07,
+      "loss": 0.7661,
+      "step": 10643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.396919498853534e-07,
+      "loss": 0.6556,
+      "step": 10644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3938602541210054e-07,
+      "loss": 0.8445,
+      "step": 10645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3908028647547874e-07,
+      "loss": 0.8236,
+      "step": 10646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.387747331005821e-07,
+      "loss": 0.8511,
+      "step": 10647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3846936531248783e-07,
+      "loss": 0.755,
+      "step": 10648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3816418313625855e-07,
+      "loss": 0.6275,
+      "step": 10649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3785918659694096e-07,
+      "loss": 0.7357,
+      "step": 10650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3755437571956835e-07,
+      "loss": 0.6133,
+      "step": 10651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.372497505291574e-07,
+      "loss": 0.7794,
+      "step": 10652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.369453110507089e-07,
+      "loss": 0.7876,
+      "step": 10653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3664105730921039e-07,
+      "loss": 0.7169,
+      "step": 10654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3633698932963245e-07,
+      "loss": 0.7257,
+      "step": 10655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3603310713693124e-07,
+      "loss": 0.6166,
+      "step": 10656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3572941075604706e-07,
+      "loss": 0.6067,
+      "step": 10657
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3542590021190608e-07,
+      "loss": 0.6993,
+      "step": 10658
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3512257552941785e-07,
+      "loss": 0.7338,
+      "step": 10659
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3481943673347717e-07,
+      "loss": 0.6255,
+      "step": 10660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.345164838489647e-07,
+      "loss": 0.6645,
+      "step": 10661
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3421371690074414e-07,
+      "loss": 0.8456,
+      "step": 10662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.339111359136642e-07,
+      "loss": 0.6232,
+      "step": 10663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3360874091255952e-07,
+      "loss": 0.6559,
+      "step": 10664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3330653192224884e-07,
+      "loss": 0.6572,
+      "step": 10665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3300450896753535e-07,
+      "loss": 0.7019,
+      "step": 10666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3270267207320645e-07,
+      "loss": 0.6991,
+      "step": 10667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3240102126403596e-07,
+      "loss": 0.7946,
+      "step": 10668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.32099556564781e-07,
+      "loss": 0.7645,
+      "step": 10669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3179827800018402e-07,
+      "loss": 0.8532,
+      "step": 10670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.314971855949713e-07,
+      "loss": 0.676,
+      "step": 10671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3119627937385563e-07,
+      "loss": 0.6063,
+      "step": 10672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3089555936153308e-07,
+      "loss": 0.7592,
+      "step": 10673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3059502558268388e-07,
+      "loss": 0.7346,
+      "step": 10674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3029467806197553e-07,
+      "loss": 0.6796,
+      "step": 10675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2999451682405777e-07,
+      "loss": 0.7532,
+      "step": 10676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2969454189356533e-07,
+      "loss": 0.6309,
+      "step": 10677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2939475329511936e-07,
+      "loss": 0.6771,
+      "step": 10678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2909515105332403e-07,
+      "loss": 0.638,
+      "step": 10679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.287957351927686e-07,
+      "loss": 0.6837,
+      "step": 10680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.284965057380273e-07,
+      "loss": 0.6626,
+      "step": 10681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2819746271365906e-07,
+      "loss": 0.7275,
+      "step": 10682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2789860614420733e-07,
+      "loss": 0.615,
+      "step": 10683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.275999360542e-07,
+      "loss": 0.583,
+      "step": 10684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2730145246815106e-07,
+      "loss": 0.7599,
+      "step": 10685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2700315541055728e-07,
+      "loss": 0.6079,
+      "step": 10686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2670504490590133e-07,
+      "loss": 0.6582,
+      "step": 10687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.264071209786492e-07,
+      "loss": 0.9815,
+      "step": 10688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2610938365325409e-07,
+      "loss": 0.6519,
+      "step": 10689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2581183295415144e-07,
+      "loss": 0.652,
+      "step": 10690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.25514468905762e-07,
+      "loss": 0.6685,
+      "step": 10691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.252172915324924e-07,
+      "loss": 1.0281,
+      "step": 10692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2492030085873278e-07,
+      "loss": 0.7198,
+      "step": 10693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2462349690885787e-07,
+      "loss": 0.7273,
+      "step": 10694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2432687970722704e-07,
+      "loss": 0.6174,
+      "step": 10695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2403044927818612e-07,
+      "loss": 0.7052,
+      "step": 10696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2373420564606312e-07,
+      "loss": 0.6651,
+      "step": 10697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2343814883517168e-07,
+      "loss": 0.6532,
+      "step": 10698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2314227886981094e-07,
+      "loss": 0.7875,
+      "step": 10699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.228465957742637e-07,
+      "loss": 0.6629,
+      "step": 10700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2255109957279725e-07,
+      "loss": 0.7312,
+      "step": 10701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2225579028966466e-07,
+      "loss": 0.7463,
+      "step": 10702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2196066794910293e-07,
+      "loss": 0.7138,
+      "step": 10703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.216657325753338e-07,
+      "loss": 0.6959,
+      "step": 10704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.213709841925632e-07,
+      "loss": 0.5842,
+      "step": 10705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2107642282498315e-07,
+      "loss": 0.6607,
+      "step": 10706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2078204849676848e-07,
+      "loss": 0.6885,
+      "step": 10707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.204878612320799e-07,
+      "loss": 0.6749,
+      "step": 10708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2019386105506163e-07,
+      "loss": 0.6094,
+      "step": 10709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1990004798984444e-07,
+      "loss": 0.6308,
+      "step": 10710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1960642206054233e-07,
+      "loss": 0.7654,
+      "step": 10711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1931298329125357e-07,
+      "loss": 0.7291,
+      "step": 10712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1901973170606272e-07,
+      "loss": 0.8643,
+      "step": 10713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1872666732903723e-07,
+      "loss": 0.6698,
+      "step": 10714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1843379018422984e-07,
+      "loss": 0.7203,
+      "step": 10715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1814110029567875e-07,
+      "loss": 0.6808,
+      "step": 10716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.178485976874059e-07,
+      "loss": 0.8272,
+      "step": 10717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1755628238341736e-07,
+      "loss": 0.7919,
+      "step": 10718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1726415440770476e-07,
+      "loss": 0.7505,
+      "step": 10719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1697221378424422e-07,
+      "loss": 0.6463,
+      "step": 10720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1668046053699653e-07,
+      "loss": 0.7117,
+      "step": 10721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.163888946899065e-07,
+      "loss": 0.6737,
+      "step": 10722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.160975162669046e-07,
+      "loss": 0.8139,
+      "step": 10723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1580632529190487e-07,
+      "loss": 0.632,
+      "step": 10724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1551532178880645e-07,
+      "loss": 0.6921,
+      "step": 10725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.152245057814925e-07,
+      "loss": 0.7841,
+      "step": 10726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1493387729383251e-07,
+      "loss": 0.7264,
+      "step": 10727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1464343634967854e-07,
+      "loss": 0.8121,
+      "step": 10728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1435318297286788e-07,
+      "loss": 0.6566,
+      "step": 10729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1406311718722345e-07,
+      "loss": 0.6294,
+      "step": 10730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1377323901655146e-07,
+      "loss": 0.671,
+      "step": 10731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.134835484846437e-07,
+      "loss": 0.6376,
+      "step": 10732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1319404561527507e-07,
+      "loss": 0.6509,
+      "step": 10733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.129047304322074e-07,
+      "loss": 0.7203,
+      "step": 10734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1261560295918526e-07,
+      "loss": 0.6418,
+      "step": 10735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1232666321993778e-07,
+      "loss": 0.839,
+      "step": 10736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.120379112381804e-07,
+      "loss": 0.7395,
+      "step": 10737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1174934703761084e-07,
+      "loss": 0.7454,
+      "step": 10738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.114609706419138e-07,
+      "loss": 0.669,
+      "step": 10739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1117278207475612e-07,
+      "loss": 0.7659,
+      "step": 10740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1088478135979167e-07,
+      "loss": 0.7206,
+      "step": 10741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1059696852065713e-07,
+      "loss": 0.8255,
+      "step": 10742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1030934358097387e-07,
+      "loss": 0.6537,
+      "step": 10743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1002190656434912e-07,
+      "loss": 0.6482,
+      "step": 10744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.097346574943737e-07,
+      "loss": 0.6623,
+      "step": 10745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.094475963946227e-07,
+      "loss": 0.5879,
+      "step": 10746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.091607232886561e-07,
+      "loss": 0.742,
+      "step": 10747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0887403820001955e-07,
+      "loss": 0.6379,
+      "step": 10748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0858754115224146e-07,
+      "loss": 0.6796,
+      "step": 10749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0830123216883574e-07,
+      "loss": 0.67,
+      "step": 10750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0801511127330143e-07,
+      "loss": 0.6787,
+      "step": 10751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.077291784891211e-07,
+      "loss": 0.7483,
+      "step": 10752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0744343383976156e-07,
+      "loss": 0.7931,
+      "step": 10753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0715787734867625e-07,
+      "loss": 0.6152,
+      "step": 10754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.068725090393009e-07,
+      "loss": 0.8211,
+      "step": 10755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0658732893505674e-07,
+      "loss": 0.8148,
+      "step": 10756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0630233705934977e-07,
+      "loss": 0.6869,
+      "step": 10757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.060175334355699e-07,
+      "loss": 0.6637,
+      "step": 10758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0573291808709288e-07,
+      "loss": 0.71,
+      "step": 10759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0544849103727698e-07,
+      "loss": 0.6626,
+      "step": 10760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0516425230946712e-07,
+      "loss": 0.7118,
+      "step": 10761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0488020192699133e-07,
+      "loss": 0.7719,
+      "step": 10762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0459633991316258e-07,
+      "loss": 0.7667,
+      "step": 10763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.043126662912781e-07,
+      "loss": 0.7078,
+      "step": 10764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0402918108462095e-07,
+      "loss": 0.7375,
+      "step": 10765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.037458843164572e-07,
+      "loss": 0.639,
+      "step": 10766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0346277601003745e-07,
+      "loss": 0.7143,
+      "step": 10767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0317985618859837e-07,
+      "loss": 0.686,
+      "step": 10768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0289712487536e-07,
+      "loss": 0.7595,
+      "step": 10769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0261458209352685e-07,
+      "loss": 0.692,
+      "step": 10770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0233222786628788e-07,
+      "loss": 0.6825,
+      "step": 10771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0205006221681756e-07,
+      "loss": 0.6711,
+      "step": 10772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0176808516827405e-07,
+      "loss": 0.7165,
+      "step": 10773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0148629674379972e-07,
+      "loss": 0.7204,
+      "step": 10774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0120469696652295e-07,
+      "loss": 0.6787,
+      "step": 10775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0092328585955446e-07,
+      "loss": 0.7754,
+      "step": 10776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.006420634459916e-07,
+      "loss": 0.6956,
+      "step": 10777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0036102974891474e-07,
+      "loss": 0.6269,
+      "step": 10778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.000801847913897e-07,
+      "loss": 0.6088,
+      "step": 10779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9979952859646656e-07,
+      "loss": 0.6315,
+      "step": 10780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9951906118717862e-07,
+      "loss": 0.6602,
+      "step": 10781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9923878258654634e-07,
+      "loss": 0.6775,
+      "step": 10782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.989586928175727e-07,
+      "loss": 0.7895,
+      "step": 10783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9867879190324514e-07,
+      "loss": 0.6977,
+      "step": 10784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9839907986653612e-07,
+      "loss": 0.6004,
+      "step": 10785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.981195567304034e-07,
+      "loss": 0.8931,
+      "step": 10786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9784022251778805e-07,
+      "loss": 0.6855,
+      "step": 10787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9756107725161566e-07,
+      "loss": 0.666,
+      "step": 10788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9728212095479705e-07,
+      "loss": 0.6895,
+      "step": 10789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9700335365022721e-07,
+      "loss": 0.6594,
+      "step": 10790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.967247753607851e-07,
+      "loss": 0.6471,
+      "step": 10791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9644638610933547e-07,
+      "loss": 0.6591,
+      "step": 10792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.961681859187259e-07,
+      "loss": 0.5958,
+      "step": 10793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9589017481178945e-07,
+      "loss": 0.7012,
+      "step": 10794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9561235281134377e-07,
+      "loss": 0.7821,
+      "step": 10795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9533471994019033e-07,
+      "loss": 0.6337,
+      "step": 10796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9505727622111585e-07,
+      "loss": 0.7346,
+      "step": 10797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9478002167689081e-07,
+      "loss": 0.5948,
+      "step": 10798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9450295633027083e-07,
+      "loss": 0.7028,
+      "step": 10799
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9422608020399553e-07,
+      "loss": 0.5373,
+      "step": 10800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9394939332078893e-07,
+      "loss": 0.7506,
+      "step": 10801
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9367289570335928e-07,
+      "loss": 0.7218,
+      "step": 10802
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9339658737440064e-07,
+      "loss": 0.7155,
+      "step": 10803
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9312046835659042e-07,
+      "loss": 0.6952,
+      "step": 10804
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9284453867258968e-07,
+      "loss": 0.6285,
+      "step": 10805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9256879834504638e-07,
+      "loss": 0.753,
+      "step": 10806
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9229324739659105e-07,
+      "loss": 0.7186,
+      "step": 10807
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9201788584983866e-07,
+      "loss": 0.7392,
+      "step": 10808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9174271372738916e-07,
+      "loss": 0.8058,
+      "step": 10809
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.914677310518276e-07,
+      "loss": 0.6865,
+      "step": 10810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9119293784572224e-07,
+      "loss": 0.8042,
+      "step": 10811
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.909183341316259e-07,
+      "loss": 0.6865,
+      "step": 10812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9064391993207754e-07,
+      "loss": 0.7176,
+      "step": 10813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9036969526959797e-07,
+      "loss": 0.6494,
+      "step": 10814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9009566016669478e-07,
+      "loss": 0.7469,
+      "step": 10815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.898218146458583e-07,
+      "loss": 0.6694,
+      "step": 10816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.895481587295647e-07,
+      "loss": 0.7104,
+      "step": 10817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8927469244027386e-07,
+      "loss": 0.7622,
+      "step": 10818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8900141580042886e-07,
+      "loss": 0.7441,
+      "step": 10819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887283288324601e-07,
+      "loss": 0.6513,
+      "step": 10820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8845543155878025e-07,
+      "loss": 0.7442,
+      "step": 10821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8818272400178667e-07,
+      "loss": 0.6089,
+      "step": 10822
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8791020618386142e-07,
+      "loss": 0.6005,
+      "step": 10823
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8763787812737132e-07,
+      "loss": 0.6927,
+      "step": 10824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8736573985466717e-07,
+      "loss": 0.6988,
+      "step": 10825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8709379138808408e-07,
+      "loss": 0.6933,
+      "step": 10826
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.868220327499426e-07,
+      "loss": 0.616,
+      "step": 10827
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8655046396254622e-07,
+      "loss": 0.7436,
+      "step": 10828
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8627908504818327e-07,
+      "loss": 0.6299,
+      "step": 10829
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8600789602912782e-07,
+      "loss": 0.6795,
+      "step": 10830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8573689692763659e-07,
+      "loss": 0.6936,
+      "step": 10831
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8546608776595143e-07,
+      "loss": 0.7115,
+      "step": 10832
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8519546856629906e-07,
+      "loss": 0.5951,
+      "step": 10833
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8492503935088947e-07,
+      "loss": 0.8312,
+      "step": 10834
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.846548001419185e-07,
+      "loss": 0.6406,
+      "step": 10835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.843847509615651e-07,
+      "loss": 0.6042,
+      "step": 10836
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8411489183199349e-07,
+      "loss": 0.6653,
+      "step": 10837
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8384522277535204e-07,
+      "loss": 0.7412,
+      "step": 10838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8357574381377335e-07,
+      "loss": 0.7869,
+      "step": 10839
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8330645496937388e-07,
+      "loss": 0.6439,
+      "step": 10840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8303735626425622e-07,
+      "loss": 0.7409,
+      "step": 10841
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8276844772050578e-07,
+      "loss": 0.6386,
+      "step": 10842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.824997293601921e-07,
+      "loss": 0.6452,
+      "step": 10843
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8223120120537119e-07,
+      "loss": 0.6593,
+      "step": 10844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8196286327808121e-07,
+      "loss": 0.6806,
+      "step": 10845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8169471560034597e-07,
+      "loss": 0.6401,
+      "step": 10846
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8142675819417282e-07,
+      "loss": 0.6834,
+      "step": 10847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8115899108155448e-07,
+      "loss": 0.621,
+      "step": 10848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8089141428446783e-07,
+      "loss": 0.6047,
+      "step": 10849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8062402782487277e-07,
+      "loss": 0.7141,
+      "step": 10850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8035683172471562e-07,
+      "loss": 0.7174,
+      "step": 10851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8008982600592555e-07,
+      "loss": 0.7043,
+      "step": 10852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7982301069041724e-07,
+      "loss": 0.6643,
+      "step": 10853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7955638580008844e-07,
+      "loss": 0.7337,
+      "step": 10854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7928995135682303e-07,
+      "loss": 0.5828,
+      "step": 10855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7902370738248741e-07,
+      "loss": 0.6193,
+      "step": 10856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7875765389893302e-07,
+      "loss": 0.6477,
+      "step": 10857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.784917909279968e-07,
+      "loss": 0.605,
+      "step": 10858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7822611849149823e-07,
+      "loss": 0.7011,
+      "step": 10859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7796063661124242e-07,
+      "loss": 0.6755,
+      "step": 10860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7769534530901743e-07,
+      "loss": 0.6425,
+      "step": 10861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7743024460659813e-07,
+      "loss": 0.5986,
+      "step": 10862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7716533452574152e-07,
+      "loss": 0.6446,
+      "step": 10863
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.769006150881894e-07,
+      "loss": 0.7131,
+      "step": 10864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7663608631566908e-07,
+      "loss": 0.716,
+      "step": 10865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7637174822989072e-07,
+      "loss": 0.5842,
+      "step": 10866
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7610760085254947e-07,
+      "loss": 0.7625,
+      "step": 10867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7584364420532518e-07,
+      "loss": 0.7941,
+      "step": 10868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7557987830988194e-07,
+      "loss": 0.6782,
+      "step": 10869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7531630318786685e-07,
+      "loss": 0.6202,
+      "step": 10870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7505291886091375e-07,
+      "loss": 0.7015,
+      "step": 10871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7478972535063866e-07,
+      "loss": 0.7173,
+      "step": 10872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7452672267864347e-07,
+      "loss": 0.7782,
+      "step": 10873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.742639108665131e-07,
+      "loss": 0.6006,
+      "step": 10874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7400128993581816e-07,
+      "loss": 0.5816,
+      "step": 10875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7373885990811245e-07,
+      "loss": 0.6448,
+      "step": 10876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7347662080493433e-07,
+      "loss": 0.7887,
+      "step": 10877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7321457264780684e-07,
+      "loss": 0.6975,
+      "step": 10878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.729527154582375e-07,
+      "loss": 0.7494,
+      "step": 10879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7269104925771773e-07,
+      "loss": 0.6443,
+      "step": 10880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7242957406772315e-07,
+      "loss": 0.6606,
+      "step": 10881
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7216828990971434e-07,
+      "loss": 0.7717,
+      "step": 10882
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7190719680513558e-07,
+      "loss": 0.7139,
+      "step": 10883
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.716462947754158e-07,
+      "loss": 0.6689,
+      "step": 10884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7138558384196784e-07,
+      "loss": 0.6526,
+      "step": 10885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7112506402618966e-07,
+      "loss": 0.5715,
+      "step": 10886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.70864735349463e-07,
+      "loss": 0.8311,
+      "step": 10887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.706045978331533e-07,
+      "loss": 0.6611,
+      "step": 10888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7034465149861178e-07,
+      "loss": 0.6145,
+      "step": 10889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.700848963671728e-07,
+      "loss": 0.6749,
+      "step": 10890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6982533246015537e-07,
+      "loss": 0.6055,
+      "step": 10891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6956595979886275e-07,
+      "loss": 0.6947,
+      "step": 10892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.693067784045832e-07,
+      "loss": 0.6678,
+      "step": 10893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6904778829858803e-07,
+      "loss": 0.8113,
+      "step": 10894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6878898950213328e-07,
+      "loss": 0.6914,
+      "step": 10895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6853038203646005e-07,
+      "loss": 0.6257,
+      "step": 10896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6827196592279326e-07,
+      "loss": 0.6869,
+      "step": 10897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6801374118234155e-07,
+      "loss": 0.8521,
+      "step": 10898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.677557078362982e-07,
+      "loss": 0.7207,
+      "step": 10899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6749786590584154e-07,
+      "loss": 0.7621,
+      "step": 10900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.672402154121336e-07,
+      "loss": 0.6433,
+      "step": 10901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.669827563763196e-07,
+      "loss": 0.7079,
+      "step": 10902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6672548881953133e-07,
+      "loss": 0.7655,
+      "step": 10903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.664684127628835e-07,
+      "loss": 0.8102,
+      "step": 10904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6621152822747426e-07,
+      "loss": 0.6677,
+      "step": 10905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6595483523438817e-07,
+      "loss": 0.7212,
+      "step": 10906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6569833380469276e-07,
+      "loss": 0.7261,
+      "step": 10907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6544202395943926e-07,
+      "loss": 0.6958,
+      "step": 10908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6518590571966474e-07,
+      "loss": 0.6371,
+      "step": 10909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6492997910638907e-07,
+      "loss": 0.6292,
+      "step": 10910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6467424414061765e-07,
+      "loss": 0.5644,
+      "step": 10911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64418700843339e-07,
+      "loss": 0.6874,
+      "step": 10912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641633492355274e-07,
+      "loss": 0.724,
+      "step": 10913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6390818933813975e-07,
+      "loss": 0.8646,
+      "step": 10914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6365322117211786e-07,
+      "loss": 0.6397,
+      "step": 10915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.633984447583878e-07,
+      "loss": 0.7136,
+      "step": 10916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.631438601178606e-07,
+      "loss": 0.7639,
+      "step": 10917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6288946727143067e-07,
+      "loss": 0.6081,
+      "step": 10918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6263526623997633e-07,
+      "loss": 0.6697,
+      "step": 10919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.623812570443617e-07,
+      "loss": 0.5608,
+      "step": 10920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6212743970543372e-07,
+      "loss": 0.6396,
+      "step": 10921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6187381424402405e-07,
+      "loss": 0.7105,
+      "step": 10922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6162038068094854e-07,
+      "loss": 0.7628,
+      "step": 10923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.613671390370078e-07,
+      "loss": 0.6605,
+      "step": 10924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.61114089332986e-07,
+      "loss": 0.5701,
+      "step": 10925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6086123158965155e-07,
+      "loss": 0.6204,
+      "step": 10926
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6060856582775815e-07,
+      "loss": 0.8038,
+      "step": 10927
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.603560920680422e-07,
+      "loss": 0.6634,
+      "step": 10928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6010381033122553e-07,
+      "loss": 0.7332,
+      "step": 10929
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5985172063801347e-07,
+      "loss": 0.7177,
+      "step": 10930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.595998230090967e-07,
+      "loss": 0.6441,
+      "step": 10931
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5934811746514873e-07,
+      "loss": 0.7353,
+      "step": 10932
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5909660402682768e-07,
+      "loss": 0.7255,
+      "step": 10933
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5884528271477683e-07,
+      "loss": 0.7149,
+      "step": 10934
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5859415354962244e-07,
+      "loss": 0.7752,
+      "step": 10935
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5834321655197605e-07,
+      "loss": 0.7727,
+      "step": 10936
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5809247174243232e-07,
+      "loss": 0.6365,
+      "step": 10937
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.578419191415717e-07,
+      "loss": 0.8912,
+      "step": 10938
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5759155876995725e-07,
+      "loss": 0.6437,
+      "step": 10939
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5734139064813663e-07,
+      "loss": 0.6507,
+      "step": 10940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.570914147966429e-07,
+      "loss": 0.7081,
+      "step": 10941
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5684163123599188e-07,
+      "loss": 0.6502,
+      "step": 10942
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5659203998668437e-07,
+      "loss": 0.763,
+      "step": 10943
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5634264106920538e-07,
+      "loss": 0.6204,
+      "step": 10944
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.560934345040238e-07,
+      "loss": 0.7894,
+      "step": 10945
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5584442031159247e-07,
+      "loss": 0.6165,
+      "step": 10946
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5559559851234946e-07,
+      "loss": 0.6066,
+      "step": 10947
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.553469691267162e-07,
+      "loss": 0.5445,
+      "step": 10948
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5509853217509892e-07,
+      "loss": 0.7506,
+      "step": 10949
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5485028767788735e-07,
+      "loss": 0.7493,
+      "step": 10950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5460223565545607e-07,
+      "loss": 0.8203,
+      "step": 10951
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5435437612816352e-07,
+      "loss": 0.6677,
+      "step": 10952
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5410670911635228e-07,
+      "loss": 0.8858,
+      "step": 10953
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5385923464034918e-07,
+      "loss": 0.7812,
+      "step": 10954
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5361195272046547e-07,
+      "loss": 0.7117,
+      "step": 10955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5336486337699684e-07,
+      "loss": 0.7354,
+      "step": 10956
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5311796663022183e-07,
+      "loss": 0.7157,
+      "step": 10957
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.528712625004053e-07,
+      "loss": 0.6939,
+      "step": 10958
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.526247510077944e-07,
+      "loss": 0.7108,
+      "step": 10959
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5237843217262126e-07,
+      "loss": 0.6948,
+      "step": 10960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.52132306015102e-07,
+      "loss": 0.6173,
+      "step": 10961
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5188637255543787e-07,
+      "loss": 0.664,
+      "step": 10962
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.8011065721511841,
+      "eval_runtime": 437.207,
+      "eval_samples_per_second": 6.859,
+      "eval_steps_per_second": 0.858,
+      "step": 10962
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5164063181381278e-07,
+      "loss": 0.5444,
+      "step": 10963
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.513950838103953e-07,
+      "loss": 0.7204,
+      "step": 10964
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5114972856533906e-07,
+      "loss": 0.669,
+      "step": 10965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5090456609878096e-07,
+      "loss": 0.7293,
+      "step": 10966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5065959643084272e-07,
+      "loss": 0.6284,
+      "step": 10967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5041481958162906e-07,
+      "loss": 0.763,
+      "step": 10968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.501702355712306e-07,
+      "loss": 0.7142,
+      "step": 10969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4992584441972103e-07,
+      "loss": 0.6195,
+      "step": 10970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4968164614715756e-07,
+      "loss": 0.5892,
+      "step": 10971
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4943764077358369e-07,
+      "loss": 0.6894,
+      "step": 10972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4919382831902502e-07,
+      "loss": 0.7675,
+      "step": 10973
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.489502088034922e-07,
+      "loss": 0.6877,
+      "step": 10974
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4870678224697955e-07,
+      "loss": 0.6272,
+      "step": 10975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.484635486694669e-07,
+      "loss": 0.7113,
+      "step": 10976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.482205080909166e-07,
+      "loss": 0.5954,
+      "step": 10977
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4797766053127605e-07,
+      "loss": 0.6579,
+      "step": 10978
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4773500601047651e-07,
+      "loss": 0.6467,
+      "step": 10979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4749254454843403e-07,
+      "loss": 0.8365,
+      "step": 10980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4725027616504738e-07,
+      "loss": 0.6903,
+      "step": 10981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4700820088020096e-07,
+      "loss": 0.6253,
+      "step": 10982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.46766318713763e-07,
+      "loss": 0.6985,
+      "step": 10983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4652462968558463e-07,
+      "loss": 0.8074,
+      "step": 10984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4628313381550352e-07,
+      "loss": 0.8111,
+      "step": 10985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4604183112333859e-07,
+      "loss": 0.6853,
+      "step": 10986
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.458007216288959e-07,
+      "loss": 0.5727,
+      "step": 10987
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4555980535196273e-07,
+      "loss": 0.7102,
+      "step": 10988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4531908231231324e-07,
+      "loss": 0.7501,
+      "step": 10989
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4507855252970382e-07,
+      "loss": 0.8171,
+      "step": 10990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4483821602387567e-07,
+      "loss": 0.7392,
+      "step": 10991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4459807281455353e-07,
+      "loss": 0.7009,
+      "step": 10992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4435812292144746e-07,
+      "loss": 0.7386,
+      "step": 10993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.441183663642512e-07,
+      "loss": 0.7302,
+      "step": 10994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4387880316264146e-07,
+      "loss": 0.6221,
+      "step": 10995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4363943333628117e-07,
+      "loss": 0.6365,
+      "step": 10996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4340025690481568e-07,
+      "loss": 0.8542,
+      "step": 10997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4316127388787516e-07,
+      "loss": 0.6264,
+      "step": 10998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4292248430507305e-07,
+      "loss": 0.7075,
+      "step": 10999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4268388817600897e-07,
+      "loss": 0.7166,
+      "step": 11000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4244548552026478e-07,
+      "loss": 0.5745,
+      "step": 11001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.422072763574067e-07,
+      "loss": 0.6068,
+      "step": 11002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4196926070698585e-07,
+      "loss": 0.7995,
+      "step": 11003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4173143858853656e-07,
+      "loss": 0.5813,
+      "step": 11004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4149381002157846e-07,
+      "loss": 0.7386,
+      "step": 11005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4125637502561352e-07,
+      "loss": 0.629,
+      "step": 11006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4101913362013024e-07,
+      "loss": 0.655,
+      "step": 11007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4078208582459895e-07,
+      "loss": 0.7441,
+      "step": 11008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.405452316584749e-07,
+      "loss": 0.6362,
+      "step": 11009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4030857114119806e-07,
+      "loss": 0.7382,
+      "step": 11010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4007210429219182e-07,
+      "loss": 0.7319,
+      "step": 11011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3983583113086397e-07,
+      "loss": 0.649,
+      "step": 11012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3959975167660566e-07,
+      "loss": 0.6791,
+      "step": 11013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3936386594879365e-07,
+      "loss": 0.6365,
+      "step": 11014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.391281739667874e-07,
+      "loss": 0.695,
+      "step": 11015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3889267574993093e-07,
+      "loss": 0.6438,
+      "step": 11016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.386573713175529e-07,
+      "loss": 0.6953,
+      "step": 11017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3842226068896515e-07,
+      "loss": 0.7564,
+      "step": 11018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3818734388346383e-07,
+      "loss": 0.679,
+      "step": 11019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3795262092033024e-07,
+      "loss": 0.76,
+      "step": 11020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.377180918188284e-07,
+      "loss": 0.6376,
+      "step": 11021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3748375659820651e-07,
+      "loss": 0.781,
+      "step": 11022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3724961527769808e-07,
+      "loss": 0.6483,
+      "step": 11023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3701566787651943e-07,
+      "loss": 0.8057,
+      "step": 11024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3678191441387183e-07,
+      "loss": 0.6903,
+      "step": 11025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3654835490893996e-07,
+      "loss": 0.7039,
+      "step": 11026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3631498938089293e-07,
+      "loss": 0.7092,
+      "step": 11027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.360818178488843e-07,
+      "loss": 0.7043,
+      "step": 11028
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.358488403320507e-07,
+      "loss": 0.6515,
+      "step": 11029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3561605684951352e-07,
+      "loss": 0.7053,
+      "step": 11030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3538346742037828e-07,
+      "loss": 0.6461,
+      "step": 11031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3515107206373445e-07,
+      "loss": 0.6436,
+      "step": 11032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.349188707986554e-07,
+      "loss": 0.7298,
+      "step": 11033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3468686364419892e-07,
+      "loss": 0.6452,
+      "step": 11034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3445505061940644e-07,
+      "loss": 0.6811,
+      "step": 11035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.342234317433039e-07,
+      "loss": 0.5871,
+      "step": 11036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3399200703490072e-07,
+      "loss": 0.6562,
+      "step": 11037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3376077651319125e-07,
+      "loss": 0.6649,
+      "step": 11038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3352974019715305e-07,
+      "loss": 0.652,
+      "step": 11039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3329889810574786e-07,
+      "loss": 0.7507,
+      "step": 11040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3306825025792253e-07,
+      "loss": 0.7252,
+      "step": 11041
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3283779667260604e-07,
+      "loss": 0.8113,
+      "step": 11042
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3260753736871385e-07,
+      "loss": 0.7279,
+      "step": 11043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3237747236514303e-07,
+      "loss": 0.8556,
+      "step": 11044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3214760168077627e-07,
+      "loss": 0.6766,
+      "step": 11045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3191792533448018e-07,
+      "loss": 0.6154,
+      "step": 11046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3168844334510462e-07,
+      "loss": 0.6556,
+      "step": 11047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3145915573148428e-07,
+      "loss": 0.6634,
+      "step": 11048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3123006251243741e-07,
+      "loss": 0.698,
+      "step": 11049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.310011637067668e-07,
+      "loss": 0.6603,
+      "step": 11050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3077245933325823e-07,
+      "loss": 0.6723,
+      "step": 11051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3054394941068332e-07,
+      "loss": 0.7309,
+      "step": 11052
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3031563395779628e-07,
+      "loss": 0.61,
+      "step": 11053
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3008751299333544e-07,
+      "loss": 0.6147,
+      "step": 11054
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2985958653602387e-07,
+      "loss": 0.8511,
+      "step": 11055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.296318546045683e-07,
+      "loss": 0.6089,
+      "step": 11056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.29404317217659e-07,
+      "loss": 0.6222,
+      "step": 11057
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2917697439397165e-07,
+      "loss": 0.7688,
+      "step": 11058
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.289498261521646e-07,
+      "loss": 0.7012,
+      "step": 11059
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.287228725108805e-07,
+      "loss": 0.7607,
+      "step": 11060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2849611348874664e-07,
+      "loss": 0.6934,
+      "step": 11061
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2826954910437339e-07,
+      "loss": 0.7494,
+      "step": 11062
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2804317937635648e-07,
+      "loss": 0.6168,
+      "step": 11063
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2781700432327404e-07,
+      "loss": 0.5972,
+      "step": 11064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.275910239636899e-07,
+      "loss": 0.6153,
+      "step": 11065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2736523831615054e-07,
+      "loss": 0.651,
+      "step": 11066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2713964739918733e-07,
+      "loss": 0.7529,
+      "step": 11067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2691425123131456e-07,
+      "loss": 0.6827,
+      "step": 11068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2668904983103246e-07,
+      "loss": 0.6324,
+      "step": 11069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.264640432168232e-07,
+      "loss": 0.6898,
+      "step": 11070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2623923140715393e-07,
+      "loss": 0.622,
+      "step": 11071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.260146144204766e-07,
+      "loss": 0.7692,
+      "step": 11072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2579019227522528e-07,
+      "loss": 0.746,
+      "step": 11073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.255659649898197e-07,
+      "loss": 0.5858,
+      "step": 11074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2534193258266243e-07,
+      "loss": 0.6357,
+      "step": 11075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2511809507214145e-07,
+      "loss": 0.6125,
+      "step": 11076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.248944524766274e-07,
+      "loss": 0.7074,
+      "step": 11077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2467100481447502e-07,
+      "loss": 0.6323,
+      "step": 11078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2444775210402438e-07,
+      "loss": 0.7312,
+      "step": 11079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2422469436359775e-07,
+      "loss": 0.6856,
+      "step": 11080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2400183161150326e-07,
+      "loss": 0.6705,
+      "step": 11081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2377916386603073e-07,
+      "loss": 0.6812,
+      "step": 11082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2355669114545664e-07,
+      "loss": 0.6162,
+      "step": 11083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.233344134680392e-07,
+      "loss": 0.6965,
+      "step": 11084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2311233085202184e-07,
+      "loss": 0.7642,
+      "step": 11085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2289044331563166e-07,
+      "loss": 0.7151,
+      "step": 11086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.226687508770802e-07,
+      "loss": 0.6496,
+      "step": 11087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2244725355456176e-07,
+      "loss": 0.7186,
+      "step": 11088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.222259513662555e-07,
+      "loss": 0.8048,
+      "step": 11089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2200484433032512e-07,
+      "loss": 0.7187,
+      "step": 11090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2178393246491727e-07,
+      "loss": 0.7952,
+      "step": 11091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2156321578816244e-07,
+      "loss": 0.6538,
+      "step": 11092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2134269431817668e-07,
+      "loss": 0.721,
+      "step": 11093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.211223680730586e-07,
+      "loss": 0.8175,
+      "step": 11094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2090223707089067e-07,
+      "loss": 0.7253,
+      "step": 11095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2068230132974035e-07,
+      "loss": 0.6233,
+      "step": 11096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.204625608676585e-07,
+      "loss": 0.6809,
+      "step": 11097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2024301570267932e-07,
+      "loss": 0.6853,
+      "step": 11098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2002366585282283e-07,
+      "loss": 0.6471,
+      "step": 11099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1980451133609072e-07,
+      "loss": 0.6399,
+      "step": 11100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1958555217047085e-07,
+      "loss": 0.6412,
+      "step": 11101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.19366788373933e-07,
+      "loss": 0.6817,
+      "step": 11102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1914821996443254e-07,
+      "loss": 0.6506,
+      "step": 11103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1892984695990816e-07,
+      "loss": 0.7032,
+      "step": 11104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1871166937828222e-07,
+      "loss": 0.7637,
+      "step": 11105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1849368723746091e-07,
+      "loss": 0.6614,
+      "step": 11106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1827590055533578e-07,
+      "loss": 0.5699,
+      "step": 11107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1805830934978086e-07,
+      "loss": 0.7512,
+      "step": 11108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1784091363865435e-07,
+      "loss": 0.644,
+      "step": 11109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1762371343979923e-07,
+      "loss": 0.6443,
+      "step": 11110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1740670877104177e-07,
+      "loss": 0.6723,
+      "step": 11111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1718989965019218e-07,
+      "loss": 0.6514,
+      "step": 11112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1697328609504427e-07,
+      "loss": 0.8428,
+      "step": 11113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1675686812337717e-07,
+      "loss": 0.6889,
+      "step": 11114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1654064575295277e-07,
+      "loss": 0.6347,
+      "step": 11115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1632461900151688e-07,
+      "loss": 0.7794,
+      "step": 11116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1610878788680003e-07,
+      "loss": 0.6243,
+      "step": 11117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1589315242651555e-07,
+      "loss": 0.6785,
+      "step": 11118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1567771263836263e-07,
+      "loss": 0.6965,
+      "step": 11119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1546246854002185e-07,
+      "loss": 0.6224,
+      "step": 11120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1524742014915991e-07,
+      "loss": 0.7013,
+      "step": 11121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1503256748342656e-07,
+      "loss": 0.7902,
+      "step": 11122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1481791056045493e-07,
+      "loss": 0.7493,
+      "step": 11123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1460344939786344e-07,
+      "loss": 0.6732,
+      "step": 11124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1438918401325327e-07,
+      "loss": 0.6208,
+      "step": 11125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1417511442421009e-07,
+      "loss": 0.6987,
+      "step": 11126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.139612406483026e-07,
+      "loss": 0.7522,
+      "step": 11127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.137475627030854e-07,
+      "loss": 0.7988,
+      "step": 11128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1353408060609527e-07,
+      "loss": 0.7452,
+      "step": 11129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1332079437485294e-07,
+      "loss": 0.5631,
+      "step": 11130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1310770402686439e-07,
+      "loss": 0.6484,
+      "step": 11131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.128948095796184e-07,
+      "loss": 0.6657,
+      "step": 11132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1268211105058741e-07,
+      "loss": 0.6747,
+      "step": 11133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1246960845722937e-07,
+      "loss": 0.7202,
+      "step": 11134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1225730181698452e-07,
+      "loss": 0.6873,
+      "step": 11135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1204519114727724e-07,
+      "loss": 0.7862,
+      "step": 11136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1183327646551723e-07,
+      "loss": 0.6934,
+      "step": 11137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1162155778909584e-07,
+      "loss": 0.7356,
+      "step": 11138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1141003513539084e-07,
+      "loss": 0.779,
+      "step": 11139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1119870852176196e-07,
+      "loss": 0.6309,
+      "step": 11140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1098757796555365e-07,
+      "loss": 0.6804,
+      "step": 11141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1077664348409429e-07,
+      "loss": 0.7544,
+      "step": 11142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1056590509469556e-07,
+      "loss": 0.7887,
+      "step": 11143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1035536281465364e-07,
+      "loss": 0.6571,
+      "step": 11144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1014501666124916e-07,
+      "loss": 0.7184,
+      "step": 11145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0993486665174524e-07,
+      "loss": 0.5959,
+      "step": 11146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0972491280338976e-07,
+      "loss": 0.7391,
+      "step": 11147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0951515513341476e-07,
+      "loss": 0.6903,
+      "step": 11148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0930559365903537e-07,
+      "loss": 0.75,
+      "step": 11149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0909622839745144e-07,
+      "loss": 0.7417,
+      "step": 11150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0888705936584559e-07,
+      "loss": 0.6646,
+      "step": 11151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0867808658138578e-07,
+      "loss": 0.821,
+      "step": 11152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0846931006122301e-07,
+      "loss": 0.6351,
+      "step": 11153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0826072982249192e-07,
+      "loss": 0.7002,
+      "step": 11154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0805234588231211e-07,
+      "loss": 0.6647,
+      "step": 11155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0784415825778577e-07,
+      "loss": 0.7177,
+      "step": 11156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0763616696600004e-07,
+      "loss": 0.6497,
+      "step": 11157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0742837202402518e-07,
+      "loss": 0.6336,
+      "step": 11158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0722077344891612e-07,
+      "loss": 0.5591,
+      "step": 11159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0701337125771066e-07,
+      "loss": 0.6286,
+      "step": 11160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.06806165467431e-07,
+      "loss": 0.6932,
+      "step": 11161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0659915609508409e-07,
+      "loss": 0.6657,
+      "step": 11162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0639234315765912e-07,
+      "loss": 0.6151,
+      "step": 11163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0618572667213029e-07,
+      "loss": 0.5997,
+      "step": 11164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0597930665545486e-07,
+      "loss": 0.7548,
+      "step": 11165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.057730831245754e-07,
+      "loss": 0.636,
+      "step": 11166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0556705609641671e-07,
+      "loss": 0.7298,
+      "step": 11167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0536122558788803e-07,
+      "loss": 0.6024,
+      "step": 11168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0515559161588307e-07,
+      "loss": 0.5958,
+      "step": 11169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0495015419727888e-07,
+      "loss": 0.6965,
+      "step": 11170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0474491334893617e-07,
+      "loss": 0.5781,
+      "step": 11171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0453986908770003e-07,
+      "loss": 0.7625,
+      "step": 11172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0433502143039925e-07,
+      "loss": 0.6593,
+      "step": 11173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0413037039384594e-07,
+      "loss": 0.6865,
+      "step": 11174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0392591599483665e-07,
+      "loss": 0.7546,
+      "step": 11175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0372165825015212e-07,
+      "loss": 0.8005,
+      "step": 11176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0351759717655646e-07,
+      "loss": 0.7317,
+      "step": 11177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.033137327907971e-07,
+      "loss": 0.7021,
+      "step": 11178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0311006510960675e-07,
+      "loss": 0.7428,
+      "step": 11179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0290659414970039e-07,
+      "loss": 0.8263,
+      "step": 11180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0270331992777826e-07,
+      "loss": 0.6218,
+      "step": 11181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0250024246052287e-07,
+      "loss": 0.7911,
+      "step": 11182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0229736176460254e-07,
+      "loss": 0.5882,
+      "step": 11183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0209467785666782e-07,
+      "loss": 0.5948,
+      "step": 11184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0189219075335377e-07,
+      "loss": 0.7823,
+      "step": 11185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.016899004712793e-07,
+      "loss": 0.6834,
+      "step": 11186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0148780702704725e-07,
+      "loss": 0.6911,
+      "step": 11187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0128591043724406e-07,
+      "loss": 0.6014,
+      "step": 11188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0108421071843954e-07,
+      "loss": 0.6816,
+      "step": 11189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.008827078871888e-07,
+      "loss": 0.5717,
+      "step": 11190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0068140196002968e-07,
+      "loss": 0.6954,
+      "step": 11191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0048029295348316e-07,
+      "loss": 0.8223,
+      "step": 11192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0027938088405604e-07,
+      "loss": 0.7753,
+      "step": 11193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0007866576823761e-07,
+      "loss": 0.6299,
+      "step": 11194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.987814762250137e-08,
+      "loss": 0.749,
+      "step": 11195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.967782646330388e-08,
+      "loss": 0.7259,
+      "step": 11196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.947770230708726e-08,
+      "loss": 0.8168,
+      "step": 11197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.92777751702756e-08,
+      "loss": 0.6566,
+      "step": 11198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.90780450692777e-08,
+      "loss": 0.6719,
+      "step": 11199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.887851202048687e-08,
+      "loss": 0.673,
+      "step": 11200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.86791760402786e-08,
+      "loss": 0.7991,
+      "step": 11201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.84800371450137e-08,
+      "loss": 0.6973,
+      "step": 11202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.828109535103547e-08,
+      "loss": 0.6714,
+      "step": 11203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.808235067467281e-08,
+      "loss": 0.6372,
+      "step": 11204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.788380313223655e-08,
+      "loss": 0.7524,
+      "step": 11205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.768545274002227e-08,
+      "loss": 0.672,
+      "step": 11206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.748729951431002e-08,
+      "loss": 0.8567,
+      "step": 11207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.728934347136231e-08,
+      "loss": 0.5768,
+      "step": 11208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.709158462742591e-08,
+      "loss": 0.6836,
+      "step": 11209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.689402299873224e-08,
+      "loss": 0.6834,
+      "step": 11210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.66966586014953e-08,
+      "loss": 0.6531,
+      "step": 11211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.649949145191406e-08,
+      "loss": 0.6738,
+      "step": 11212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.630252156616976e-08,
+      "loss": 0.725,
+      "step": 11213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.610574896042973e-08,
+      "loss": 0.7334,
+      "step": 11214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.590917365084301e-08,
+      "loss": 0.6131,
+      "step": 11215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.571279565354308e-08,
+      "loss": 0.7613,
+      "step": 11216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.551661498464815e-08,
+      "loss": 0.6785,
+      "step": 11217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.532063166025868e-08,
+      "loss": 0.683,
+      "step": 11218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.512484569646013e-08,
+      "loss": 0.7005,
+      "step": 11219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.492925710932105e-08,
+      "loss": 0.584,
+      "step": 11220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.473386591489441e-08,
+      "loss": 0.6532,
+      "step": 11221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.453867212921658e-08,
+      "loss": 0.607,
+      "step": 11222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.43436757683075e-08,
+      "loss": 0.7514,
+      "step": 11223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.414887684817187e-08,
+      "loss": 0.8327,
+      "step": 11224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.39542753847969e-08,
+      "loss": 0.7269,
+      "step": 11225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.375987139415455e-08,
+      "loss": 0.7019,
+      "step": 11226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.35656648922001e-08,
+      "loss": 0.6587,
+      "step": 11227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.337165589487274e-08,
+      "loss": 0.7657,
+      "step": 11228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.317784441809585e-08,
+      "loss": 0.7272,
+      "step": 11229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.298423047777561e-08,
+      "loss": 0.687,
+      "step": 11230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.279081408980262e-08,
+      "loss": 0.6613,
+      "step": 11231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.259759527005224e-08,
+      "loss": 0.8007,
+      "step": 11232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.24045740343818e-08,
+      "loss": 0.7118,
+      "step": 11233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.221175039863307e-08,
+      "loss": 0.7515,
+      "step": 11234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.201912437863253e-08,
+      "loss": 0.6213,
+      "step": 11235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.182669599018895e-08,
+      "loss": 0.7639,
+      "step": 11236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.163446524909581e-08,
+      "loss": 0.725,
+      "step": 11237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.144243217113047e-08,
+      "loss": 0.604,
+      "step": 11238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.125059677205366e-08,
+      "loss": 0.679,
+      "step": 11239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.105895906761003e-08,
+      "loss": 0.7405,
+      "step": 11240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.086751907352726e-08,
+      "loss": 0.7046,
+      "step": 11241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.06762768055186e-08,
+      "loss": 0.6252,
+      "step": 11242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.048523227927957e-08,
+      "loss": 0.7142,
+      "step": 11243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.029438551048931e-08,
+      "loss": 0.7019,
+      "step": 11244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.010373651481191e-08,
+      "loss": 0.7937,
+      "step": 11245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.991328530789462e-08,
+      "loss": 0.6874,
+      "step": 11246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.972303190536797e-08,
+      "loss": 0.7526,
+      "step": 11247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.953297632284753e-08,
+      "loss": 0.624,
+      "step": 11248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.934311857593081e-08,
+      "loss": 0.8354,
+      "step": 11249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.915345868020093e-08,
+      "loss": 0.6652,
+      "step": 11250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.896399665122373e-08,
+      "loss": 0.7484,
+      "step": 11251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.877473250454904e-08,
+      "loss": 0.5777,
+      "step": 11252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.858566625571025e-08,
+      "loss": 0.6053,
+      "step": 11253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.839679792022465e-08,
+      "loss": 0.696,
+      "step": 11254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.820812751359404e-08,
+      "loss": 0.6742,
+      "step": 11255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.801965505130266e-08,
+      "loss": 0.6426,
+      "step": 11256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.783138054881901e-08,
+      "loss": 0.7959,
+      "step": 11257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.764330402159543e-08,
+      "loss": 0.7035,
+      "step": 11258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.745542548506875e-08,
+      "loss": 0.7934,
+      "step": 11259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.726774495465829e-08,
+      "loss": 0.6834,
+      "step": 11260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.708026244576728e-08,
+      "loss": 0.638,
+      "step": 11261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.68929779737837e-08,
+      "loss": 0.6952,
+      "step": 11262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.67058915540786e-08,
+      "loss": 0.6107,
+      "step": 11263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.651900320200662e-08,
+      "loss": 0.7786,
+      "step": 11264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.633231293290606e-08,
+      "loss": 0.6867,
+      "step": 11265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.614582076209993e-08,
+      "loss": 0.7144,
+      "step": 11266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.595952670489378e-08,
+      "loss": 0.8689,
+      "step": 11267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.577343077657785e-08,
+      "loss": 0.777,
+      "step": 11268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.558753299242522e-08,
+      "loss": 0.5881,
+      "step": 11269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.54018333676937e-08,
+      "loss": 0.7235,
+      "step": 11270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.521633191762413e-08,
+      "loss": 0.6953,
+      "step": 11271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.503102865744073e-08,
+      "loss": 0.6723,
+      "step": 11272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.484592360235299e-08,
+      "loss": 0.7031,
+      "step": 11273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.466101676755239e-08,
+      "loss": 0.6608,
+      "step": 11274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.447630816821484e-08,
+      "loss": 0.614,
+      "step": 11275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.42917978195007e-08,
+      "loss": 0.7339,
+      "step": 11276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.410748573655286e-08,
+      "loss": 0.7268,
+      "step": 11277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.392337193449868e-08,
+      "loss": 0.7814,
+      "step": 11278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.373945642844883e-08,
+      "loss": 0.7768,
+      "step": 11279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.355573923349819e-08,
+      "loss": 0.665,
+      "step": 11280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.337222036472498e-08,
+      "loss": 0.6799,
+      "step": 11281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.318889983719075e-08,
+      "loss": 0.6609,
+      "step": 11282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.300577766594209e-08,
+      "loss": 0.6976,
+      "step": 11283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.282285386600836e-08,
+      "loss": 0.6564,
+      "step": 11284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.2640128452402e-08,
+      "loss": 0.8091,
+      "step": 11285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.245760144012072e-08,
+      "loss": 0.646,
+      "step": 11286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.227527284414477e-08,
+      "loss": 0.6104,
+      "step": 11287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.209314267943885e-08,
+      "loss": 0.6758,
+      "step": 11288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.191121096095045e-08,
+      "loss": 0.7868,
+      "step": 11289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.172947770361206e-08,
+      "loss": 0.6801,
+      "step": 11290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.15479429223387e-08,
+      "loss": 0.7235,
+      "step": 11291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.136660663202984e-08,
+      "loss": 0.6991,
+      "step": 11292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.118546884756828e-08,
+      "loss": 0.618,
+      "step": 11293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.100452958382049e-08,
+      "loss": 0.6444,
+      "step": 11294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.082378885563708e-08,
+      "loss": 0.6914,
+      "step": 11295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.064324667785173e-08,
+      "loss": 0.6663,
+      "step": 11296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.046290306528259e-08,
+      "loss": 0.668,
+      "step": 11297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.028275803273089e-08,
+      "loss": 0.6598,
+      "step": 11298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.010281159498146e-08,
+      "loss": 0.6302,
+      "step": 11299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.992306376680386e-08,
+      "loss": 0.6231,
+      "step": 11300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.97435145629502e-08,
+      "loss": 0.7507,
+      "step": 11301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.956416399815675e-08,
+      "loss": 0.6642,
+      "step": 11302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.93850120871431e-08,
+      "loss": 0.596,
+      "step": 11303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.920605884461364e-08,
+      "loss": 0.7046,
+      "step": 11304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.902730428525518e-08,
+      "loss": 0.6467,
+      "step": 11305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.88487484237388e-08,
+      "loss": 0.6495,
+      "step": 11306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.867039127471915e-08,
+      "loss": 0.7057,
+      "step": 11307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.849223285283535e-08,
+      "loss": 0.6995,
+      "step": 11308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.831427317270852e-08,
+      "loss": 0.6238,
+      "step": 11309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.81365122489447e-08,
+      "loss": 0.6532,
+      "step": 11310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.79589500961339e-08,
+      "loss": 0.632,
+      "step": 11311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.778158672884867e-08,
+      "loss": 0.7338,
+      "step": 11312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.760442216164593e-08,
+      "loss": 0.6383,
+      "step": 11313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.742745640906658e-08,
+      "loss": 0.6405,
+      "step": 11314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.725068948563453e-08,
+      "loss": 0.6911,
+      "step": 11315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.707412140585791e-08,
+      "loss": 0.6384,
+      "step": 11316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.689775218422762e-08,
+      "loss": 0.7192,
+      "step": 11317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.672158183521961e-08,
+      "loss": 0.5466,
+      "step": 11318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.654561037329284e-08,
+      "loss": 0.6271,
+      "step": 11319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.636983781288937e-08,
+      "loss": 0.6548,
+      "step": 11320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.6194264168436e-08,
+      "loss": 0.7532,
+      "step": 11321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.601888945434233e-08,
+      "loss": 0.6339,
+      "step": 11322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.58437136850021e-08,
+      "loss": 0.7977,
+      "step": 11323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.56687368747927e-08,
+      "loss": 0.8924,
+      "step": 11324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.549395903807488e-08,
+      "loss": 0.6333,
+      "step": 11325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.531938018919355e-08,
+      "loss": 0.5451,
+      "step": 11326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.514500034247668e-08,
+      "loss": 0.6895,
+      "step": 11327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.497081951223672e-08,
+      "loss": 0.691,
+      "step": 11328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.47968377127692e-08,
+      "loss": 0.7538,
+      "step": 11329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.462305495835292e-08,
+      "loss": 0.7684,
+      "step": 11330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.444947126325124e-08,
+      "loss": 0.6606,
+      "step": 11331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.427608664171109e-08,
+      "loss": 0.8463,
+      "step": 11332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.41029011079622e-08,
+      "loss": 0.6769,
+      "step": 11333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.392991467621874e-08,
+      "loss": 0.6257,
+      "step": 11334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.375712736067825e-08,
+      "loss": 0.6969,
+      "step": 11335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.358453917552243e-08,
+      "loss": 0.7303,
+      "step": 11336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.341215013491526e-08,
+      "loss": 0.6239,
+      "step": 11337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.32399602530065e-08,
+      "loss": 0.7746,
+      "step": 11338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.306796954392764e-08,
+      "loss": 0.677,
+      "step": 11339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.289617802179489e-08,
+      "loss": 0.6173,
+      "step": 11340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.272458570070723e-08,
+      "loss": 0.7355,
+      "step": 11341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.255319259474842e-08,
+      "loss": 0.7574,
+      "step": 11342
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.238199871798496e-08,
+      "loss": 0.657,
+      "step": 11343
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.221100408446784e-08,
+      "loss": 0.6832,
+      "step": 11344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.204020870823059e-08,
+      "loss": 0.5974,
+      "step": 11345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.18696126032914e-08,
+      "loss": 0.6366,
+      "step": 11346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.169921578365185e-08,
+      "loss": 0.676,
+      "step": 11347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.152901826329633e-08,
+      "loss": 0.7853,
+      "step": 11348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.135902005619422e-08,
+      "loss": 0.7477,
+      "step": 11349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.118922117629768e-08,
+      "loss": 0.727,
+      "step": 11350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.101962163754251e-08,
+      "loss": 0.692,
+      "step": 11351
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.085022145384868e-08,
+      "loss": 0.751,
+      "step": 11352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.068102063911925e-08,
+      "loss": 0.7226,
+      "step": 11353
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.051201920724116e-08,
+      "loss": 0.7815,
+      "step": 11354
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.03432171720847e-08,
+      "loss": 0.83,
+      "step": 11355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.017461454750434e-08,
+      "loss": 0.6803,
+      "step": 11356
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.00062113473382e-08,
+      "loss": 0.7708,
+      "step": 11357
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.983800758540715e-08,
+      "loss": 0.6297,
+      "step": 11358
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.967000327551654e-08,
+      "loss": 0.7853,
+      "step": 11359
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.950219843145534e-08,
+      "loss": 0.6412,
+      "step": 11360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.933459306699502e-08,
+      "loss": 0.6202,
+      "step": 11361
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.916718719589266e-08,
+      "loss": 0.6442,
+      "step": 11362
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.899998083188697e-08,
+      "loss": 0.7636,
+      "step": 11363
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.883297398870197e-08,
+      "loss": 0.6983,
+      "step": 11364
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.866616668004366e-08,
+      "loss": 0.7973,
+      "step": 11365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.84995589196033e-08,
+      "loss": 0.7167,
+      "step": 11366
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.83331507210544e-08,
+      "loss": 0.6321,
+      "step": 11367
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.816694209805491e-08,
+      "loss": 0.7181,
+      "step": 11368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.800093306424616e-08,
+      "loss": 0.7615,
+      "step": 11369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.783512363325306e-08,
+      "loss": 0.6441,
+      "step": 11370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.766951381868448e-08,
+      "loss": 0.6174,
+      "step": 11371
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.750410363413173e-08,
+      "loss": 0.6324,
+      "step": 11372
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.733889309317176e-08,
+      "loss": 0.7507,
+      "step": 11373
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.717388220936316e-08,
+      "loss": 0.7767,
+      "step": 11374
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.700907099624925e-08,
+      "loss": 0.6874,
+      "step": 11375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.684445946735674e-08,
+      "loss": 0.6542,
+      "step": 11376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.668004763619562e-08,
+      "loss": 0.6122,
+      "step": 11377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.65158355162604e-08,
+      "loss": 0.6456,
+      "step": 11378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.63518231210275e-08,
+      "loss": 0.7668,
+      "step": 11379
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.618801046395896e-08,
+      "loss": 0.6041,
+      "step": 11380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.602439755849871e-08,
+      "loss": 0.5406,
+      "step": 11381
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.586098441807576e-08,
+      "loss": 0.6516,
+      "step": 11382
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.569777105610131e-08,
+      "loss": 0.5865,
+      "step": 11383
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.55347574859716e-08,
+      "loss": 0.7083,
+      "step": 11384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.537194372106537e-08,
+      "loss": 0.7775,
+      "step": 11385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.520932977474497e-08,
+      "loss": 0.7561,
+      "step": 11386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.504691566035726e-08,
+      "loss": 0.7271,
+      "step": 11387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.488470139123209e-08,
+      "loss": 0.7959,
+      "step": 11388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.472268698068245e-08,
+      "loss": 0.6744,
+      "step": 11389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.456087244200632e-08,
+      "loss": 0.7517,
+      "step": 11390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.439925778848361e-08,
+      "loss": 0.8617,
+      "step": 11391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.42378430333787e-08,
+      "loss": 0.7809,
+      "step": 11392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.407662818993965e-08,
+      "loss": 0.6568,
+      "step": 11393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.391561327139833e-08,
+      "loss": 0.6932,
+      "step": 11394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.37547982909692e-08,
+      "loss": 0.6396,
+      "step": 11395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.359418326185085e-08,
+      "loss": 0.6746,
+      "step": 11396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.343376819722608e-08,
+      "loss": 0.6388,
+      "step": 11397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.327355311026046e-08,
+      "loss": 0.5959,
+      "step": 11398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.311353801410291e-08,
+      "loss": 0.6422,
+      "step": 11399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.29537229218874e-08,
+      "loss": 0.859,
+      "step": 11400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.279410784672979e-08,
+      "loss": 0.6626,
+      "step": 11401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.263469280173074e-08,
+      "loss": 0.7183,
+      "step": 11402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.247547779997365e-08,
+      "loss": 0.8135,
+      "step": 11403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.231646285452614e-08,
+      "loss": 0.663,
+      "step": 11404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.215764797843915e-08,
+      "loss": 0.6172,
+      "step": 11405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.199903318474642e-08,
+      "loss": 0.6335,
+      "step": 11406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.184061848646727e-08,
+      "loss": 0.7171,
+      "step": 11407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.16824038966024e-08,
+      "loss": 0.677,
+      "step": 11408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.152438942813754e-08,
+      "loss": 0.67,
+      "step": 11409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.13665750940412e-08,
+      "loss": 0.6216,
+      "step": 11410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.120896090726581e-08,
+      "loss": 0.7122,
+      "step": 11411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.105154688074765e-08,
+      "loss": 0.7959,
+      "step": 11412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.089433302740561e-08,
+      "loss": 0.7237,
+      "step": 11413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.073731936014348e-08,
+      "loss": 0.8127,
+      "step": 11414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.058050589184766e-08,
+      "loss": 0.6674,
+      "step": 11415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.042389263538839e-08,
+      "loss": 0.6476,
+      "step": 11416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.026747960361929e-08,
+      "loss": 0.6249,
+      "step": 11417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.011126680937785e-08,
+      "loss": 0.6952,
+      "step": 11418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.995525426548493e-08,
+      "loss": 0.7011,
+      "step": 11419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.979944198474558e-08,
+      "loss": 0.6613,
+      "step": 11420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.964382997994706e-08,
+      "loss": 0.6447,
+      "step": 11421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.9488418263861646e-08,
+      "loss": 0.7402,
+      "step": 11422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.933320684924415e-08,
+      "loss": 0.7173,
+      "step": 11423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.917819574883355e-08,
+      "loss": 0.7345,
+      "step": 11424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.902338497535187e-08,
+      "loss": 0.6212,
+      "step": 11425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.886877454150536e-08,
+      "loss": 0.6239,
+      "step": 11426
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.871436445998302e-08,
+      "loss": 0.6202,
+      "step": 11427
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.856015474345833e-08,
+      "loss": 0.6334,
+      "step": 11428
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8406145404587554e-08,
+      "loss": 0.7171,
+      "step": 11429
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.825233645601086e-08,
+      "loss": 0.6479,
+      "step": 11430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8098727910351484e-08,
+      "loss": 0.6686,
+      "step": 11431
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.794531978021739e-08,
+      "loss": 0.703,
+      "step": 11432
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.779211207819879e-08,
+      "loss": 0.7382,
+      "step": 11433
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.76391048168698e-08,
+      "loss": 0.7488,
+      "step": 11434
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.7486298008788964e-08,
+      "loss": 0.7547,
+      "step": 11435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.7333691666497384e-08,
+      "loss": 0.8507,
+      "step": 11436
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.718128580251947e-08,
+      "loss": 0.7555,
+      "step": 11437
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.702908042936439e-08,
+      "loss": 0.6378,
+      "step": 11438
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.687707555952382e-08,
+      "loss": 0.8032,
+      "step": 11439
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.672527120547361e-08,
+      "loss": 0.7071,
+      "step": 11440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.657366737967268e-08,
+      "loss": 0.6586,
+      "step": 11441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.642226409456386e-08,
+      "loss": 0.7332,
+      "step": 11442
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.627106136257332e-08,
+      "loss": 0.6761,
+      "step": 11443
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.612005919611058e-08,
+      "loss": 0.8232,
+      "step": 11444
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.596925760756933e-08,
+      "loss": 0.689,
+      "step": 11445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.581865660932606e-08,
+      "loss": 0.7585,
+      "step": 11446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.566825621374117e-08,
+      "loss": 0.6843,
+      "step": 11447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.551805643315866e-08,
+      "loss": 0.6592,
+      "step": 11448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.536805727990591e-08,
+      "loss": 0.8023,
+      "step": 11449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.521825876629389e-08,
+      "loss": 0.7026,
+      "step": 11450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.506866090461693e-08,
+      "loss": 0.7268,
+      "step": 11451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.491926370715328e-08,
+      "loss": 0.6756,
+      "step": 11452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.477006718616479e-08,
+      "loss": 0.6775,
+      "step": 11453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.462107135389583e-08,
+      "loss": 0.6579,
+      "step": 11454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4472276222575226e-08,
+      "loss": 0.6562,
+      "step": 11455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.432368180441572e-08,
+      "loss": 0.7655,
+      "step": 11456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.417528811161227e-08,
+      "loss": 0.6462,
+      "step": 11457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.402709515634458e-08,
+      "loss": 0.6701,
+      "step": 11458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.3879102950774864e-08,
+      "loss": 0.7067,
+      "step": 11459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.37313115070498e-08,
+      "loss": 0.594,
+      "step": 11460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.3583720837299125e-08,
+      "loss": 0.7953,
+      "step": 11461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.343633095363565e-08,
+      "loss": 0.6361,
+      "step": 11462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.328914186815693e-08,
+      "loss": 0.724,
+      "step": 11463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.314215359294272e-08,
+      "loss": 0.712,
+      "step": 11464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2995366140057014e-08,
+      "loss": 0.762,
+      "step": 11465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.284877952154738e-08,
+      "loss": 0.7029,
+      "step": 11466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.270239374944475e-08,
+      "loss": 0.795,
+      "step": 11467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2556208835763125e-08,
+      "loss": 0.5909,
+      "step": 11468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2410224792500406e-08,
+      "loss": 0.6877,
+      "step": 11469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2264441631638685e-08,
+      "loss": 0.7971,
+      "step": 11470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2118859365142275e-08,
+      "loss": 0.673,
+      "step": 11471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.197347800495967e-08,
+      "loss": 0.8071,
+      "step": 11472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.182829756302327e-08,
+      "loss": 0.7524,
+      "step": 11473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.168331805124799e-08,
+      "loss": 0.6922,
+      "step": 11474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.15385394815332e-08,
+      "loss": 0.7078,
+      "step": 11475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.139396186576134e-08,
+      "loss": 0.6819,
+      "step": 11476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.12495852157982e-08,
+      "loss": 0.6231,
+      "step": 11477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.110540954349374e-08,
+      "loss": 0.6243,
+      "step": 11478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.096143486068017e-08,
+      "loss": 0.6769,
+      "step": 11479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.081766117917497e-08,
+      "loss": 0.7049,
+      "step": 11480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.06740885107776e-08,
+      "loss": 0.6399,
+      "step": 11481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.05307168672714e-08,
+      "loss": 0.6612,
+      "step": 11482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.0387546260423914e-08,
+      "loss": 0.6218,
+      "step": 11483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.024457670198546e-08,
+      "loss": 0.7048,
+      "step": 11484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.010180820368998e-08,
+      "loss": 0.6329,
+      "step": 11485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.995924077725478e-08,
+      "loss": 0.7751,
+      "step": 11486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.98168744343816e-08,
+      "loss": 0.6156,
+      "step": 11487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.967470918675416e-08,
+      "loss": 0.6302,
+      "step": 11488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.953274504604061e-08,
+      "loss": 0.6249,
+      "step": 11489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.939098202389275e-08,
+      "loss": 0.7001,
+      "step": 11490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9249420131945725e-08,
+      "loss": 0.7344,
+      "step": 11491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.910805938181745e-08,
+      "loss": 0.5816,
+      "step": 11492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.896689978511032e-08,
+      "loss": 0.6291,
+      "step": 11493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.882594135340951e-08,
+      "loss": 0.7366,
+      "step": 11494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.868518409828438e-08,
+      "loss": 0.6178,
+      "step": 11495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8544628031287064e-08,
+      "loss": 0.6376,
+      "step": 11496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.840427316395335e-08,
+      "loss": 0.7032,
+      "step": 11497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.826411950780319e-08,
+      "loss": 0.7553,
+      "step": 11498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.812416707433931e-08,
+      "loss": 0.6722,
+      "step": 11499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.798441587504754e-08,
+      "loss": 0.7719,
+      "step": 11500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.78448659213987e-08,
+      "loss": 0.8266,
+      "step": 11501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7705517224845285e-08,
+      "loss": 0.8273,
+      "step": 11502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.756636979682455e-08,
+      "loss": 0.8415,
+      "step": 11503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.742742364875708e-08,
+      "loss": 0.6376,
+      "step": 11504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.728867879204624e-08,
+      "loss": 0.6258,
+      "step": 11505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7150135238079596e-08,
+      "loss": 0.6788,
+      "step": 11506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.70117929982275e-08,
+      "loss": 0.6955,
+      "step": 11507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6873652083844756e-08,
+      "loss": 0.8029,
+      "step": 11508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6735712506268684e-08,
+      "loss": 0.5893,
+      "step": 11509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6597974276820224e-08,
+      "loss": 0.7294,
+      "step": 11510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.646043740680478e-08,
+      "loss": 0.853,
+      "step": 11511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.632310190751027e-08,
+      "loss": 0.7653,
+      "step": 11512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.618596779020768e-08,
+      "loss": 0.7187,
+      "step": 11513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6049035066152994e-08,
+      "loss": 0.8238,
+      "step": 11514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5912303746583894e-08,
+      "loss": 0.7346,
+      "step": 11515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.577577384272336e-08,
+      "loss": 0.6357,
+      "step": 11516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.563944536577575e-08,
+      "loss": 0.6335,
+      "step": 11517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.550331832693128e-08,
+      "loss": 0.7553,
+      "step": 11518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5367392737361574e-08,
+      "loss": 0.7547,
+      "step": 11519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.523166860822242e-08,
+      "loss": 0.5962,
+      "step": 11520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5096145950653815e-08,
+      "loss": 0.7411,
+      "step": 11521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.496082477577823e-08,
+      "loss": 0.6294,
+      "step": 11522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.482570509470208e-08,
+      "loss": 0.6835,
+      "step": 11523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4690786918514525e-08,
+      "loss": 0.8527,
+      "step": 11524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.455607025828951e-08,
+      "loss": 0.7645,
+      "step": 11525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4421555125083724e-08,
+      "loss": 0.6498,
+      "step": 11526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4287241529936686e-08,
+      "loss": 0.6276,
+      "step": 11527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4153129483872345e-08,
+      "loss": 0.7707,
+      "step": 11528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.4019218997898014e-08,
+      "loss": 0.6639,
+      "step": 11529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.388551008300379e-08,
+      "loss": 0.7817,
+      "step": 11530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.3752002750163666e-08,
+      "loss": 0.6856,
+      "step": 11531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.3618697010335557e-08,
+      "loss": 0.7027,
+      "step": 11532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.348559287445958e-08,
+      "loss": 0.6773,
+      "step": 11533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.335269035346063e-08,
+      "loss": 0.6,
+      "step": 11534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.321998945824607e-08,
+      "loss": 0.6386,
+      "step": 11535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.308749019970776e-08,
+      "loss": 0.7399,
+      "step": 11536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.295519258871977e-08,
+      "loss": 0.6742,
+      "step": 11537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.282309663614037e-08,
+      "loss": 0.7259,
+      "step": 11538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.269120235281144e-08,
+      "loss": 0.6253,
+      "step": 11539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2559509749557915e-08,
+      "loss": 0.6824,
+      "step": 11540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2428018837188115e-08,
+      "loss": 0.7318,
+      "step": 11541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.229672962649395e-08,
+      "loss": 0.6308,
+      "step": 11542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.216564212825097e-08,
+      "loss": 0.658,
+      "step": 11543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2034756353217795e-08,
+      "loss": 0.6755,
+      "step": 11544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.190407231213667e-08,
+      "loss": 0.5853,
+      "step": 11545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.1773590015733725e-08,
+      "loss": 0.585,
+      "step": 11546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.164330947471762e-08,
+      "loss": 0.7802,
+      "step": 11547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.1513230699781196e-08,
+      "loss": 0.6451,
+      "step": 11548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.1383353701600346e-08,
+      "loss": 0.7271,
+      "step": 11549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.125367849083461e-08,
+      "loss": 0.6582,
+      "step": 11550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.1124205078126865e-08,
+      "loss": 0.6495,
+      "step": 11551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0994933474103325e-08,
+      "loss": 0.7403,
+      "step": 11552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.086586368937412e-08,
+      "loss": 0.6651,
+      "step": 11553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.073699573453216e-08,
+      "loss": 0.8424,
+      "step": 11554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0608329620154263e-08,
+      "loss": 0.7456,
+      "step": 11555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0479865356800316e-08,
+      "loss": 0.6207,
+      "step": 11556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.035160295501411e-08,
+      "loss": 0.6263,
+      "step": 11557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.02235424253225e-08,
+      "loss": 0.6461,
+      "step": 11558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.009568377823569e-08,
+      "loss": 0.6289,
+      "step": 11559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.996802702424779e-08,
+      "loss": 0.8241,
+      "step": 11560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9840572173835965e-08,
+      "loss": 0.7258,
+      "step": 11561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.971331923746047e-08,
+      "loss": 0.7342,
+      "step": 11562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.958626822556599e-08,
+      "loss": 0.683,
+      "step": 11563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.945941914857976e-08,
+      "loss": 0.685,
+      "step": 11564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.93327720169126e-08,
+      "loss": 0.6398,
+      "step": 11565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.920632684095954e-08,
+      "loss": 0.596,
+      "step": 11566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9080083631097554e-08,
+      "loss": 0.7588,
+      "step": 11567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8954042397688354e-08,
+      "loss": 0.5861,
+      "step": 11568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8828203151076436e-08,
+      "loss": 0.767,
+      "step": 11569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.870256590158994e-08,
+      "loss": 0.766,
+      "step": 11570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.857713065954005e-08,
+      "loss": 0.6544,
+      "step": 11571
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.8010957837104797,
+      "eval_runtime": 437.0502,
+      "eval_samples_per_second": 6.862,
+      "eval_steps_per_second": 0.858,
+      "step": 11571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.845189743522243e-08,
+      "loss": 0.6371,
+      "step": 11572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.832686623891468e-08,
+      "loss": 0.7828,
+      "step": 11573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8202037080878884e-08,
+      "loss": 0.664,
+      "step": 11574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.807740997136017e-08,
+      "loss": 0.7011,
+      "step": 11575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.795298492058702e-08,
+      "loss": 0.6927,
+      "step": 11576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.782876193877183e-08,
+      "loss": 0.6449,
+      "step": 11577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.770474103610949e-08,
+      "loss": 0.6501,
+      "step": 11578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7580922222779094e-08,
+      "loss": 0.614,
+      "step": 11579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.745730550894305e-08,
+      "loss": 0.6761,
+      "step": 11580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7333890904746596e-08,
+      "loss": 0.6189,
+      "step": 11581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.721067842031939e-08,
+      "loss": 0.7491,
+      "step": 11582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7087668065773353e-08,
+      "loss": 0.7573,
+      "step": 11583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.696485985120457e-08,
+      "loss": 0.748,
+      "step": 11584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.684225378669248e-08,
+      "loss": 0.7608,
+      "step": 11585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6719849882299596e-08,
+      "loss": 0.711,
+      "step": 11586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.659764814807232e-08,
+      "loss": 0.789,
+      "step": 11587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.647564859404012e-08,
+      "loss": 0.6946,
+      "step": 11588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6353851230215566e-08,
+      "loss": 0.5903,
+      "step": 11589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.623225606659564e-08,
+      "loss": 0.7764,
+      "step": 11590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.611086311315931e-08,
+      "loss": 0.6325,
+      "step": 11591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.598967237987028e-08,
+      "loss": 0.6342,
+      "step": 11592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.586868387667503e-08,
+      "loss": 0.6762,
+      "step": 11593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.574789761350339e-08,
+      "loss": 0.5728,
+      "step": 11594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.562731360026911e-08,
+      "loss": 0.7544,
+      "step": 11595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5506931846868156e-08,
+      "loss": 0.6508,
+      "step": 11596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.538675236318123e-08,
+      "loss": 0.726,
+      "step": 11597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.526677515907212e-08,
+      "loss": 0.666,
+      "step": 11598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5147000244387386e-08,
+      "loss": 0.6568,
+      "step": 11599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.502742762895722e-08,
+      "loss": 0.7764,
+      "step": 11600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4908057322595726e-08,
+      "loss": 0.7231,
+      "step": 11601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4788889335100054e-08,
+      "loss": 0.7653,
+      "step": 11602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4669923676250436e-08,
+      "loss": 0.7116,
+      "step": 11603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.45511603558113e-08,
+      "loss": 0.7081,
+      "step": 11604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.443259938352955e-08,
+      "loss": 0.666,
+      "step": 11605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4314240769136035e-08,
+      "loss": 0.6438,
+      "step": 11606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419608452234463e-08,
+      "loss": 0.6552,
+      "step": 11607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.407813065285315e-08,
+      "loss": 0.7259,
+      "step": 11608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.396037917034245e-08,
+      "loss": 0.6686,
+      "step": 11609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.384283008447703e-08,
+      "loss": 0.7466,
+      "step": 11610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.372548340490389e-08,
+      "loss": 0.6081,
+      "step": 11611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.360833914125505e-08,
+      "loss": 0.7299,
+      "step": 11612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3491397303143936e-08,
+      "loss": 0.7603,
+      "step": 11613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.337465790016897e-08,
+      "loss": 0.7838,
+      "step": 11614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.325812094191139e-08,
+      "loss": 0.6338,
+      "step": 11615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.314178643793575e-08,
+      "loss": 0.7369,
+      "step": 11616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.302565439778971e-08,
+      "loss": 0.6659,
+      "step": 11617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.290972483100507e-08,
+      "loss": 0.6878,
+      "step": 11618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2793997747096465e-08,
+      "loss": 0.73,
+      "step": 11619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.267847315556183e-08,
+      "loss": 0.7503,
+      "step": 11620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2563151065882756e-08,
+      "loss": 0.6824,
+      "step": 11621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.244803148752418e-08,
+      "loss": 0.6876,
+      "step": 11622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.233311442993464e-08,
+      "loss": 0.643,
+      "step": 11623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.221839990254522e-08,
+      "loss": 0.7836,
+      "step": 11624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2103887914771424e-08,
+      "loss": 0.6253,
+      "step": 11625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.198957847601159e-08,
+      "loss": 0.8067,
+      "step": 11626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1875471595647075e-08,
+      "loss": 0.7144,
+      "step": 11627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.176156728304375e-08,
+      "loss": 0.6689,
+      "step": 11628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1647865547549394e-08,
+      "loss": 0.6542,
+      "step": 11629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.153436639849655e-08,
+      "loss": 0.6508,
+      "step": 11630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1421069845199994e-08,
+      "loss": 0.7022,
+      "step": 11631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.130797589695894e-08,
+      "loss": 0.6757,
+      "step": 11632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.119508456305514e-08,
+      "loss": 0.6562,
+      "step": 11633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.108239585275341e-08,
+      "loss": 0.7643,
+      "step": 11634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.096990977530356e-08,
+      "loss": 0.744,
+      "step": 11635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.085762633993711e-08,
+      "loss": 0.7378,
+      "step": 11636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.074554555586945e-08,
+      "loss": 0.6372,
+      "step": 11637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.06336674322999e-08,
+      "loss": 0.7267,
+      "step": 11638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.052199197841027e-08,
+      "loss": 0.8069,
+      "step": 11639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.041051920336629e-08,
+      "loss": 0.7101,
+      "step": 11640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.029924911631704e-08,
+      "loss": 0.6502,
+      "step": 11641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.018818172639493e-08,
+      "loss": 0.6737,
+      "step": 11642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.007731704271544e-08,
+      "loss": 0.7722,
+      "step": 11643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9966655074377684e-08,
+      "loss": 0.8049,
+      "step": 11644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9856195830463843e-08,
+      "loss": 0.8275,
+      "step": 11645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9745939320040273e-08,
+      "loss": 0.7788,
+      "step": 11646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9635885552155574e-08,
+      "loss": 0.7929,
+      "step": 11647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.952603453584252e-08,
+      "loss": 0.6691,
+      "step": 11648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9416386280117228e-08,
+      "loss": 0.6083,
+      "step": 11649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.930694079397861e-08,
+      "loss": 0.67,
+      "step": 11650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9197698086408932e-08,
+      "loss": 0.6939,
+      "step": 11651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.908865816637463e-08,
+      "loss": 0.6615,
+      "step": 11652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8979821042824928e-08,
+      "loss": 0.7544,
+      "step": 11653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8871186724692135e-08,
+      "loss": 0.7073,
+      "step": 11654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8762755220892725e-08,
+      "loss": 0.6232,
+      "step": 11655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.865452654032569e-08,
+      "loss": 0.7428,
+      "step": 11656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.85465006918742e-08,
+      "loss": 0.6767,
+      "step": 11657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.843867768440367e-08,
+      "loss": 0.595,
+      "step": 11658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8331057526763962e-08,
+      "loss": 0.7078,
+      "step": 11659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.822364022778773e-08,
+      "loss": 0.7034,
+      "step": 11660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.811642579629098e-08,
+      "loss": 0.7346,
+      "step": 11661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.800941424107334e-08,
+      "loss": 0.7488,
+      "step": 11662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7902605570917784e-08,
+      "loss": 0.6637,
+      "step": 11663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.779599979459008e-08,
+      "loss": 0.6084,
+      "step": 11664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.768959692083989e-08,
+      "loss": 0.7376,
+      "step": 11665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7583396958400234e-08,
+      "loss": 0.6527,
+      "step": 11666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7477399915987468e-08,
+      "loss": 0.6199,
+      "step": 11667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.737160580230075e-08,
+      "loss": 0.6161,
+      "step": 11668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7266014626022853e-08,
+      "loss": 0.6294,
+      "step": 11669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7160626395820734e-08,
+      "loss": 0.7003,
+      "step": 11670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7055441120343306e-08,
+      "loss": 0.7531,
+      "step": 11671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6950458808223668e-08,
+      "loss": 0.751,
+      "step": 11672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6845679468078256e-08,
+      "loss": 0.7468,
+      "step": 11673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.674110310850686e-08,
+      "loss": 0.7472,
+      "step": 11674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6636729738091783e-08,
+      "loss": 0.7098,
+      "step": 11675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6532559365399778e-08,
+      "loss": 0.6932,
+      "step": 11676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.64285919989804e-08,
+      "loss": 0.6336,
+      "step": 11677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6324827647366824e-08,
+      "loss": 0.8956,
+      "step": 11678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.622126631907501e-08,
+      "loss": 0.6267,
+      "step": 11679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.611790802260483e-08,
+      "loss": 0.7909,
+      "step": 11680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6014752766438946e-08,
+      "loss": 0.7408,
+      "step": 11681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5911800559044187e-08,
+      "loss": 0.6119,
+      "step": 11682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5809051408869636e-08,
+      "loss": 0.6904,
+      "step": 11683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5706505324348818e-08,
+      "loss": 0.5729,
+      "step": 11684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5604162313897785e-08,
+      "loss": 0.6588,
+      "step": 11685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5502022385915926e-08,
+      "loss": 0.6421,
+      "step": 11686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.540008554878709e-08,
+      "loss": 0.6904,
+      "step": 11687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.529835181087681e-08,
+      "loss": 0.7137,
+      "step": 11688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5196821180534792e-08,
+      "loss": 0.6045,
+      "step": 11689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5095493666094374e-08,
+      "loss": 0.7565,
+      "step": 11690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4994369275871676e-08,
+      "loss": 0.7508,
+      "step": 11691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4893448018166177e-08,
+      "loss": 0.6621,
+      "step": 11692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.479272990126097e-08,
+      "loss": 0.7428,
+      "step": 11693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.46922149334225e-08,
+      "loss": 0.6259,
+      "step": 11694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4591903122900274e-08,
+      "loss": 0.6667,
+      "step": 11695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4491794477927434e-08,
+      "loss": 0.5777,
+      "step": 11696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4391889006719627e-08,
+      "loss": 0.7269,
+      "step": 11697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.429218671747724e-08,
+      "loss": 0.8104,
+      "step": 11698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.419268761838289e-08,
+      "loss": 0.7879,
+      "step": 11699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4093391717602278e-08,
+      "loss": 0.6596,
+      "step": 11700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3994299023285815e-08,
+      "loss": 0.6754,
+      "step": 11701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3895409543566173e-08,
+      "loss": 0.7721,
+      "step": 11702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3796723286559077e-08,
+      "loss": 0.7193,
+      "step": 11703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3698240260364437e-08,
+      "loss": 0.6675,
+      "step": 11704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.359996047306523e-08,
+      "loss": 0.7191,
+      "step": 11705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.350188393272751e-08,
+      "loss": 0.739,
+      "step": 11706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3404010647400664e-08,
+      "loss": 0.8355,
+      "step": 11707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.330634062511744e-08,
+      "loss": 0.671,
+      "step": 11708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.32088738738942e-08,
+      "loss": 0.6492,
+      "step": 11709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3111610401730376e-08,
+      "loss": 0.725,
+      "step": 11710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3014550216608477e-08,
+      "loss": 0.6838,
+      "step": 11711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2917693326494906e-08,
+      "loss": 0.6294,
+      "step": 11712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2821039739338858e-08,
+      "loss": 0.74,
+      "step": 11713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2724589463072877e-08,
+      "loss": 0.6907,
+      "step": 11714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.262834250561341e-08,
+      "loss": 0.8107,
+      "step": 11715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2532298874859414e-08,
+      "loss": 0.7198,
+      "step": 11716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2436458578693476e-08,
+      "loss": 0.637,
+      "step": 11717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.23408216249818e-08,
+      "loss": 0.6975,
+      "step": 11718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2245388021573667e-08,
+      "loss": 0.6307,
+      "step": 11719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2150157776301696e-08,
+      "loss": 0.6867,
+      "step": 11720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2055130896981302e-08,
+      "loss": 0.7411,
+      "step": 11721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1960307391412083e-08,
+      "loss": 0.5736,
+      "step": 11722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.18656872673767e-08,
+      "loss": 0.7269,
+      "step": 11723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.177127053264061e-08,
+      "loss": 0.5775,
+      "step": 11724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1677057194953167e-08,
+      "loss": 0.7105,
+      "step": 11725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.158304726204652e-08,
+      "loss": 0.6629,
+      "step": 11726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.148924074163672e-08,
+      "loss": 0.6821,
+      "step": 11727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1395637641422607e-08,
+      "loss": 0.6194,
+      "step": 11728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1302237969086647e-08,
+      "loss": 0.6272,
+      "step": 11729
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.120904173229438e-08,
+      "loss": 0.6379,
+      "step": 11730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.11160489386944e-08,
+      "loss": 0.866,
+      "step": 11731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1023259595919775e-08,
+      "loss": 0.5616,
+      "step": 11732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0930673711585526e-08,
+      "loss": 0.8125,
+      "step": 11733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.083829129329057e-08,
+      "loss": 0.7845,
+      "step": 11734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.07461123486169e-08,
+      "loss": 0.84,
+      "step": 11735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0654136885130128e-08,
+      "loss": 0.6961,
+      "step": 11736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.056236491037922e-08,
+      "loss": 0.713,
+      "step": 11737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0470796431895922e-08,
+      "loss": 0.8587,
+      "step": 11738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.03794314571959e-08,
+      "loss": 0.6464,
+      "step": 11739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0288269993777308e-08,
+      "loss": 0.6634,
+      "step": 11740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0197312049122507e-08,
+      "loss": 0.603,
+      "step": 11741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0106557630696633e-08,
+      "loss": 0.6852,
+      "step": 11742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0016006745948168e-08,
+      "loss": 0.6804,
+      "step": 11743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9925659402309227e-08,
+      "loss": 0.7832,
+      "step": 11744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9835515607194434e-08,
+      "loss": 0.5956,
+      "step": 11745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.974557536800259e-08,
+      "loss": 0.6689,
+      "step": 11746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9655838692115292e-08,
+      "loss": 0.7022,
+      "step": 11747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9566305586897482e-08,
+      "loss": 0.8192,
+      "step": 11748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.947697605969745e-08,
+      "loss": 0.788,
+      "step": 11749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9387850117847108e-08,
+      "loss": 0.7616,
+      "step": 11750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.929892776866088e-08,
+      "loss": 0.6053,
+      "step": 11751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9210209019437376e-08,
+      "loss": 0.6773,
+      "step": 11752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9121693877457714e-08,
+      "loss": 0.6315,
+      "step": 11753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.903338234998664e-08,
+      "loss": 0.7016,
+      "step": 11754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.894527444427252e-08,
+      "loss": 0.6832,
+      "step": 11755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.885737016754652e-08,
+      "loss": 0.6031,
+      "step": 11756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8769669527023138e-08,
+      "loss": 0.654,
+      "step": 11757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8682172529900232e-08,
+      "loss": 0.6553,
+      "step": 11758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8594879183359283e-08,
+      "loss": 0.6954,
+      "step": 11759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8507789494564553e-08,
+      "loss": 0.6451,
+      "step": 11760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.842090347066394e-08,
+      "loss": 0.659,
+      "step": 11761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8334221118788408e-08,
+      "loss": 0.7086,
+      "step": 11762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8247742446052264e-08,
+      "loss": 0.633,
+      "step": 11763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8161467459553162e-08,
+      "loss": 0.7902,
+      "step": 11764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8075396166372106e-08,
+      "loss": 0.7252,
+      "step": 11765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.798952857357289e-08,
+      "loss": 0.6009,
+      "step": 11766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.790386468820321e-08,
+      "loss": 0.5901,
+      "step": 11767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.781840451729411e-08,
+      "loss": 0.7614,
+      "step": 11768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.773314806785914e-08,
+      "loss": 0.6317,
+      "step": 11769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7648095346895767e-08,
+      "loss": 0.6849,
+      "step": 11770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.756324636138479e-08,
+      "loss": 0.5929,
+      "step": 11771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7478601118289808e-08,
+      "loss": 0.7652,
+      "step": 11772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.739415962455776e-08,
+      "loss": 0.7367,
+      "step": 11773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7309921887119495e-08,
+      "loss": 0.6131,
+      "step": 11774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.722588791288865e-08,
+      "loss": 0.7685,
+      "step": 11775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7142057708761652e-08,
+      "loss": 0.7441,
+      "step": 11776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.705843128161966e-08,
+      "loss": 0.6532,
+      "step": 11777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.697500863832524e-08,
+      "loss": 0.6653,
+      "step": 11778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6891789785725975e-08,
+      "loss": 0.7104,
+      "step": 11779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.68087747306514e-08,
+      "loss": 0.6817,
+      "step": 11780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.672596347991523e-08,
+      "loss": 0.746,
+      "step": 11781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6643356040313696e-08,
+      "loss": 0.639,
+      "step": 11782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.656095241862693e-08,
+      "loss": 0.6989,
+      "step": 11783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6478752621618132e-08,
+      "loss": 0.695,
+      "step": 11784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6396756656033853e-08,
+      "loss": 0.7388,
+      "step": 11785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.631496452860315e-08,
+      "loss": 0.7429,
+      "step": 11786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6233376246039822e-08,
+      "loss": 0.742,
+      "step": 11787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6151991815039625e-08,
+      "loss": 0.7376,
+      "step": 11788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6070811242281937e-08,
+      "loss": 0.7425,
+      "step": 11789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5989834534430038e-08,
+      "loss": 0.5927,
+      "step": 11790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5909061698129445e-08,
+      "loss": 0.7106,
+      "step": 11791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582849274000986e-08,
+      "loss": 0.7267,
+      "step": 11792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5748127666683765e-08,
+      "loss": 0.557,
+      "step": 11793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5667966484747e-08,
+      "loss": 0.7031,
+      "step": 11794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5588009200778463e-08,
+      "loss": 0.6454,
+      "step": 11795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5508255821340966e-08,
+      "loss": 0.7089,
+      "step": 11796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.542870635297983e-08,
+      "loss": 0.6543,
+      "step": 11797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5349360802223713e-08,
+      "loss": 0.6662,
+      "step": 11798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5270219175585467e-08,
+      "loss": 0.7023,
+      "step": 11799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.519128147956017e-08,
+      "loss": 0.6429,
+      "step": 11800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.511254772062626e-08,
+      "loss": 0.7316,
+      "step": 11801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5034017905246335e-08,
+      "loss": 0.6876,
+      "step": 11802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4955692039864967e-08,
+      "loss": 0.7439,
+      "step": 11803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4877570130911178e-08,
+      "loss": 0.7495,
+      "step": 11804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4799652184796232e-08,
+      "loss": 0.6479,
+      "step": 11805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4721938207915288e-08,
+      "loss": 0.7609,
+      "step": 11806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4644428206646854e-08,
+      "loss": 0.7608,
+      "step": 11807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4567122187352235e-08,
+      "loss": 0.7675,
+      "step": 11808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4490020156376072e-08,
+      "loss": 0.6973,
+      "step": 11809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4413122120046918e-08,
+      "loss": 0.753,
+      "step": 11810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.433642808467528e-08,
+      "loss": 0.7725,
+      "step": 11811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.42599380565564e-08,
+      "loss": 0.74,
+      "step": 11812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4183652041967754e-08,
+      "loss": 0.6145,
+      "step": 11813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4107570047170449e-08,
+      "loss": 0.7393,
+      "step": 11814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4031692078408932e-08,
+      "loss": 0.6421,
+      "step": 11815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.395601814191072e-08,
+      "loss": 0.7054,
+      "step": 11816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3880548243886683e-08,
+      "loss": 0.6501,
+      "step": 11817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.380528239053075e-08,
+      "loss": 0.5906,
+      "step": 11818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3730220588020205e-08,
+      "loss": 0.6579,
+      "step": 11819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.365536284251595e-08,
+      "loss": 0.7779,
+      "step": 11820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3580709160161687e-08,
+      "loss": 0.5923,
+      "step": 11821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3506259547084454e-08,
+      "loss": 0.6349,
+      "step": 11822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3432014009394367e-08,
+      "loss": 0.964,
+      "step": 11823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.335797255318544e-08,
+      "loss": 0.6831,
+      "step": 11824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3284135184534474e-08,
+      "loss": 0.7159,
+      "step": 11825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.321050190950135e-08,
+      "loss": 0.7106,
+      "step": 11826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3137072734129564e-08,
+      "loss": 0.5987,
+      "step": 11827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3063847664445406e-08,
+      "loss": 0.653,
+      "step": 11828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2990826706459347e-08,
+      "loss": 0.7203,
+      "step": 11829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2918009866163817e-08,
+      "loss": 0.6667,
+      "step": 11830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2845397149535421e-08,
+      "loss": 0.7124,
+      "step": 11831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.277298856253384e-08,
+      "loss": 0.8013,
+      "step": 11832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2700784111101816e-08,
+      "loss": 0.6207,
+      "step": 11833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2628783801165445e-08,
+      "loss": 0.8441,
+      "step": 11834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2556987638633888e-08,
+      "loss": 0.7146,
+      "step": 11835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2485395629399932e-08,
+      "loss": 0.6201,
+      "step": 11836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2414007779339155e-08,
+      "loss": 0.6573,
+      "step": 11837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2342824094311034e-08,
+      "loss": 0.8455,
+      "step": 11838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2271844580157566e-08,
+      "loss": 0.6759,
+      "step": 11839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.220106924270409e-08,
+      "loss": 0.6483,
+      "step": 11840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2130498087759846e-08,
+      "loss": 0.6401,
+      "step": 11841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2060131121116591e-08,
+      "loss": 0.683,
+      "step": 11842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.198996834854943e-08,
+      "loss": 0.7658,
+      "step": 11843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1920009775817365e-08,
+      "loss": 0.7394,
+      "step": 11844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1850255408661637e-08,
+      "loss": 0.595,
+      "step": 11845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1780705252807667e-08,
+      "loss": 0.7271,
+      "step": 11846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1711359313963388e-08,
+      "loss": 0.5736,
+      "step": 11847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.164221759782036e-08,
+      "loss": 0.8482,
+      "step": 11848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.157328011005321e-08,
+      "loss": 0.592,
+      "step": 11849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.150454685631991e-08,
+      "loss": 0.7488,
+      "step": 11850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1436017842261782e-08,
+      "loss": 0.6372,
+      "step": 11851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.136769307350294e-08,
+      "loss": 0.6892,
+      "step": 11852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1299572555651672e-08,
+      "loss": 0.6039,
+      "step": 11853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1231656294298231e-08,
+      "loss": 0.6682,
+      "step": 11854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1163944295016772e-08,
+      "loss": 0.6409,
+      "step": 11855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1096436563365065e-08,
+      "loss": 0.741,
+      "step": 11856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1029133104883405e-08,
+      "loss": 0.7654,
+      "step": 11857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0962033925095983e-08,
+      "loss": 0.754,
+      "step": 11858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0895139029509227e-08,
+      "loss": 0.5932,
+      "step": 11859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0828448423614025e-08,
+      "loss": 0.8091,
+      "step": 11860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0761962112883773e-08,
+      "loss": 0.6679,
+      "step": 11861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0695680102774942e-08,
+      "loss": 0.581,
+      "step": 11862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0629602398727901e-08,
+      "loss": 0.6098,
+      "step": 11863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0563729006165813e-08,
+      "loss": 0.6139,
+      "step": 11864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0498059930495186e-08,
+      "loss": 0.9279,
+      "step": 11865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.043259517710532e-08,
+      "loss": 0.6957,
+      "step": 11866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0367334751369418e-08,
+      "loss": 0.7285,
+      "step": 11867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.030227865864375e-08,
+      "loss": 0.8046,
+      "step": 11868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0237426904267656e-08,
+      "loss": 0.8201,
+      "step": 11869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0172779493563544e-08,
+      "loss": 0.6867,
+      "step": 11870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.010833643183745e-08,
+      "loss": 0.7778,
+      "step": 11871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0044097724378199e-08,
+      "loss": 0.7695,
+      "step": 11872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.980063376458515e-09,
+      "loss": 0.6127,
+      "step": 11873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.916233393333641e-09,
+      "loss": 0.7624,
+      "step": 11874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.85260778024244e-09,
+      "loss": 0.6625,
+      "step": 11875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.789186542406848e-09,
+      "loss": 0.5958,
+      "step": 11876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.725969685031867e-09,
+      "loss": 0.6179,
+      "step": 11877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.662957213306401e-09,
+      "loss": 0.7035,
+      "step": 11878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.600149132401592e-09,
+      "loss": 0.6232,
+      "step": 11879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.537545447472484e-09,
+      "loss": 0.6478,
+      "step": 11880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.475146163657467e-09,
+      "loss": 0.6964,
+      "step": 11881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.41295128607772e-09,
+      "loss": 0.6904,
+      "step": 11882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.350960819837496e-09,
+      "loss": 0.6716,
+      "step": 11883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.289174770024944e-09,
+      "loss": 0.7204,
+      "step": 11884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.22759314171101e-09,
+      "loss": 0.65,
+      "step": 11885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.166215939949707e-09,
+      "loss": 0.7812,
+      "step": 11886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.105043169778672e-09,
+      "loss": 0.6666,
+      "step": 11887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.044074836218608e-09,
+      "loss": 0.719,
+      "step": 11888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.983310944273294e-09,
+      "loss": 0.6734,
+      "step": 11889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.922751498929572e-09,
+      "loss": 0.6515,
+      "step": 11890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.862396505158188e-09,
+      "loss": 0.637,
+      "step": 11891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.802245967912682e-09,
+      "loss": 0.746,
+      "step": 11892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.742299892129657e-09,
+      "loss": 0.6123,
+      "step": 11893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.68255828272907e-09,
+      "loss": 0.631,
+      "step": 11894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.623021144614497e-09,
+      "loss": 0.8061,
+      "step": 11895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.56368848267175e-09,
+      "loss": 0.5302,
+      "step": 11896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.504560301770826e-09,
+      "loss": 0.7312,
+      "step": 11897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.445636606764784e-09,
+      "loss": 0.7277,
+      "step": 11898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.386917402489203e-09,
+      "loss": 0.6355,
+      "step": 11899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.328402693763837e-09,
+      "loss": 0.6481,
+      "step": 11900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.270092485390957e-09,
+      "loss": 0.7485,
+      "step": 11901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.211986782156179e-09,
+      "loss": 0.6288,
+      "step": 11902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.154085588828741e-09,
+      "loss": 0.6278,
+      "step": 11903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.096388910160402e-09,
+      "loss": 0.6794,
+      "step": 11904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.038896750887093e-09,
+      "loss": 0.7639,
+      "step": 11905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.981609115726984e-09,
+      "loss": 0.917,
+      "step": 11906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.92452600938215e-09,
+      "loss": 0.5962,
+      "step": 11907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.867647436537173e-09,
+      "loss": 0.7096,
+      "step": 11908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.810973401860822e-09,
+      "loss": 0.7462,
+      "step": 11909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.75450391000382e-09,
+      "loss": 0.6585,
+      "step": 11910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.698238965601624e-09,
+      "loss": 0.6334,
+      "step": 11911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.642178573271653e-09,
+      "loss": 0.6038,
+      "step": 11912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.586322737615226e-09,
+      "loss": 0.5936,
+      "step": 11913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.530671463216177e-09,
+      "loss": 0.608,
+      "step": 11914
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.475224754642519e-09,
+      "loss": 0.687,
+      "step": 11915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.4199826164444985e-09,
+      "loss": 0.6223,
+      "step": 11916
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.364945053156547e-09,
+      "loss": 0.7419,
+      "step": 11917
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.310112069295328e-09,
+      "loss": 0.6205,
+      "step": 11918
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.255483669361407e-09,
+      "loss": 0.7004,
+      "step": 11919
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.201059857838422e-09,
+      "loss": 0.9129,
+      "step": 11920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.146840639192798e-09,
+      "loss": 0.7574,
+      "step": 11921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.092826017874865e-09,
+      "loss": 0.6272,
+      "step": 11922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.039015998317744e-09,
+      "loss": 0.7339,
+      "step": 11923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.985410584937624e-09,
+      "loss": 0.6653,
+      "step": 11924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.932009782134041e-09,
+      "loss": 0.6332,
+      "step": 11925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.87881359428988e-09,
+      "loss": 0.5785,
+      "step": 11926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.825822025771367e-09,
+      "loss": 0.8409,
+      "step": 11927
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.773035080927248e-09,
+      "loss": 0.7134,
+      "step": 11928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.720452764090446e-09,
+      "loss": 0.7937,
+      "step": 11929
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.6680750795763975e-09,
+      "loss": 0.7946,
+      "step": 11930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.615902031683608e-09,
+      "loss": 0.7838,
+      "step": 11931
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.563933624694485e-09,
+      "loss": 0.6567,
+      "step": 11932
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.512169862874229e-09,
+      "loss": 0.7036,
+      "step": 11933
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.460610750471385e-09,
+      "loss": 0.6052,
+      "step": 11934
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.409256291717014e-09,
+      "loss": 0.689,
+      "step": 11935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.358106490826632e-09,
+      "loss": 0.7926,
+      "step": 11936
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.307161351997715e-09,
+      "loss": 0.6522,
+      "step": 11937
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.256420879411917e-09,
+      "loss": 0.6539,
+      "step": 11938
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.205885077233687e-09,
+      "loss": 0.7194,
+      "step": 11939
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.155553949610538e-09,
+      "loss": 0.8653,
+      "step": 11940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.105427500673056e-09,
+      "loss": 0.6985,
+      "step": 11941
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.055505734535727e-09,
+      "loss": 0.8,
+      "step": 11942
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.00578865529583e-09,
+      "loss": 0.6939,
+      "step": 11943
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.956276267033712e-09,
+      "loss": 0.6883,
+      "step": 11944
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.906968573812788e-09,
+      "loss": 0.6798,
+      "step": 11945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8578655796801e-09,
+      "loss": 0.6917,
+      "step": 11946
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.808967288665757e-09,
+      "loss": 0.8147,
+      "step": 11947
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.760273704782937e-09,
+      "loss": 0.6439,
+      "step": 11948
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.711784832028166e-09,
+      "loss": 0.7744,
+      "step": 11949
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.663500674381039e-09,
+      "loss": 0.5658,
+      "step": 11950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.615421235804497e-09,
+      "loss": 0.6709,
+      "step": 11951
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.567546520244271e-09,
+      "loss": 0.7545,
+      "step": 11952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.519876531630275e-09,
+      "loss": 0.7328,
+      "step": 11953
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4724112738743785e-09,
+      "loss": 0.8887,
+      "step": 11954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4251507508723545e-09,
+      "loss": 0.8058,
+      "step": 11955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.378094966503045e-09,
+      "loss": 0.7122,
+      "step": 11956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.33124392462836e-09,
+      "loss": 0.7066,
+      "step": 11957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.284597629093835e-09,
+      "loss": 0.7904,
+      "step": 11958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.238156083727797e-09,
+      "loss": 0.6979,
+      "step": 11959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.191919292341918e-09,
+      "loss": 0.7559,
+      "step": 11960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.145887258730664e-09,
+      "loss": 0.6731,
+      "step": 11961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.100059986672401e-09,
+      "loss": 0.6582,
+      "step": 11962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.054437479928287e-09,
+      "loss": 0.7037,
+      "step": 11963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.009019742242827e-09,
+      "loss": 0.6491,
+      "step": 11964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9638067773433165e-09,
+      "loss": 0.6768,
+      "step": 11965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.918798588940399e-09,
+      "loss": 0.7011,
+      "step": 11966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.873995180728897e-09,
+      "loss": 0.7328,
+      "step": 11967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.829396556385036e-09,
+      "loss": 0.7416,
+      "step": 11968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.785002719569776e-09,
+      "loss": 0.7402,
+      "step": 11969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7408136739263144e-09,
+      "loss": 0.6542,
+      "step": 11970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.69682942308175e-09,
+      "loss": 0.6208,
+      "step": 11971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.653049970645696e-09,
+      "loss": 0.789,
+      "step": 11972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6094753202113876e-09,
+      "loss": 0.7224,
+      "step": 11973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.566105475355409e-09,
+      "loss": 0.8416,
+      "step": 11974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.522940439637135e-09,
+      "loss": 0.7513,
+      "step": 11975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4799802165990116e-09,
+      "loss": 0.8135,
+      "step": 11976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.437224809767382e-09,
+      "loss": 0.682,
+      "step": 11977
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.39467422265083e-09,
+      "loss": 0.573,
+      "step": 11978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3523284587421165e-09,
+      "loss": 0.6191,
+      "step": 11979
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.31018752151624e-09,
+      "loss": 0.6764,
+      "step": 11980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.268251414432378e-09,
+      "loss": 0.7139,
+      "step": 11981
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2265201409319445e-09,
+      "loss": 0.711,
+      "step": 11982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.184993704439977e-09,
+      "loss": 0.71,
+      "step": 11983
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.143672108365138e-09,
+      "loss": 0.7756,
+      "step": 11984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1025553560983255e-09,
+      "loss": 0.607,
+      "step": 11985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.061643451014619e-09,
+      "loss": 0.8019,
+      "step": 11986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.020936396471331e-09,
+      "loss": 0.7354,
+      "step": 11987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.980434195809679e-09,
+      "loss": 0.6397,
+      "step": 11988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.940136852353948e-09,
+      "loss": 0.6899,
+      "step": 11989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.900044369411215e-09,
+      "loss": 0.6826,
+      "step": 11990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.860156750272181e-09,
+      "loss": 0.6703,
+      "step": 11991
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.820473998210617e-09,
+      "loss": 0.7988,
+      "step": 11992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.780996116483082e-09,
+      "loss": 0.742,
+      "step": 11993
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.741723108330042e-09,
+      "loss": 0.889,
+      "step": 11994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.702654976974751e-09,
+      "loss": 0.6704,
+      "step": 11995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6637917256235332e-09,
+      "loss": 0.7216,
+      "step": 11996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.625133357465782e-09,
+      "loss": 0.7102,
+      "step": 11997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5866798756750698e-09,
+      "loss": 0.7336,
+      "step": 11998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5484312834066505e-09,
+      "loss": 0.7875,
+      "step": 11999
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5103875837999568e-09,
+      "loss": 0.7325,
+      "step": 12000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4725487799774915e-09,
+      "loss": 0.7269,
+      "step": 12001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4349148750448257e-09,
+      "loss": 0.6056,
+      "step": 12002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3974858720906e-09,
+      "loss": 0.6562,
+      "step": 12003
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3602617741868015e-09,
+      "loss": 0.685,
+      "step": 12004
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3232425843884865e-09,
+      "loss": 0.8051,
+      "step": 12005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2864283057340574e-09,
+      "loss": 0.6581,
+      "step": 12006
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2498189412449867e-09,
+      "loss": 0.6218,
+      "step": 12007
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.213414493925815e-09,
+      "loss": 0.636,
+      "step": 12008
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.17721496676443e-09,
+      "loss": 0.7225,
+      "step": 12009
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.141220362732067e-09,
+      "loss": 0.7536,
+      "step": 12010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.105430684782473e-09,
+      "loss": 0.7205,
+      "step": 12011
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0698459358532996e-09,
+      "loss": 0.7363,
+      "step": 12012
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.034466118865265e-09,
+      "loss": 0.7212,
+      "step": 12013
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9992912367218795e-09,
+      "loss": 0.8221,
+      "step": 12014
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.964321292310279e-09,
+      "loss": 0.653,
+      "step": 12015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9295562885003904e-09,
+      "loss": 0.716,
+      "step": 12016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.894996228145486e-09,
+      "loss": 0.6868,
+      "step": 12017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8606411140821855e-09,
+      "loss": 0.7535,
+      "step": 12018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.826490949129901e-09,
+      "loss": 0.7221,
+      "step": 12019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.792545736091945e-09,
+      "loss": 0.7974,
+      "step": 12020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.75880547775359e-09,
+      "loss": 0.8209,
+      "step": 12021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7252701768848424e-09,
+      "loss": 0.7078,
+      "step": 12022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6919398362373893e-09,
+      "loss": 0.8564,
+      "step": 12023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.658814458547099e-09,
+      "loss": 0.6725,
+      "step": 12024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.625894046532629e-09,
+      "loss": 0.564,
+      "step": 12025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5931786028957073e-09,
+      "loss": 0.7019,
+      "step": 12026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.560668130321686e-09,
+      "loss": 0.6875,
+      "step": 12027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.528362631478709e-09,
+      "loss": 0.7148,
+      "step": 12028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4962621090179885e-09,
+      "loss": 0.6933,
+      "step": 12029
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4643665655746384e-09,
+      "loss": 0.7246,
+      "step": 12030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4326760037657326e-09,
+      "loss": 0.7795,
+      "step": 12031
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4011904261930784e-09,
+      "loss": 0.7433,
+      "step": 12032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3699098354398875e-09,
+      "loss": 0.6367,
+      "step": 12033
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.338834234074383e-09,
+      "loss": 0.7312,
+      "step": 12034
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3079636246461924e-09,
+      "loss": 0.7683,
+      "step": 12035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.277298009689677e-09,
+      "loss": 0.7863,
+      "step": 12036
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.246837391721157e-09,
+      "loss": 0.6584,
+      "step": 12037
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.216581773241133e-09,
+      "loss": 0.6174,
+      "step": 12038
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.186531156732341e-09,
+      "loss": 0.6116,
+      "step": 12039
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1566855446614187e-09,
+      "loss": 0.6165,
+      "step": 12040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1270449394777957e-09,
+      "loss": 0.7622,
+      "step": 12041
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0976093436139712e-09,
+      "loss": 0.7731,
+      "step": 12042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0683787594863445e-09,
+      "loss": 0.6604,
+      "step": 12043
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0393531894935536e-09,
+      "loss": 0.6371,
+      "step": 12044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010532636017859e-09,
+      "loss": 0.762,
+      "step": 12045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9819171014248683e-09,
+      "loss": 0.7517,
+      "step": 12046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.953506588062981e-09,
+      "loss": 0.7704,
+      "step": 12047
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9253010982642205e-09,
+      "loss": 0.6178,
+      "step": 12048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8973006343431244e-09,
+      "loss": 0.7782,
+      "step": 12049
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8695051985981317e-09,
+      "loss": 0.6827,
+      "step": 12050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8419147933101956e-09,
+      "loss": 0.6856,
+      "step": 12051
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.814529420744171e-09,
+      "loss": 0.6285,
+      "step": 12052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7873490831471495e-09,
+      "loss": 0.5791,
+      "step": 12053
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7603737827504019e-09,
+      "loss": 0.6389,
+      "step": 12054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7336035217677127e-09,
+      "loss": 0.7729,
+      "step": 12055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7070383023959359e-09,
+      "loss": 0.7786,
+      "step": 12056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6806781268158267e-09,
+      "loss": 0.7461,
+      "step": 12057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.654522997190655e-09,
+      "loss": 0.6198,
+      "step": 12058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6285729156670372e-09,
+      "loss": 0.7208,
+      "step": 12059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602827884374658e-09,
+      "loss": 0.6345,
+      "step": 12060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5772879054268276e-09,
+      "loss": 0.6972,
+      "step": 12061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.551952980919369e-09,
+      "loss": 0.6681,
+      "step": 12062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5268231129320078e-09,
+      "loss": 0.7206,
+      "step": 12063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5018983035267055e-09,
+      "loss": 0.7311,
+      "step": 12064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.477178554749603e-09,
+      "loss": 0.7109,
+      "step": 12065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4526638686293558e-09,
+      "loss": 0.8285,
+      "step": 12066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.428354247178243e-09,
+      "loss": 0.683,
+      "step": 12067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4042496923907801e-09,
+      "loss": 0.7328,
+      "step": 12068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3803502062459394e-09,
+      "loss": 0.7382,
+      "step": 12069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3566557907049306e-09,
+      "loss": 0.6816,
+      "step": 12070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.333166447712586e-09,
+      "loss": 0.6311,
+      "step": 12071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.309882179196531e-09,
+      "loss": 0.7323,
+      "step": 12072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2868029870682918e-09,
+      "loss": 0.6989,
+      "step": 12073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.263928873221354e-09,
+      "loss": 0.734,
+      "step": 12074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2412598395336594e-09,
+      "loss": 0.7226,
+      "step": 12075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.218795887865387e-09,
+      "loss": 0.7543,
+      "step": 12076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1965370200606175e-09,
+      "loss": 0.7456,
+      "step": 12077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1744832379456672e-09,
+      "loss": 0.6384,
+      "step": 12078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1526345433310327e-09,
+      "loss": 0.6664,
+      "step": 12079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.130990938009724e-09,
+      "loss": 0.6506,
+      "step": 12080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1095524237580978e-09,
+      "loss": 0.5857,
+      "step": 12081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0883190023355806e-09,
+      "loss": 0.6174,
+      "step": 12082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0672906754852218e-09,
+      "loss": 0.6842,
+      "step": 12083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0464674449325863e-09,
+      "loss": 0.7317,
+      "step": 12084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0258493123868619e-09,
+      "loss": 0.6564,
+      "step": 12085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0054362795403061e-09,
+      "loss": 0.6084,
+      "step": 12086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.852283480679681e-10,
+      "loss": 0.7539,
+      "step": 12087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.652255196287984e-10,
+      "loss": 0.7371,
+      "step": 12088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.454277958642622e-10,
+      "loss": 0.7517,
+      "step": 12089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.258351783991704e-10,
+      "loss": 0.6265,
+      "step": 12090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.064476688416813e-10,
+      "loss": 0.62,
+      "step": 12091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.872652687830219e-10,
+      "loss": 0.9293,
+      "step": 12092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.682879797974886e-10,
+      "loss": 0.7169,
+      "step": 12093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.495158034427242e-10,
+      "loss": 0.8044,
+      "step": 12094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.309487412591633e-10,
+      "loss": 0.7333,
+      "step": 12095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.125867947711418e-10,
+      "loss": 0.7165,
+      "step": 12096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.944299654852328e-10,
+      "loss": 0.7417,
+      "step": 12097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.764782548916328e-10,
+      "loss": 0.6753,
+      "step": 12098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.587316644641629e-10,
+      "loss": 0.6252,
+      "step": 12099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.411901956588807e-10,
+      "loss": 0.6687,
+      "step": 12100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.23853849915468e-10,
+      "loss": 0.754,
+      "step": 12101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.067226286572304e-10,
+      "loss": 0.7335,
+      "step": 12102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.89796533289988e-10,
+      "loss": 0.8085,
+      "step": 12103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7307556520263e-10,
+      "loss": 0.6604,
+      "step": 12104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.565597257679468e-10,
+      "loss": 0.6983,
+      "step": 12105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.402490163409659e-10,
+      "loss": 0.6494,
+      "step": 12106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.241434382608935e-10,
+      "loss": 0.6012,
+      "step": 12107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.082429928491729e-10,
+      "loss": 0.7234,
+      "step": 12108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.925476814108711e-10,
+      "loss": 0.797,
+      "step": 12109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.770575052344018e-10,
+      "loss": 0.6696,
+      "step": 12110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.617724655909706e-10,
+      "loss": 0.696,
+      "step": 12111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.466925637348519e-10,
+      "loss": 0.6946,
+      "step": 12112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.318178009042219e-10,
+      "loss": 0.598,
+      "step": 12113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.171481783192156e-10,
+      "loss": 0.7887,
+      "step": 12114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.026836971844252e-10,
+      "loss": 0.7471,
+      "step": 12115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.884243586866788e-10,
+      "loss": 0.692,
+      "step": 12116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.743701639964293e-10,
+      "loss": 0.7306,
+      "step": 12117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6052111426692083e-10,
+      "loss": 0.6903,
+      "step": 12118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4687721063502164e-10,
+      "loss": 0.6074,
+      "step": 12119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.334384542206693e-10,
+      "loss": 0.6204,
+      "step": 12120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2020484612659283e-10,
+      "loss": 0.6559,
+      "step": 12121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.071763874388679e-10,
+      "loss": 0.625,
+      "step": 12122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.943530792269168e-10,
+      "loss": 0.6445,
+      "step": 12123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.81734922543231e-10,
+      "loss": 0.5497,
+      "step": 12124
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6932191842337095e-10,
+      "loss": 0.7318,
+      "step": 12125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5711406788596635e-10,
+      "loss": 0.6909,
+      "step": 12126
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4511137193327103e-10,
+      "loss": 0.7184,
+      "step": 12127
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3331383155033035e-10,
+      "loss": 0.5969,
+      "step": 12128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2172144770498124e-10,
+      "loss": 0.7798,
+      "step": 12129
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1033422134924e-10,
+      "loss": 0.6086,
+      "step": 12130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9915215341735914e-10,
+      "loss": 1.0291,
+      "step": 12131
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.881752448272157e-10,
+      "loss": 0.7041,
+      "step": 12132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.774034964794781e-10,
+      "loss": 0.7037,
+      "step": 12133
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6683690925843886e-10,
+      "loss": 0.5795,
+      "step": 12134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5647548403145983e-10,
+      "loss": 0.6469,
+      "step": 12135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.463192216486943e-10,
+      "loss": 0.813,
+      "step": 12136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3636812294391965e-10,
+      "loss": 0.7451,
+      "step": 12137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.26622188733705e-10,
+      "loss": 0.7156,
+      "step": 12138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1708141981796604e-10,
+      "loss": 0.6788,
+      "step": 12139
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0774581697968754e-10,
+      "loss": 0.64,
+      "step": 12140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9861538098520094e-10,
+      "loss": 0.7822,
+      "step": 12141
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8969011258362923e-10,
+      "loss": 0.6555,
+      "step": 12142
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8097001250799718e-10,
+      "loss": 0.6707,
+      "step": 12143
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7245508147356593e-10,
+      "loss": 0.626,
+      "step": 12144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6414532017922092e-10,
+      "loss": 0.8251,
+      "step": 12145
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5604072930691662e-10,
+      "loss": 0.721,
+      "step": 12146
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.481413095222317e-10,
+      "loss": 0.6671,
+      "step": 12147
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4044706147325892e-10,
+      "loss": 0.6631,
+      "step": 12148
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3295798579116004e-10,
+      "loss": 0.7274,
+      "step": 12149
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.256740830909986e-10,
+      "loss": 0.762,
+      "step": 12150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1859535397062971e-10,
+      "loss": 0.5832,
+      "step": 12151
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1172179901097757e-10,
+      "loss": 0.635,
+      "step": 12152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0505341877575792e-10,
+      "loss": 0.7071,
+      "step": 12153
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.859021381286582e-11,
+      "loss": 0.5705,
+      "step": 12154
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.233218465243276e-11,
+      "loss": 0.6833,
+      "step": 12155
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.627933180793691e-11,
+      "loss": 0.7816,
+      "step": 12156
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.04316557764806e-11,
+      "loss": 0.6692,
+      "step": 12157
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.478915703795775e-11,
+      "loss": 0.5681,
+      "step": 12158
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.935183605533135e-11,
+      "loss": 0.6996,
+      "step": 12159
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.411969327491108e-11,
+      "loss": 0.7355,
+      "step": 12160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.909272912607567e-11,
+      "loss": 0.7055,
+      "step": 12161
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.427094402155053e-11,
+      "loss": 0.7225,
+      "step": 12162
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9654338356852626e-11,
+      "loss": 0.7267,
+      "step": 12163
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.524291251112312e-11,
+      "loss": 0.7042,
+      "step": 12164
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1036666846294704e-11,
+      "loss": 0.6294,
+      "step": 12165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.703560170764675e-11,
+      "loss": 0.7053,
+      "step": 12166
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3239717423527716e-11,
+      "loss": 0.5941,
+      "step": 12167
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.964901430535516e-11,
+      "loss": 0.6974,
+      "step": 12168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.626349264817085e-11,
+      "loss": 0.7816,
+      "step": 12169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3083152729530545e-11,
+      "loss": 0.7473,
+      "step": 12170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.010799481033665e-11,
+      "loss": 0.6571,
+      "step": 12171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.733801913511579e-11,
+      "loss": 0.731,
+      "step": 12172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.477322593118613e-11,
+      "loss": 0.6662,
+      "step": 12173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.241361540893493e-11,
+      "loss": 0.6257,
+      "step": 12174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0259187761818556e-11,
+      "loss": 0.5782,
+      "step": 12175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.309943166917578e-12,
+      "loss": 0.7574,
+      "step": 12176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.565881784104111e-12,
+      "loss": 0.6059,
+      "step": 12177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0270037568744824e-12,
+      "loss": 0.7379,
+      "step": 12178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.693309210961449e-12,
+      "loss": 0.6935,
+      "step": 12179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5647982559995345e-12,
+      "loss": 0.6782,
+      "step": 12180
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.8010936379432678,
+      "eval_runtime": 437.2975,
+      "eval_samples_per_second": 6.858,
+      "eval_steps_per_second": 0.858,
+      "step": 12180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6414709849699173e-12,
+      "loss": 0.8354,
+      "step": 12181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.233274733677633e-13,
+      "loss": 0.6146,
+      "step": 12182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1036778003489266e-13,
+      "loss": 0.657,
+      "step": 12183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0259194715978027e-13,
+      "loss": 0.7198,
+      "step": 12184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8385,
+      "step": 12185
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 12185,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 5.179326738878546e+18,
+  "trial_name": null,
+  "trial_params": null
+}