diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32617 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 5432,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.226993865030675e-07,
+      "loss": 5.6191,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.45398773006135e-07,
+      "loss": 5.459,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.680981595092025e-07,
+      "loss": 5.5078,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9079754601227e-07,
+      "loss": 5.4121,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.134969325153375e-07,
+      "loss": 5.4102,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.36196319018405e-07,
+      "loss": 5.4805,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.588957055214725e-07,
+      "loss": 5.375,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.8159509202454e-07,
+      "loss": 5.0977,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1042944785276075e-06,
+      "loss": 5.0977,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.226993865030675e-06,
+      "loss": 4.0967,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3496932515337425e-06,
+      "loss": 4.0986,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.47239263803681e-06,
+      "loss": 3.9102,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5950920245398775e-06,
+      "loss": 2.9072,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.717791411042945e-06,
+      "loss": 2.7236,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8404907975460124e-06,
+      "loss": 2.7129,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.96319018404908e-06,
+      "loss": 2.6094,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.085889570552147e-06,
+      "loss": 2.5098,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.208588957055215e-06,
+      "loss": 2.0381,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.331288343558282e-06,
+      "loss": 1.8618,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.45398773006135e-06,
+      "loss": 1.814,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5766871165644175e-06,
+      "loss": 1.7739,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.699386503067485e-06,
+      "loss": 1.731,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.822085889570552e-06,
+      "loss": 1.6543,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.94478527607362e-06,
+      "loss": 1.4785,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0674846625766875e-06,
+      "loss": 1.3569,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.190184049079755e-06,
+      "loss": 1.3218,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.312883435582822e-06,
+      "loss": 1.2734,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.43558282208589e-06,
+      "loss": 1.1904,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5582822085889574e-06,
+      "loss": 1.1934,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.680981595092025e-06,
+      "loss": 1.1431,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8036809815950928e-06,
+      "loss": 1.1094,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.92638036809816e-06,
+      "loss": 1.1187,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.049079754601227e-06,
+      "loss": 1.0376,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.171779141104294e-06,
+      "loss": 1.0481,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.294478527607362e-06,
+      "loss": 0.9932,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.41717791411043e-06,
+      "loss": 0.9639,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.539877300613497e-06,
+      "loss": 1.0088,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.662576687116564e-06,
+      "loss": 0.9438,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.785276073619632e-06,
+      "loss": 0.9041,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9079754601227e-06,
+      "loss": 0.9097,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.030674846625767e-06,
+      "loss": 0.9021,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.153374233128835e-06,
+      "loss": 0.8604,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.276073619631902e-06,
+      "loss": 0.8594,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.39877300613497e-06,
+      "loss": 0.8293,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.521472392638038e-06,
+      "loss": 0.8284,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.644171779141104e-06,
+      "loss": 0.8594,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.766871165644172e-06,
+      "loss": 0.8159,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.88957055214724e-06,
+      "loss": 0.8164,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.012269938650307e-06,
+      "loss": 0.7974,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.134969325153375e-06,
+      "loss": 0.7915,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.257668711656443e-06,
+      "loss": 0.7668,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.38036809815951e-06,
+      "loss": 0.7778,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.503067484662578e-06,
+      "loss": 0.7979,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.625766871165644e-06,
+      "loss": 0.7427,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.748466257668712e-06,
+      "loss": 0.7512,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.87116564417178e-06,
+      "loss": 0.729,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.993865030674847e-06,
+      "loss": 0.7603,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.116564417177915e-06,
+      "loss": 0.7236,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.239263803680983e-06,
+      "loss": 0.6987,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.36196319018405e-06,
+      "loss": 0.7061,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.484662576687118e-06,
+      "loss": 0.6929,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6073619631901856e-06,
+      "loss": 0.7073,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.730061349693252e-06,
+      "loss": 0.7019,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.85276073619632e-06,
+      "loss": 0.7009,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.975460122699386e-06,
+      "loss": 0.7126,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.098159509202455e-06,
+      "loss": 0.6753,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.220858895705522e-06,
+      "loss": 0.656,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.343558282208589e-06,
+      "loss": 0.6504,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.466257668711658e-06,
+      "loss": 0.6499,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.588957055214725e-06,
+      "loss": 0.6697,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.711656441717792e-06,
+      "loss": 0.6562,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.83435582822086e-06,
+      "loss": 0.6636,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.957055214723927e-06,
+      "loss": 0.6799,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.079754601226994e-06,
+      "loss": 0.6697,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.202453987730062e-06,
+      "loss": 0.6589,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.325153374233129e-06,
+      "loss": 0.6626,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.447852760736197e-06,
+      "loss": 0.6313,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.570552147239264e-06,
+      "loss": 0.6672,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.693251533742331e-06,
+      "loss": 0.6584,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.8159509202454e-06,
+      "loss": 0.6198,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.938650306748467e-06,
+      "loss": 0.6443,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0061349693251534e-05,
+      "loss": 0.6201,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0184049079754601e-05,
+      "loss": 0.6116,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.030674846625767e-05,
+      "loss": 0.6042,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0429447852760737e-05,
+      "loss": 0.6177,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0552147239263804e-05,
+      "loss": 0.6206,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0674846625766873e-05,
+      "loss": 0.601,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.079754601226994e-05,
+      "loss": 0.6265,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0920245398773005e-05,
+      "loss": 0.6072,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1042944785276076e-05,
+      "loss": 0.6061,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1165644171779141e-05,
+      "loss": 0.6177,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1288343558282208e-05,
+      "loss": 0.6221,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1411042944785277e-05,
+      "loss": 0.6115,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1533742331288344e-05,
+      "loss": 0.5967,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1656441717791411e-05,
+      "loss": 0.6177,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.177914110429448e-05,
+      "loss": 0.5868,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1901840490797547e-05,
+      "loss": 0.6049,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2024539877300614e-05,
+      "loss": 0.5989,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2147239263803683e-05,
+      "loss": 0.5823,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.226993865030675e-05,
+      "loss": 0.5879,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2392638036809817e-05,
+      "loss": 0.571,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2515337423312886e-05,
+      "loss": 0.5649,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2638036809815953e-05,
+      "loss": 0.5782,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.276073619631902e-05,
+      "loss": 0.5922,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2883435582822085e-05,
+      "loss": 0.5791,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3006134969325156e-05,
+      "loss": 0.5771,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3128834355828221e-05,
+      "loss": 0.5918,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3251533742331288e-05,
+      "loss": 0.5508,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3374233128834357e-05,
+      "loss": 0.5767,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3496932515337424e-05,
+      "loss": 0.5599,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3619631901840491e-05,
+      "loss": 0.6125,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.374233128834356e-05,
+      "loss": 0.5693,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3865030674846627e-05,
+      "loss": 0.577,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3987730061349694e-05,
+      "loss": 0.5549,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4110429447852763e-05,
+      "loss": 0.5734,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.423312883435583e-05,
+      "loss": 0.5967,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4355828220858897e-05,
+      "loss": 0.5316,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4478527607361965e-05,
+      "loss": 0.5511,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4601226993865032e-05,
+      "loss": 0.5439,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.47239263803681e-05,
+      "loss": 0.5718,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4846625766871168e-05,
+      "loss": 0.5605,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4969325153374235e-05,
+      "loss": 0.5261,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.50920245398773e-05,
+      "loss": 0.5425,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5214723926380371e-05,
+      "loss": 0.5587,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5337423312883436e-05,
+      "loss": 0.5302,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5460122699386504e-05,
+      "loss": 0.5448,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5582822085889574e-05,
+      "loss": 0.5402,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.570552147239264e-05,
+      "loss": 0.5444,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5828220858895708e-05,
+      "loss": 0.5311,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5950920245398772e-05,
+      "loss": 0.532,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6073619631901842e-05,
+      "loss": 0.5441,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.619631901840491e-05,
+      "loss": 0.5278,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6319018404907976e-05,
+      "loss": 0.5342,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6441717791411043e-05,
+      "loss": 0.5269,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.656441717791411e-05,
+      "loss": 0.5259,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6687116564417178e-05,
+      "loss": 0.5171,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6809815950920248e-05,
+      "loss": 0.512,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6932515337423315e-05,
+      "loss": 0.5437,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7055214723926382e-05,
+      "loss": 0.5323,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.717791411042945e-05,
+      "loss": 0.5145,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7300613496932516e-05,
+      "loss": 0.5315,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7423312883435583e-05,
+      "loss": 0.5255,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7546012269938654e-05,
+      "loss": 0.5074,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.766871165644172e-05,
+      "loss": 0.5444,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7791411042944788e-05,
+      "loss": 0.5254,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7914110429447855e-05,
+      "loss": 0.5217,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8036809815950922e-05,
+      "loss": 0.5194,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.815950920245399e-05,
+      "loss": 0.5201,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.828220858895706e-05,
+      "loss": 0.5082,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8404907975460123e-05,
+      "loss": 0.5293,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.852760736196319e-05,
+      "loss": 0.5054,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8650306748466257e-05,
+      "loss": 0.5289,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8773006134969328e-05,
+      "loss": 0.5063,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8895705521472395e-05,
+      "loss": 0.5043,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9018404907975462e-05,
+      "loss": 0.5327,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.914110429447853e-05,
+      "loss": 0.5205,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9263803680981596e-05,
+      "loss": 0.5114,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9386503067484663e-05,
+      "loss": 0.498,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9509202453987733e-05,
+      "loss": 0.5044,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.96319018404908e-05,
+      "loss": 0.4749,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9754601226993868e-05,
+      "loss": 0.5128,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9877300613496935e-05,
+      "loss": 0.489,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5159,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998222484733e-05,
+      "loss": 0.514,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992889939564e-05,
+      "loss": 0.5018,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999984002366384e-05,
+      "loss": 0.4913,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999971559768356e-05,
+      "loss": 0.4934,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999955562149907e-05,
+      "loss": 0.5024,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999936009516717e-05,
+      "loss": 0.4919,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999912901875743e-05,
+      "loss": 0.5031,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999886239235193e-05,
+      "loss": 0.4999,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999856021604556e-05,
+      "loss": 0.4739,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999822248994562e-05,
+      "loss": 0.5088,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999978492141723e-05,
+      "loss": 0.5106,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999974403888582e-05,
+      "loss": 0.5088,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999969960141487e-05,
+      "loss": 0.4948,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999651609020176e-05,
+      "loss": 0.4825,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999600061718806e-05,
+      "loss": 0.4823,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999544959529074e-05,
+      "loss": 0.4822,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999948630247058e-05,
+      "loss": 0.4733,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999942409056417e-05,
+      "loss": 0.4766,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999935832383196e-05,
+      "loss": 0.4667,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999289002297333e-05,
+      "loss": 0.4929,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999216125984937e-05,
+      "loss": 0.4807,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999913969492067e-05,
+      "loss": 0.4757,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999905970913171e-05,
+      "loss": 0.4626,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999897616864649e-05,
+      "loss": 0.4583,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999888907349471e-05,
+      "loss": 0.4926,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998798423707337e-05,
+      "loss": 0.4951,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999870421931659e-05,
+      "loss": 0.4823,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998606460355964e-05,
+      "loss": 0.4736,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998505146860204e-05,
+      "loss": 0.4755,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999840027886534e-05,
+      "loss": 0.4663,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998291856408643e-05,
+      "loss": 0.4695,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998179879528662e-05,
+      "loss": 0.4585,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99980643482652e-05,
+      "loss": 0.4623,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999794526265934e-05,
+      "loss": 0.4744,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997822622753406e-05,
+      "loss": 0.4608,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997696428591e-05,
+      "loss": 0.4734,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997566680216988e-05,
+      "loss": 0.4501,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997433377677493e-05,
+      "loss": 0.473,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997296521019905e-05,
+      "loss": 0.4749,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997156110292873e-05,
+      "loss": 0.4813,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997012145546323e-05,
+      "loss": 0.4764,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996864626831428e-05,
+      "loss": 0.4806,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996713554200627e-05,
+      "loss": 0.4777,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996558927707637e-05,
+      "loss": 0.4749,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999640074740742e-05,
+      "loss": 0.4529,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996239013356216e-05,
+      "loss": 0.4784,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999607372561152e-05,
+      "loss": 0.4801,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999590488423209e-05,
+      "loss": 0.4733,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995732489277947e-05,
+      "loss": 0.4677,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999555654081039e-05,
+      "loss": 0.4691,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995377038891954e-05,
+      "loss": 0.4794,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999519398358646e-05,
+      "loss": 0.4738,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995007374958985e-05,
+      "loss": 0.4641,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999481721307587e-05,
+      "loss": 0.427,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994623498004717e-05,
+      "loss": 0.4579,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994426229814388e-05,
+      "loss": 0.4496,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994225408575017e-05,
+      "loss": 0.4711,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994021034357997e-05,
+      "loss": 0.4604,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999381310723598e-05,
+      "loss": 0.4596,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999360162728289e-05,
+      "loss": 0.459,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993386594573906e-05,
+      "loss": 0.4602,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993168009185472e-05,
+      "loss": 0.4497,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992945871195293e-05,
+      "loss": 0.4427,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992720180682347e-05,
+      "loss": 0.4678,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999249093772686e-05,
+      "loss": 0.457,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992258142410335e-05,
+      "loss": 0.4539,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992021794815526e-05,
+      "loss": 0.4927,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999178189502646e-05,
+      "loss": 0.4486,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999153844312842e-05,
+      "loss": 0.4528,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991291439207954e-05,
+      "loss": 0.4598,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999104088335287e-05,
+      "loss": 0.4417,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990786775652248e-05,
+      "loss": 0.48,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999052911619642e-05,
+      "loss": 0.4895,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990267905076985e-05,
+      "loss": 0.4659,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990003142386802e-05,
+      "loss": 0.4332,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998973482822e-05,
+      "loss": 0.4535,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989462962671957e-05,
+      "loss": 0.4546,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998918754583933e-05,
+      "loss": 0.4673,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998890857782003e-05,
+      "loss": 0.4701,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998862605871323e-05,
+      "loss": 0.458,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988339988619363e-05,
+      "loss": 0.4305,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988050367640132e-05,
+      "loss": 0.4584,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987757195878498e-05,
+      "loss": 0.476,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987460473438676e-05,
+      "loss": 0.4518,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987160200426167e-05,
+      "loss": 0.4554,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986856376947706e-05,
+      "loss": 0.4484,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986549003111308e-05,
+      "loss": 0.4521,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998623807902625e-05,
+      "loss": 0.4521,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985923604803053e-05,
+      "loss": 0.4452,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985605580553527e-05,
+      "loss": 0.4597,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985284006390728e-05,
+      "loss": 0.452,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998495888242897e-05,
+      "loss": 0.4386,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998463020878384e-05,
+      "loss": 0.4508,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998429798557219e-05,
+      "loss": 0.4385,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998396221291211e-05,
+      "loss": 0.4413,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998362289092298e-05,
+      "loss": 0.4612,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983280019725426e-05,
+      "loss": 0.4441,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982933599441342e-05,
+      "loss": 0.4475,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982583630193882e-05,
+      "loss": 0.4525,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982230112107455e-05,
+      "loss": 0.4497,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981873045307747e-05,
+      "loss": 0.432,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981512429921687e-05,
+      "loss": 0.4674,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981148266077483e-05,
+      "loss": 0.4576,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998078055390459e-05,
+      "loss": 0.4456,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980409293533736e-05,
+      "loss": 0.455,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99800344850969e-05,
+      "loss": 0.4336,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997965612872733e-05,
+      "loss": 0.4106,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979274224559536e-05,
+      "loss": 0.429,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978888772729282e-05,
+      "loss": 0.4297,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978499773373596e-05,
+      "loss": 0.4399,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978107226630774e-05,
+      "loss": 0.4441,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977711132640362e-05,
+      "loss": 0.447,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997731149154318e-05,
+      "loss": 0.4536,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997690830348129e-05,
+      "loss": 0.4386,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976501568598036e-05,
+      "loss": 0.4342,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976091287038014e-05,
+      "loss": 0.4166,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975677458947072e-05,
+      "loss": 0.4462,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975260084472338e-05,
+      "loss": 0.4413,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997483916376218e-05,
+      "loss": 0.4271,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974414696966242e-05,
+      "loss": 0.446,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997398668423542e-05,
+      "loss": 0.4502,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973555125721876e-05,
+      "loss": 0.4393,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973120021579037e-05,
+      "loss": 0.4182,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997268137196157e-05,
+      "loss": 0.4382,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972239177025425e-05,
+      "loss": 0.4384,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971793436927802e-05,
+      "loss": 0.4496,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971344151827166e-05,
+      "loss": 0.4326,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970891321883234e-05,
+      "loss": 0.4187,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997043494725699e-05,
+      "loss": 0.4534,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969975028110677e-05,
+      "loss": 0.4249,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969511564607798e-05,
+      "loss": 0.4288,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969044556913116e-05,
+      "loss": 0.4541,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968574005192654e-05,
+      "loss": 0.4438,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968099909613693e-05,
+      "loss": 0.4346,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967622270344777e-05,
+      "loss": 0.4293,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967141087555708e-05,
+      "loss": 0.4218,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966656361417547e-05,
+      "loss": 0.4464,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966168092102613e-05,
+      "loss": 0.4353,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965676279784495e-05,
+      "loss": 0.4314,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965180924638026e-05,
+      "loss": 0.4325,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996468202683931e-05,
+      "loss": 0.4412,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964179586565705e-05,
+      "loss": 0.416,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963673603995833e-05,
+      "loss": 0.4165,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996316407930957e-05,
+      "loss": 0.4263,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962651012688053e-05,
+      "loss": 0.4404,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996213440431368e-05,
+      "loss": 0.4373,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961614254370107e-05,
+      "loss": 0.4301,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961090563042248e-05,
+      "loss": 0.4115,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960563330516276e-05,
+      "loss": 0.4203,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996003255697963e-05,
+      "loss": 0.4301,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959498242620992e-05,
+      "loss": 0.4158,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958960387630317e-05,
+      "loss": 0.4336,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958418992198817e-05,
+      "loss": 0.4447,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957874056518954e-05,
+      "loss": 0.4434,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957325580784462e-05,
+      "loss": 0.4209,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956773565190318e-05,
+      "loss": 0.4352,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956218009932765e-05,
+      "loss": 0.4191,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955658915209312e-05,
+      "loss": 0.433,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955096281218717e-05,
+      "loss": 0.4065,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954530108160994e-05,
+      "loss": 0.4314,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953960396237418e-05,
+      "loss": 0.426,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953387145650526e-05,
+      "loss": 0.4355,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952810356604114e-05,
+      "loss": 0.4183,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952230029303227e-05,
+      "loss": 0.4116,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951646163954177e-05,
+      "loss": 0.4285,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951058760764527e-05,
+      "loss": 0.424,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950467819943104e-05,
+      "loss": 0.4275,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949873341699982e-05,
+      "loss": 0.4115,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994927532624651e-05,
+      "loss": 0.4059,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948673773795274e-05,
+      "loss": 0.4137,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948068684560138e-05,
+      "loss": 0.4094,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947460058756205e-05,
+      "loss": 0.4225,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946847896599847e-05,
+      "loss": 0.4119,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946232198308688e-05,
+      "loss": 0.4252,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945612964101612e-05,
+      "loss": 0.4254,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944990194198758e-05,
+      "loss": 0.412,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994436388882152e-05,
+      "loss": 0.4429,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994373404819256e-05,
+      "loss": 0.4225,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994310067253578e-05,
+      "loss": 0.4353,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994246376207635e-05,
+      "loss": 0.4229,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994182331704069e-05,
+      "loss": 0.4119,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994117933765648e-05,
+      "loss": 0.3922,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940531824152668e-05,
+      "loss": 0.4249,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939880776759435e-05,
+      "loss": 0.415,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939226195708235e-05,
+      "loss": 0.417,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993856808123177e-05,
+      "loss": 0.4226,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993790643356401e-05,
+      "loss": 0.4086,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993724125294016e-05,
+      "loss": 0.4133,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936572539596706e-05,
+      "loss": 0.4069,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9935900293771372e-05,
+      "loss": 0.4309,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935224515703143e-05,
+      "loss": 0.4021,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993454520563226e-05,
+      "loss": 0.4272,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933862363800222e-05,
+      "loss": 0.4258,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993317599044978e-05,
+      "loss": 0.4196,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932486085824944e-05,
+      "loss": 0.3927,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931792650170973e-05,
+      "loss": 0.4175,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993109568373439e-05,
+      "loss": 0.4128,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930395186762967e-05,
+      "loss": 0.434,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992969115950573e-05,
+      "loss": 0.4152,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928983602212963e-05,
+      "loss": 0.4258,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992827251513621e-05,
+      "loss": 0.4091,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927557898528262e-05,
+      "loss": 0.4113,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926839752643167e-05,
+      "loss": 0.4056,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926118077736227e-05,
+      "loss": 0.3995,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925392874064004e-05,
+      "loss": 0.4055,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99246641418843e-05,
+      "loss": 0.417,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923931881456196e-05,
+      "loss": 0.4088,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992319609304e-05,
+      "loss": 0.4315,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922456776897295e-05,
+      "loss": 0.4121,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99217139332909e-05,
+      "loss": 0.436,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920967562484915e-05,
+      "loss": 0.3909,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920217664744663e-05,
+      "loss": 0.4019,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991946424033674e-05,
+      "loss": 0.4149,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918707289528992e-05,
+      "loss": 0.4008,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917946812590514e-05,
+      "loss": 0.3868,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991718280979166e-05,
+      "loss": 0.3877,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916415281404032e-05,
+      "loss": 0.3999,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915644227700495e-05,
+      "loss": 0.4044,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914869648955156e-05,
+      "loss": 0.4113,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991409154544338e-05,
+      "loss": 0.4007,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991330991744179e-05,
+      "loss": 0.4111,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912524765228252e-05,
+      "loss": 0.4104,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991173608908189e-05,
+      "loss": 0.4138,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910943889283083e-05,
+      "loss": 0.386,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991014816611346e-05,
+      "loss": 0.4083,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909348919855908e-05,
+      "loss": 0.4064,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908546150794548e-05,
+      "loss": 0.4033,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907739859214785e-05,
+      "loss": 0.4053,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906930045403242e-05,
+      "loss": 0.3859,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906116709647814e-05,
+      "loss": 0.4148,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905299852237654e-05,
+      "loss": 0.3867,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904479473463145e-05,
+      "loss": 0.3823,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903655573615944e-05,
+      "loss": 0.4465,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902828152988945e-05,
+      "loss": 0.3962,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901997211876295e-05,
+      "loss": 0.4049,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901162750573406e-05,
+      "loss": 0.4048,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900324769376918e-05,
+      "loss": 0.4004,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989948326858475e-05,
+      "loss": 0.4076,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898638248496048e-05,
+      "loss": 0.3955,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897789709411226e-05,
+      "loss": 0.4021,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896937651631937e-05,
+      "loss": 0.4049,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896082075461093e-05,
+      "loss": 0.4225,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895222981202853e-05,
+      "loss": 0.3961,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894360369162628e-05,
+      "loss": 0.4216,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989349423964708e-05,
+      "loss": 0.4183,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989262459296412e-05,
+      "loss": 0.4244,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891751429422906e-05,
+      "loss": 0.392,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890874749333858e-05,
+      "loss": 0.3956,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889994553008632e-05,
+      "loss": 0.411,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889110840760143e-05,
+      "loss": 0.4011,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888223612902554e-05,
+      "loss": 0.4147,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887332869751278e-05,
+      "loss": 0.4038,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886438611622973e-05,
+      "loss": 0.3857,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885540838835556e-05,
+      "loss": 0.3983,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884639551708184e-05,
+      "loss": 0.4016,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988373475056127e-05,
+      "loss": 0.4031,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882826435716465e-05,
+      "loss": 0.4229,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881914607496694e-05,
+      "loss": 0.3805,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98809992662261e-05,
+      "loss": 0.4279,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880080412230093e-05,
+      "loss": 0.41,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879158045835334e-05,
+      "loss": 0.4003,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878232167369722e-05,
+      "loss": 0.4082,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987730277716241e-05,
+      "loss": 0.4092,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876369875543805e-05,
+      "loss": 0.3958,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875433462845545e-05,
+      "loss": 0.3866,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987449353940054e-05,
+      "loss": 0.4012,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873550105542926e-05,
+      "loss": 0.4084,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98726031616081e-05,
+      "loss": 0.4189,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871652707932706e-05,
+      "loss": 0.3856,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870698744854628e-05,
+      "loss": 0.3992,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869741272713007e-05,
+      "loss": 0.4253,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868780291848228e-05,
+      "loss": 0.4122,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867815802601916e-05,
+      "loss": 0.4042,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986684780531696e-05,
+      "loss": 0.3921,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986587630033748e-05,
+      "loss": 0.4275,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864901288008847e-05,
+      "loss": 0.4059,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863922768677685e-05,
+      "loss": 0.4082,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862940742691856e-05,
+      "loss": 0.4078,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861955210400478e-05,
+      "loss": 0.3973,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986096617215391e-05,
+      "loss": 0.4087,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859973628303754e-05,
+      "loss": 0.4049,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985897757920287e-05,
+      "loss": 0.4065,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857978025205348e-05,
+      "loss": 0.3901,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856974966666538e-05,
+      "loss": 0.406,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985596840394303e-05,
+      "loss": 0.4027,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854958337392654e-05,
+      "loss": 0.3995,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98539447673745e-05,
+      "loss": 0.3969,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852927694248893e-05,
+      "loss": 0.3929,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851907118377403e-05,
+      "loss": 0.3962,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985088304012285e-05,
+      "loss": 0.4207,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849855459849293e-05,
+      "loss": 0.4021,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984882437792205e-05,
+      "loss": 0.392,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847789794707663e-05,
+      "loss": 0.4015,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846751710573934e-05,
+      "loss": 0.405,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845710125889902e-05,
+      "loss": 0.3923,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984466504102586e-05,
+      "loss": 0.3811,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843616456353333e-05,
+      "loss": 0.3977,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98425643722451e-05,
+      "loss": 0.3704,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841508789075173e-05,
+      "loss": 0.4115,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840449707218824e-05,
+      "loss": 0.3948,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839387127052556e-05,
+      "loss": 0.4016,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838321048954117e-05,
+      "loss": 0.4011,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837251473302505e-05,
+      "loss": 0.4026,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836178400477958e-05,
+      "loss": 0.4141,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835101830861952e-05,
+      "loss": 0.3776,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834021764837213e-05,
+      "loss": 0.3938,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983293820278771e-05,
+      "loss": 0.4001,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983185114509865e-05,
+      "loss": 0.3961,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830760592156485e-05,
+      "loss": 0.3933,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829666544348916e-05,
+      "loss": 0.3919,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828569002064867e-05,
+      "loss": 0.3728,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827467965694534e-05,
+      "loss": 0.405,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826363435629325e-05,
+      "loss": 0.3812,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982525541226191e-05,
+      "loss": 0.3926,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98241438959862e-05,
+      "loss": 0.3856,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982302888719733e-05,
+      "loss": 0.402,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821910386291698e-05,
+      "loss": 0.386,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820788393666933e-05,
+      "loss": 0.3735,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819662909721905e-05,
+      "loss": 0.4058,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981853393485673e-05,
+      "loss": 0.3706,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817401469472758e-05,
+      "loss": 0.3862,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816265513972585e-05,
+      "loss": 0.3914,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815126068760048e-05,
+      "loss": 0.3859,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813983134240225e-05,
+      "loss": 0.3859,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812836710819427e-05,
+      "loss": 0.3827,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811686798905217e-05,
+      "loss": 0.3802,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810533398906392e-05,
+      "loss": 0.4092,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980937651123298e-05,
+      "loss": 0.396,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980821613629627e-05,
+      "loss": 0.389,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807052274508773e-05,
+      "loss": 0.3712,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805884926284246e-05,
+      "loss": 0.3961,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804714092037684e-05,
+      "loss": 0.3811,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803539772185326e-05,
+      "loss": 0.3923,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802361967144644e-05,
+      "loss": 0.3883,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980118067733435e-05,
+      "loss": 0.3816,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799995903174392e-05,
+      "loss": 0.3944,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979880764508597e-05,
+      "loss": 0.3846,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979761590349151e-05,
+      "loss": 0.3805,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796420678814678e-05,
+      "loss": 0.4048,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795221971480384e-05,
+      "loss": 0.3893,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794019781914764e-05,
+      "loss": 0.3833,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979281411054521e-05,
+      "loss": 0.3927,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791604957800333e-05,
+      "loss": 0.373,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790392324109996e-05,
+      "loss": 0.3827,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789176209905296e-05,
+      "loss": 0.3876,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787956615618556e-05,
+      "loss": 0.3922,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786733541683355e-05,
+      "loss": 0.3928,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785506988534492e-05,
+      "loss": 0.4017,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784276956608017e-05,
+      "loss": 0.3929,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783043446341207e-05,
+      "loss": 0.3661,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978180645817258e-05,
+      "loss": 0.3768,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780565992541887e-05,
+      "loss": 0.4073,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.977932204989012e-05,
+      "loss": 0.3848,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97780746306595e-05,
+      "loss": 0.3811,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776823735293492e-05,
+      "loss": 0.3799,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775569364236794e-05,
+      "loss": 0.3951,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774311517935336e-05,
+      "loss": 0.3835,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773050196836286e-05,
+      "loss": 0.3843,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977178540138805e-05,
+      "loss": 0.3868,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770517132040266e-05,
+      "loss": 0.3787,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769245389243806e-05,
+      "loss": 0.3849,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976797017345078e-05,
+      "loss": 0.3639,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976669148511453e-05,
+      "loss": 0.3815,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765409324689636e-05,
+      "loss": 0.3854,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764123692631908e-05,
+      "loss": 0.3694,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976283458939839e-05,
+      "loss": 0.3903,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761542015447362e-05,
+      "loss": 0.3727,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760245971238346e-05,
+      "loss": 0.3813,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758946457232082e-05,
+      "loss": 0.403,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757643473890553e-05,
+      "loss": 0.3983,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756337021676973e-05,
+      "loss": 0.3873,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975502710105579e-05,
+      "loss": 0.3834,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975371371249269e-05,
+      "loss": 0.3855,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752396856454576e-05,
+      "loss": 0.3943,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751076533409602e-05,
+      "loss": 0.3828,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749752743827147e-05,
+      "loss": 0.3885,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748425488177817e-05,
+      "loss": 0.382,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747094766933463e-05,
+      "loss": 0.3774,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745760580567153e-05,
+      "loss": 0.3807,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97444229295532e-05,
+      "loss": 0.3762,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974308181436714e-05,
+      "loss": 0.3689,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974173723548574e-05,
+      "loss": 0.4071,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740389193387012e-05,
+      "loss": 0.3928,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973903768855018e-05,
+      "loss": 0.4037,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973768272145571e-05,
+      "loss": 0.401,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736324292585304e-05,
+      "loss": 0.3948,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973496240242188e-05,
+      "loss": 0.364,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733597051449593e-05,
+      "loss": 0.401,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732228240153834e-05,
+      "loss": 0.4034,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730855969021217e-05,
+      "loss": 0.3818,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729480238539592e-05,
+      "loss": 0.3799,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972810104919803e-05,
+      "loss": 0.3826,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726718401486842e-05,
+      "loss": 0.3859,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972533229589756e-05,
+      "loss": 0.3888,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723942732922953e-05,
+      "loss": 0.3785,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722549713057008e-05,
+      "loss": 0.382,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721153236794953e-05,
+      "loss": 0.3678,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971975330463324e-05,
+      "loss": 0.38,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718349917069546e-05,
+      "loss": 0.3984,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971694307460278e-05,
+      "loss": 0.3857,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715532777733088e-05,
+      "loss": 0.3866,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714119026961818e-05,
+      "loss": 0.3781,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712701822791574e-05,
+      "loss": 0.387,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711281165726174e-05,
+      "loss": 0.3868,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709857056270667e-05,
+      "loss": 0.3759,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708429494931328e-05,
+      "loss": 0.3801,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9706998482215654e-05,
+      "loss": 0.3694,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705564018632385e-05,
+      "loss": 0.3677,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704126104691467e-05,
+      "loss": 0.3892,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702684740904086e-05,
+      "loss": 0.3857,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701239927782655e-05,
+      "loss": 0.3728,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969979166584081e-05,
+      "loss": 0.3705,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698339955593403e-05,
+      "loss": 0.3748,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969688479755653e-05,
+      "loss": 0.3737,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695426192247503e-05,
+      "loss": 0.3917,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693964140184857e-05,
+      "loss": 0.3807,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692498641888357e-05,
+      "loss": 0.3744,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691029697878998e-05,
+      "loss": 0.373,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689557308678987e-05,
+      "loss": 0.3636,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688081474811766e-05,
+      "loss": 0.3859,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686602196801998e-05,
+      "loss": 0.3907,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968511947517557e-05,
+      "loss": 0.3911,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683633310459594e-05,
+      "loss": 0.3982,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968214370318241e-05,
+      "loss": 0.3853,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968065065387357e-05,
+      "loss": 0.364,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679154163063868e-05,
+      "loss": 0.3652,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96776542312853e-05,
+      "loss": 0.3691,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676150859071105e-05,
+      "loss": 0.3672,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674644046955732e-05,
+      "loss": 0.3733,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967313379547486e-05,
+      "loss": 0.3737,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671620105165386e-05,
+      "loss": 0.3677,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670102976565432e-05,
+      "loss": 0.3798,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668582410214344e-05,
+      "loss": 0.37,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966705840665268e-05,
+      "loss": 0.3711,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966553096642224e-05,
+      "loss": 0.3893,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664000090066023e-05,
+      "loss": 0.3901,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662465778128265e-05,
+      "loss": 0.3821,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966092803115442e-05,
+      "loss": 0.3938,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965938684969116e-05,
+      "loss": 0.3815,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965784223428638e-05,
+      "loss": 0.353,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656294185489194e-05,
+      "loss": 0.3704,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965474270384994e-05,
+      "loss": 0.3755,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653187789920173e-05,
+      "loss": 0.3759,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965162944425267e-05,
+      "loss": 0.3927,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965006766740143e-05,
+      "loss": 0.3896,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648502459921667e-05,
+      "loss": 0.3879,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646933822369815e-05,
+      "loss": 0.3893,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645361755303535e-05,
+      "loss": 0.3844,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643786259281696e-05,
+      "loss": 0.3688,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642207334864395e-05,
+      "loss": 0.3885,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640624982612943e-05,
+      "loss": 0.389,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639039203089877e-05,
+      "loss": 0.3759,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637449996858934e-05,
+      "loss": 0.3884,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635857364485093e-05,
+      "loss": 0.371,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634261306534538e-05,
+      "loss": 0.3854,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632661823574664e-05,
+      "loss": 0.3799,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96310589161741e-05,
+      "loss": 0.3623,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629452584902685e-05,
+      "loss": 0.3866,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962784283033147e-05,
+      "loss": 0.3809,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962622965303273e-05,
+      "loss": 0.3943,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624613053579954e-05,
+      "loss": 0.3712,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962299303254785e-05,
+      "loss": 0.3834,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962136959051234e-05,
+      "loss": 0.3704,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619742728050557e-05,
+      "loss": 0.3624,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618112445740864e-05,
+      "loss": 0.373,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616478744162823e-05,
+      "loss": 0.3975,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614841623897228e-05,
+      "loss": 0.3837,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613201085526075e-05,
+      "loss": 0.3633,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611557129632585e-05,
+      "loss": 0.3794,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609909756801185e-05,
+      "loss": 0.3651,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608258967617518e-05,
+      "loss": 0.3573,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606604762668453e-05,
+      "loss": 0.3922,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960494714254206e-05,
+      "loss": 0.3607,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603286107827627e-05,
+      "loss": 0.3921,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601621659115662e-05,
+      "loss": 0.4001,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599953796997878e-05,
+      "loss": 0.3748,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95982825220672e-05,
+      "loss": 0.4075,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959660783491778e-05,
+      "loss": 0.3634,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.3973,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959324822634535e-05,
+      "loss": 0.376,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959156330611668e-05,
+      "loss": 0.3499,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589874976057975e-05,
+      "loss": 0.3876,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588183236769426e-05,
+      "loss": 0.3715,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586488088852465e-05,
+      "loss": 0.3843,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958478953290971e-05,
+      "loss": 0.3887,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958308756954501e-05,
+      "loss": 0.391,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958138219936342e-05,
+      "loss": 0.3711,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579673422971202e-05,
+      "loss": 0.3933,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577961240975826e-05,
+      "loss": 0.3995,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576245653985984e-05,
+      "loss": 0.3875,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574526662611572e-05,
+      "loss": 0.3948,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572804267463688e-05,
+      "loss": 0.3777,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571078469154666e-05,
+      "loss": 0.3816,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569349268298018e-05,
+      "loss": 0.3812,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567616665508483e-05,
+      "loss": 0.3726,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956588066140201e-05,
+      "loss": 0.3926,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564141256595753e-05,
+      "loss": 0.3665,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562398451708077e-05,
+      "loss": 0.3789,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956065224735855e-05,
+      "loss": 0.3777,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558902644167956e-05,
+      "loss": 0.3765,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955714964275828e-05,
+      "loss": 0.3561,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555393243752728e-05,
+      "loss": 0.384,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553633447775696e-05,
+      "loss": 0.3608,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551870255452804e-05,
+      "loss": 0.3708,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955010366741087e-05,
+      "loss": 0.3615,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954833368427792e-05,
+      "loss": 0.3792,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546560306683187e-05,
+      "loss": 0.3601,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544783535257114e-05,
+      "loss": 0.3486,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954300337063135e-05,
+      "loss": 0.3883,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954121981343875e-05,
+      "loss": 0.3778,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539432864313367e-05,
+      "loss": 0.3401,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537642523890477e-05,
+      "loss": 0.3619,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535848792806543e-05,
+      "loss": 0.3823,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534051671699247e-05,
+      "loss": 0.3542,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953225116120747e-05,
+      "loss": 0.3695,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9530447261971296e-05,
+      "loss": 0.3732,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952863997463202e-05,
+      "loss": 0.3705,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526829299832137e-05,
+      "loss": 0.3732,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525015238215346e-05,
+      "loss": 0.3859,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523197790426557e-05,
+      "loss": 0.3605,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952137695711187e-05,
+      "loss": 0.3601,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519552738918598e-05,
+      "loss": 0.3768,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517725136495266e-05,
+      "loss": 0.3512,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515894150491582e-05,
+      "loss": 0.3748,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951405978155847e-05,
+      "loss": 0.3762,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512222030348055e-05,
+      "loss": 0.3707,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510380897513662e-05,
+      "loss": 0.386,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950853638370982e-05,
+      "loss": 0.389,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950668848959226e-05,
+      "loss": 0.3629,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504837215817915e-05,
+      "loss": 0.3785,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950298256304491e-05,
+      "loss": 0.3706,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501124531932592e-05,
+      "loss": 0.3748,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949926312314149e-05,
+      "loss": 0.3566,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497398337333336e-05,
+      "loss": 0.3608,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495530175171078e-05,
+      "loss": 0.3674,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493658637318846e-05,
+      "loss": 0.3754,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491783724441983e-05,
+      "loss": 0.3771,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489905437207018e-05,
+      "loss": 0.3492,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488023776281695e-05,
+      "loss": 0.3732,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486138742334945e-05,
+      "loss": 0.3699,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484250336036905e-05,
+      "loss": 0.3622,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482358558058913e-05,
+      "loss": 0.3544,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480463409073498e-05,
+      "loss": 0.3749,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947856488975439e-05,
+      "loss": 0.3715,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476663000776524e-05,
+      "loss": 0.3894,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474757742816024e-05,
+      "loss": 0.3821,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472849116550217e-05,
+      "loss": 0.3605,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947093712265762e-05,
+      "loss": 0.3768,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469021761817953e-05,
+      "loss": 0.3698,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946710303471214e-05,
+      "loss": 0.3813,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946518094202229e-05,
+      "loss": 0.3684,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463255484431713e-05,
+      "loss": 0.3821,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461326662624914e-05,
+      "loss": 0.3862,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459394477287598e-05,
+      "loss": 0.3766,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457458929106655e-05,
+      "loss": 0.3644,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455520018770187e-05,
+      "loss": 0.3727,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453577746967483e-05,
+      "loss": 0.3794,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945163211438902e-05,
+      "loss": 0.3875,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449683121726483e-05,
+      "loss": 0.3645,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447730769672736e-05,
+      "loss": 0.3639,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445775058921856e-05,
+      "loss": 0.3739,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443815990169094e-05,
+      "loss": 0.3777,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441853564110915e-05,
+      "loss": 0.3647,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943988778144496e-05,
+      "loss": 0.3773,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943791864287008e-05,
+      "loss": 0.3635,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435946149086297e-05,
+      "loss": 0.3529,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943397030079485e-05,
+      "loss": 0.3615,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943199109869815e-05,
+      "loss": 0.373,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430008543499816e-05,
+      "loss": 0.3577,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942802263590465e-05,
+      "loss": 0.376,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426033376618645e-05,
+      "loss": 0.3754,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424040766348992e-05,
+      "loss": 0.356,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422044805804074e-05,
+      "loss": 0.3776,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420045495693456e-05,
+      "loss": 0.3757,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418042836727898e-05,
+      "loss": 0.3468,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416036829619353e-05,
+      "loss": 0.3691,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414027475080964e-05,
+      "loss": 0.3674,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412014773827062e-05,
+      "loss": 0.3678,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940999872657317e-05,
+      "loss": 0.3785,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407979334035995e-05,
+      "loss": 0.3738,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940595659693344e-05,
+      "loss": 0.3923,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940393051598459e-05,
+      "loss": 0.373,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401901091909733e-05,
+      "loss": 0.3772,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399868325430322e-05,
+      "loss": 0.3533,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397832217269023e-05,
+      "loss": 0.3618,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939579276814967e-05,
+      "loss": 0.3746,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393749978797302e-05,
+      "loss": 0.3741,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939170384993813e-05,
+      "loss": 0.3441,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938965438229956e-05,
+      "loss": 0.3549,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387601576610188e-05,
+      "loss": 0.3757,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938554543359979e-05,
+      "loss": 0.3654,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938348595399933e-05,
+      "loss": 0.3646,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381423138540963e-05,
+      "loss": 0.3723,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379356987958022e-05,
+      "loss": 0.3577,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377287502985033e-05,
+      "loss": 0.3859,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93752146843577e-05,
+      "loss": 0.3735,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937313853281292e-05,
+      "loss": 0.3647,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937105904908877e-05,
+      "loss": 0.3652,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368976233924518e-05,
+      "loss": 0.354,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366890088060603e-05,
+      "loss": 0.3696,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936480061223866e-05,
+      "loss": 0.3622,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362707807201503e-05,
+      "loss": 0.3486,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360611673693132e-05,
+      "loss": 0.3862,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935851221245873e-05,
+      "loss": 0.3766,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356409424244655e-05,
+      "loss": 0.3623,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354303309798462e-05,
+      "loss": 0.3662,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935219386986888e-05,
+      "loss": 0.3624,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935008110520582e-05,
+      "loss": 0.3693,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347965016560374e-05,
+      "loss": 0.364,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934584560468482e-05,
+      "loss": 0.3673,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934372287033262e-05,
+      "loss": 0.3615,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341596814258407e-05,
+      "loss": 0.3606,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339467437217997e-05,
+      "loss": 0.3636,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337334739968398e-05,
+      "loss": 0.3566,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335198723267788e-05,
+      "loss": 0.3596,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333059387875527e-05,
+      "loss": 0.35,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330916734552154e-05,
+      "loss": 0.3518,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932877076405939e-05,
+      "loss": 0.36,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932662147716013e-05,
+      "loss": 0.3746,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932446887461846e-05,
+      "loss": 0.3536,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932231295719963e-05,
+      "loss": 0.3732,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932015372567008e-05,
+      "loss": 0.365,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931799118079742e-05,
+      "loss": 0.3615,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315825323350444e-05,
+      "loss": 0.3661,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313656154099116e-05,
+      "loss": 0.3735,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931148367381459e-05,
+      "loss": 0.3591,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930930788326918e-05,
+      "loss": 0.347,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307128783236394e-05,
+      "loss": 0.3633,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930494637449091e-05,
+      "loss": 0.3676,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930276065780857e-05,
+      "loss": 0.3551,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300571633966414e-05,
+      "loss": 0.3604,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929837930374264e-05,
+      "loss": 0.3596,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296183667916633e-05,
+      "loss": 0.3716,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293984727268948e-05,
+      "loss": 0.3539,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929178248258131e-05,
+      "loss": 0.3652,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289576934636622e-05,
+      "loss": 0.3623,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287368084218975e-05,
+      "loss": 0.3772,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285155932113612e-05,
+      "loss": 0.3539,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282940479106962e-05,
+      "loss": 0.3656,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280721725986627e-05,
+      "loss": 0.3577,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927849967354138e-05,
+      "loss": 0.3691,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276274322561167e-05,
+      "loss": 0.369,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274045673837104e-05,
+      "loss": 0.3472,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271813728161488e-05,
+      "loss": 0.3524,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269578486327777e-05,
+      "loss": 0.3741,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267339949130613e-05,
+      "loss": 0.3698,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265098117365797e-05,
+      "loss": 0.3683,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926285299183031e-05,
+      "loss": 0.3474,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260604573322298e-05,
+      "loss": 0.3739,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925835286264108e-05,
+      "loss": 0.3528,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256097860587157e-05,
+      "loss": 0.3488,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253839567962176e-05,
+      "loss": 0.3727,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925157798556897e-05,
+      "loss": 0.3552,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249313114211545e-05,
+      "loss": 0.3532,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924704495469506e-05,
+      "loss": 0.3699,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244773507825863e-05,
+      "loss": 0.3613,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924249877441145e-05,
+      "loss": 0.3583,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240220755260502e-05,
+      "loss": 0.3678,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923793945118286e-05,
+      "loss": 0.3594,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235654862989537e-05,
+      "loss": 0.3383,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233366991492707e-05,
+      "loss": 0.3566,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231075837505715e-05,
+      "loss": 0.3672,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922878140184308e-05,
+      "loss": 0.3422,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226483685320473e-05,
+      "loss": 0.3665,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224182688754743e-05,
+      "loss": 0.3627,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221878412963903e-05,
+      "loss": 0.3646,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921957085876712e-05,
+      "loss": 0.3702,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217260026984756e-05,
+      "loss": 0.3652,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92149459184383e-05,
+      "loss": 0.3566,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212628533950433e-05,
+      "loss": 0.3638,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921030787434499e-05,
+      "loss": 0.3687,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207983940446976e-05,
+      "loss": 0.3485,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205656733082555e-05,
+      "loss": 0.3798,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203326253079053e-05,
+      "loss": 0.3745,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200992501264967e-05,
+      "loss": 0.3595,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198655478469952e-05,
+      "loss": 0.3593,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196315185524822e-05,
+      "loss": 0.3381,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193971623261568e-05,
+      "loss": 0.3542,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191624792513326e-05,
+      "loss": 0.3683,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91892746941144e-05,
+      "loss": 0.3446,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186921328900263e-05,
+      "loss": 0.3707,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184564697707545e-05,
+      "loss": 0.3784,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182204801374028e-05,
+      "loss": 0.356,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179841640738664e-05,
+      "loss": 0.3566,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177475216641567e-05,
+      "loss": 0.3552,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175105529924008e-05,
+      "loss": 0.3556,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172732581428414e-05,
+      "loss": 0.3733,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170356371998383e-05,
+      "loss": 0.3633,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167976902478654e-05,
+      "loss": 0.3516,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165594173715146e-05,
+      "loss": 0.3738,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916320818655492e-05,
+      "loss": 0.3647,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160818941846204e-05,
+      "loss": 0.371,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915842644043838e-05,
+      "loss": 0.3643,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156030683181992e-05,
+      "loss": 0.361,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915363167092874e-05,
+      "loss": 0.3643,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151229404531473e-05,
+      "loss": 0.3649,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914882388484421e-05,
+      "loss": 0.3657,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146415112722126e-05,
+      "loss": 0.354,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144003089021533e-05,
+      "loss": 0.375,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141587814599928e-05,
+      "loss": 0.373,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139169290315935e-05,
+      "loss": 0.3568,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136747517029356e-05,
+      "loss": 0.3611,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134322495601133e-05,
+      "loss": 0.3417,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913189422689337e-05,
+      "loss": 0.3667,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129462711769327e-05,
+      "loss": 0.3549,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912702795109341e-05,
+      "loss": 0.3364,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912458994573119e-05,
+      "loss": 0.3574,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912214869654938e-05,
+      "loss": 0.3625,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119704204415854e-05,
+      "loss": 0.3535,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117256470199638e-05,
+      "loss": 0.3557,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114805494770907e-05,
+      "loss": 0.3582,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112351279000987e-05,
+      "loss": 0.3578,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109893823762366e-05,
+      "loss": 0.3502,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910743312992867e-05,
+      "loss": 0.3582,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910496919837469e-05,
+      "loss": 0.3496,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102502029976355e-05,
+      "loss": 0.384,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100031625610754e-05,
+      "loss": 0.3634,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097557986156124e-05,
+      "loss": 0.3669,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095081112491848e-05,
+      "loss": 0.3521,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092601005498467e-05,
+      "loss": 0.3705,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090117666057665e-05,
+      "loss": 0.359,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087631095052274e-05,
+      "loss": 0.3536,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085141293366277e-05,
+      "loss": 0.3604,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908264826188481e-05,
+      "loss": 0.3658,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080152001494152e-05,
+      "loss": 0.3752,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907765251308173e-05,
+      "loss": 0.3549,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075149797536127e-05,
+      "loss": 0.3409,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907264385574705e-05,
+      "loss": 0.3596,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070134688605387e-05,
+      "loss": 0.3635,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067622297003142e-05,
+      "loss": 0.3712,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065106681833484e-05,
+      "loss": 0.3629,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062587843990717e-05,
+      "loss": 0.3425,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060065784370302e-05,
+      "loss": 0.3483,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057540503868832e-05,
+      "loss": 0.3362,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905501200338406e-05,
+      "loss": 0.3657,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052480283814865e-05,
+      "loss": 0.3665,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049945346061292e-05,
+      "loss": 0.3447,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047407191024513e-05,
+      "loss": 0.3544,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904486581960685e-05,
+      "loss": 0.3699,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042321232711768e-05,
+      "loss": 0.3629,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903977343124388e-05,
+      "loss": 0.3582,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037222416108935e-05,
+      "loss": 0.3778,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034668188213823e-05,
+      "loss": 0.3456,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903211074846658e-05,
+      "loss": 0.3483,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029550097776393e-05,
+      "loss": 0.3689,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902698623705357e-05,
+      "loss": 0.3512,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024419167209575e-05,
+      "loss": 0.3522,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902184888915701e-05,
+      "loss": 0.3512,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019275403809616e-05,
+      "loss": 0.3524,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016698712082277e-05,
+      "loss": 0.3623,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014118814891012e-05,
+      "loss": 0.3602,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011535713152985e-05,
+      "loss": 0.3488,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008949407786493e-05,
+      "loss": 0.374,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900635989971098e-05,
+      "loss": 0.3558,
+      "step": 917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003767189847018e-05,
+      "loss": 0.3431,
+      "step": 918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001171279116327e-05,
+      "loss": 0.3597,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998572168441766e-05,
+      "loss": 0.3552,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995969858747316e-05,
+      "loss": 0.3457,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993364350958115e-05,
+      "loss": 0.3491,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990755646000422e-05,
+      "loss": 0.3496,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988143744801647e-05,
+      "loss": 0.3833,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985528648290327e-05,
+      "loss": 0.3599,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982910357396128e-05,
+      "loss": 0.3721,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980288873049875e-05,
+      "loss": 0.3684,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89776641961835e-05,
+      "loss": 0.3632,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897503632773009e-05,
+      "loss": 0.3632,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972405268623863e-05,
+      "loss": 0.3662,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969771019800168e-05,
+      "loss": 0.3438,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896713358219548e-05,
+      "loss": 0.3529,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964492956747427e-05,
+      "loss": 0.3619,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896184914439475e-05,
+      "loss": 0.3511,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895920214607734e-05,
+      "loss": 0.3591,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956551962736208e-05,
+      "loss": 0.3613,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953898595313505e-05,
+      "loss": 0.3444,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951242044752512e-05,
+      "loss": 0.3397,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948582311997638e-05,
+      "loss": 0.363,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945919397994425e-05,
+      "loss": 0.3547,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943253303689552e-05,
+      "loss": 0.3442,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894058403003082e-05,
+      "loss": 0.3627,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937911577967165e-05,
+      "loss": 0.3599,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935235948448653e-05,
+      "loss": 0.3745,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932557142426477e-05,
+      "loss": 0.3638,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892987516085296e-05,
+      "loss": 0.3475,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927190004681558e-05,
+      "loss": 0.3293,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924501674866847e-05,
+      "loss": 0.353,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892181017236454e-05,
+      "loss": 0.3558,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919115498131475e-05,
+      "loss": 0.3461,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916417653125615e-05,
+      "loss": 0.3451,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913716638306054e-05,
+      "loss": 0.3512,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911012454633007e-05,
+      "loss": 0.3577,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908305103067823e-05,
+      "loss": 0.3528,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905594584572973e-05,
+      "loss": 0.3473,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902880900112057e-05,
+      "loss": 0.338,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900164050649793e-05,
+      "loss": 0.3489,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897444037152034e-05,
+      "loss": 0.3622,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889472086058575e-05,
+      "loss": 0.3558,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889199452191904e-05,
+      "loss": 0.3538,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889265022121126e-05,
+      "loss": 0.3534,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886532362162354e-05,
+      "loss": 0.3594,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888379654301419e-05,
+      "loss": 0.3447,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888105756564923e-05,
+      "loss": 0.3419,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878315431041185e-05,
+      "loss": 0.3613,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875570140164895e-05,
+      "loss": 0.3567,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872821693996317e-05,
+      "loss": 0.3607,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870070093512534e-05,
+      "loss": 0.3534,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886731533969175e-05,
+      "loss": 0.3383,
+      "step": 969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864557433513283e-05,
+      "loss": 0.3738,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861796375957585e-05,
+      "loss": 0.3547,
+      "step": 971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885903216800621e-05,
+      "loss": 0.3412,
+      "step": 972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885626481064185e-05,
+      "loss": 0.3469,
+      "step": 973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853494304848305e-05,
+      "loss": 0.351,
+      "step": 974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850720651610503e-05,
+      "loss": 0.3698,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847943851914483e-05,
+      "loss": 0.3538,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845163906747407e-05,
+      "loss": 0.3691,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884238081709755e-05,
+      "loss": 0.3593,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839594583954314e-05,
+      "loss": 0.3542,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836805208308213e-05,
+      "loss": 0.3345,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834012691150874e-05,
+      "loss": 0.3396,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883121703347505e-05,
+      "loss": 0.3574,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828418236274603e-05,
+      "loss": 0.3378,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825616300544517e-05,
+      "loss": 0.3481,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822811227280887e-05,
+      "loss": 0.3623,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820003017480925e-05,
+      "loss": 0.3461,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817191672142957e-05,
+      "loss": 0.3566,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814377192266424e-05,
+      "loss": 0.3519,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811559578851887e-05,
+      "loss": 0.3481,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808738832901015e-05,
+      "loss": 0.3425,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805914955416583e-05,
+      "loss": 0.343,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803087947402503e-05,
+      "loss": 0.3418,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800257809863772e-05,
+      "loss": 0.3612,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797424543806522e-05,
+      "loss": 0.3488,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879458815023798e-05,
+      "loss": 0.3586,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87917486301665e-05,
+      "loss": 0.3484,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788905984601533e-05,
+      "loss": 0.3763,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786060214553652e-05,
+      "loss": 0.3623,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783211321034537e-05,
+      "loss": 0.3472,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780359305056976e-05,
+      "loss": 0.3365,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777504167634873e-05,
+      "loss": 0.3477,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774645909783233e-05,
+      "loss": 0.3391,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877178453251818e-05,
+      "loss": 0.3615,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768920036856945e-05,
+      "loss": 0.3516,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766052423817853e-05,
+      "loss": 0.3462,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876318169442036e-05,
+      "loss": 0.3441,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760307849685017e-05,
+      "loss": 0.3441,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757430890633483e-05,
+      "loss": 0.3529,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875455081828853e-05,
+      "loss": 0.3574,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751667633674023e-05,
+      "loss": 0.35,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874878133781495e-05,
+      "loss": 0.3467,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87458919317374e-05,
+      "loss": 0.3649,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874299941646856e-05,
+      "loss": 0.354,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740103793036727e-05,
+      "loss": 0.3555,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737205062471317e-05,
+      "loss": 0.3511,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873430322580282e-05,
+      "loss": 0.3241,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731398284062856e-05,
+      "loss": 0.345,
+      "step": 1017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728490238284142e-05,
+      "loss": 0.3483,
+      "step": 1018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725579089500498e-05,
+      "loss": 0.3456,
+      "step": 1019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722664838746842e-05,
+      "loss": 0.353,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87197474870592e-05,
+      "loss": 0.3459,
+      "step": 1021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87168270354747e-05,
+      "loss": 0.3582,
+      "step": 1022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713903485031573e-05,
+      "loss": 0.3589,
+      "step": 1023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871097683676915e-05,
+      "loss": 0.3557,
+      "step": 1024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870804709172786e-05,
+      "loss": 0.354,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705114250949238e-05,
+      "loss": 0.3518,
+      "step": 1026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702178315475922e-05,
+      "loss": 0.3699,
+      "step": 1027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869923928635164e-05,
+      "loss": 0.3372,
+      "step": 1028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869629716462123e-05,
+      "loss": 0.338,
+      "step": 1029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869335195133062e-05,
+      "loss": 0.344,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869040364752685e-05,
+      "loss": 0.3478,
+      "step": 1031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687452254258046e-05,
+      "loss": 0.3472,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684497772573434e-05,
+      "loss": 0.3448,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681540203523352e-05,
+      "loss": 0.3403,
+      "step": 1034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678579548159215e-05,
+      "loss": 0.3573,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675615807533547e-05,
+      "loss": 0.3447,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867264898269997e-05,
+      "loss": 0.343,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86696790747132e-05,
+      "loss": 0.3337,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866670608462904e-05,
+      "loss": 0.3628,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663730013504406e-05,
+      "loss": 0.3641,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660750862397297e-05,
+      "loss": 0.3582,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865776863236681e-05,
+      "loss": 0.3354,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654783324473138e-05,
+      "loss": 0.3544,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651794939777566e-05,
+      "loss": 0.3561,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648803479342475e-05,
+      "loss": 0.3577,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645808944231342e-05,
+      "loss": 0.3465,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642811335508724e-05,
+      "loss": 0.3489,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639810654240284e-05,
+      "loss": 0.3519,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636806901492775e-05,
+      "loss": 0.3411,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633800078334042e-05,
+      "loss": 0.3409,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630790185833015e-05,
+      "loss": 0.3645,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627777225059726e-05,
+      "loss": 0.3472,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624761197085287e-05,
+      "loss": 0.3518,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8621742102981906e-05,
+      "loss": 0.3446,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861871994382288e-05,
+      "loss": 0.3428,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615694720682597e-05,
+      "loss": 0.3442,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612666434636535e-05,
+      "loss": 0.3477,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609635086761254e-05,
+      "loss": 0.3589,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860660067813441e-05,
+      "loss": 0.3365,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603563209834745e-05,
+      "loss": 0.3539,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600522682942085e-05,
+      "loss": 0.3572,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859747909853735e-05,
+      "loss": 0.3358,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594432457702547e-05,
+      "loss": 0.3562,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591382761520757e-05,
+      "loss": 0.3409,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588330011076167e-05,
+      "loss": 0.3462,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585274207454027e-05,
+      "loss": 0.3522,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582215351740695e-05,
+      "loss": 0.3545,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85791534450236e-05,
+      "loss": 0.3473,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576088488391253e-05,
+      "loss": 0.3412,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573020482933265e-05,
+      "loss": 0.3636,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569949429740316e-05,
+      "loss": 0.3271,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566875329904176e-05,
+      "loss": 0.3341,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85637981845177e-05,
+      "loss": 0.3313,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560717994674818e-05,
+      "loss": 0.3662,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8557634761470547e-05,
+      "loss": 0.3451,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855454848600099e-05,
+      "loss": 0.3345,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855145916936332e-05,
+      "loss": 0.3341,
+      "step": 1077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854836681265581e-05,
+      "loss": 0.3502,
+      "step": 1078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854527141697779e-05,
+      "loss": 0.3519,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854217298342969e-05,
+      "loss": 0.338,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853907151311301e-05,
+      "loss": 0.3472,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535967007130337e-05,
+      "loss": 0.3629,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532859466585322e-05,
+      "loss": 0.3527,
+      "step": 1083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529748892582714e-05,
+      "loss": 0.345,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852663528622833e-05,
+      "loss": 0.338,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852351864862906e-05,
+      "loss": 0.3383,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852039898089289e-05,
+      "loss": 0.3328,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851727628412886e-05,
+      "loss": 0.3571,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514150559447108e-05,
+      "loss": 0.3409,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851102180795883e-05,
+      "loss": 0.358,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850789003077631e-05,
+      "loss": 0.3425,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850475522901291e-05,
+      "loss": 0.3506,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850161740378305e-05,
+      "loss": 0.3376,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849847655620225e-05,
+      "loss": 0.3599,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849533268738708e-05,
+      "loss": 0.3545,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84921857984552e-05,
+      "loss": 0.3439,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489035890525336e-05,
+      "loss": 0.3374,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485882964717292e-05,
+      "loss": 0.3438,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482727022151943e-05,
+      "loss": 0.3444,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8479568063951236e-05,
+      "loss": 0.3365,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847640609123819e-05,
+      "loss": 0.3441,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847324110513689e-05,
+      "loss": 0.3444,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847007310677251e-05,
+      "loss": 0.3393,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466902097271275e-05,
+      "loss": 0.3441,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463728077760488e-05,
+      "loss": 0.3378,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846055104936853e-05,
+      "loss": 0.3503,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457371013224835e-05,
+      "loss": 0.3604,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454187970459923e-05,
+      "loss": 0.3402,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845100192220537e-05,
+      "loss": 0.3279,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447812869593832e-05,
+      "loss": 0.3549,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444620813759017e-05,
+      "loss": 0.3438,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441425755835723e-05,
+      "loss": 0.3501,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438227696959793e-05,
+      "loss": 0.3652,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435026638268152e-05,
+      "loss": 0.3322,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431822580898785e-05,
+      "loss": 0.3505,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428615525990743e-05,
+      "loss": 0.3372,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425405474684143e-05,
+      "loss": 0.3364,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842219242812017e-05,
+      "loss": 0.3417,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418976387441075e-05,
+      "loss": 0.359,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841575735379016e-05,
+      "loss": 0.3591,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.3525,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840931031215147e-05,
+      "loss": 0.3634,
+      "step": 1122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840608230645563e-05,
+      "loss": 0.3418,
+      "step": 1123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402851312371862e-05,
+      "loss": 0.3362,
+      "step": 1124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399617331048793e-05,
+      "loss": 0.3605,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396380363636117e-05,
+      "loss": 0.3516,
+      "step": 1126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839314041128458e-05,
+      "loss": 0.3577,
+      "step": 1127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389897475145998e-05,
+      "loss": 0.3434,
+      "step": 1128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838665155637325e-05,
+      "loss": 0.3453,
+      "step": 1129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838340265612026e-05,
+      "loss": 0.3435,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838015077554203e-05,
+      "loss": 0.3324,
+      "step": 1131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376895915794607e-05,
+      "loss": 0.3451,
+      "step": 1132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373638078035106e-05,
+      "loss": 0.3406,
+      "step": 1133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370377263421707e-05,
+      "loss": 0.3571,
+      "step": 1134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367113473113622e-05,
+      "loss": 0.3428,
+      "step": 1135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363846708271154e-05,
+      "loss": 0.3489,
+      "step": 1136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836057697005564e-05,
+      "loss": 0.3784,
+      "step": 1137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357304259629485e-05,
+      "loss": 0.3577,
+      "step": 1138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354028578156146e-05,
+      "loss": 0.3423,
+      "step": 1139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350749926800135e-05,
+      "loss": 0.368,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834746830672703e-05,
+      "loss": 0.3344,
+      "step": 1141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344183719103454e-05,
+      "loss": 0.351,
+      "step": 1142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834089616509708e-05,
+      "loss": 0.3529,
+      "step": 1143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337605645876655e-05,
+      "loss": 0.3629,
+      "step": 1144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334312162611963e-05,
+      "loss": 0.3347,
+      "step": 1145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833101571647385e-05,
+      "loss": 0.3477,
+      "step": 1146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327716308634206e-05,
+      "loss": 0.3431,
+      "step": 1147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324413940265988e-05,
+      "loss": 0.3481,
+      "step": 1148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321108612543194e-05,
+      "loss": 0.3365,
+      "step": 1149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831780032664088e-05,
+      "loss": 0.3776,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314489083735155e-05,
+      "loss": 0.3386,
+      "step": 1151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311174885003168e-05,
+      "loss": 0.3505,
+      "step": 1152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307857731623133e-05,
+      "loss": 0.3445,
+      "step": 1153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304537624774306e-05,
+      "loss": 0.3392,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301214565636993e-05,
+      "loss": 0.3456,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297888555392555e-05,
+      "loss": 0.3363,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294559595223402e-05,
+      "loss": 0.3448,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291227686312978e-05,
+      "loss": 0.3542,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287892829845798e-05,
+      "loss": 0.3344,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828455502700741e-05,
+      "loss": 0.3483,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828121427898441e-05,
+      "loss": 0.3403,
+      "step": 1161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277870586964452e-05,
+      "loss": 0.3464,
+      "step": 1162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827452395213622e-05,
+      "loss": 0.3314,
+      "step": 1163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271174375689454e-05,
+      "loss": 0.354,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267821858814944e-05,
+      "loss": 0.3396,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264466402704516e-05,
+      "loss": 0.3373,
+      "step": 1166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826110800855105e-05,
+      "loss": 0.3237,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257746677548457e-05,
+      "loss": 0.3279,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254382410891705e-05,
+      "loss": 0.3531,
+      "step": 1169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251015209776803e-05,
+      "loss": 0.3339,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82476450754008e-05,
+      "loss": 0.3447,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244272008961786e-05,
+      "loss": 0.3563,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82408960116589e-05,
+      "loss": 0.3367,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237517084692316e-05,
+      "loss": 0.3431,
+      "step": 1174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234135229263258e-05,
+      "loss": 0.3605,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230750446573983e-05,
+      "loss": 0.3582,
+      "step": 1176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822736273782779e-05,
+      "loss": 0.3552,
+      "step": 1177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822397210422902e-05,
+      "loss": 0.3444,
+      "step": 1178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220578546983055e-05,
+      "loss": 0.3435,
+      "step": 1179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217182067296317e-05,
+      "loss": 0.3394,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213782666376262e-05,
+      "loss": 0.3396,
+      "step": 1181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210380345431386e-05,
+      "loss": 0.3417,
+      "step": 1182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206975105671225e-05,
+      "loss": 0.3624,
+      "step": 1183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203566948306358e-05,
+      "loss": 0.3319,
+      "step": 1184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200155874548388e-05,
+      "loss": 0.3552,
+      "step": 1185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819674188560997e-05,
+      "loss": 0.3373,
+      "step": 1186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193324982704776e-05,
+      "loss": 0.3242,
+      "step": 1187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189905167047535e-05,
+      "loss": 0.351,
+      "step": 1188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186482439854e-05,
+      "loss": 0.3416,
+      "step": 1189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818305680234096e-05,
+      "loss": 0.3446,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179628255726238e-05,
+      "loss": 0.3417,
+      "step": 1191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176196801228695e-05,
+      "loss": 0.3644,
+      "step": 1192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172762440068225e-05,
+      "loss": 0.3442,
+      "step": 1193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816932517346575e-05,
+      "loss": 0.3519,
+      "step": 1194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165885002643228e-05,
+      "loss": 0.3257,
+      "step": 1195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162441928823658e-05,
+      "loss": 0.3116,
+      "step": 1196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158995953231055e-05,
+      "loss": 0.3561,
+      "step": 1197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155547077090475e-05,
+      "loss": 0.3291,
+      "step": 1198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815209530162801e-05,
+      "loss": 0.3408,
+      "step": 1199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814864062807077e-05,
+      "loss": 0.3418,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145183057646908e-05,
+      "loss": 0.3369,
+      "step": 1201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141722591585594e-05,
+      "loss": 0.3531,
+      "step": 1202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138259231117036e-05,
+      "loss": 0.3413,
+      "step": 1203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134792977472476e-05,
+      "loss": 0.3531,
+      "step": 1204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813132383188417e-05,
+      "loss": 0.3594,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812785179558541e-05,
+      "loss": 0.3519,
+      "step": 1206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124376869810524e-05,
+      "loss": 0.342,
+      "step": 1207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120899055794848e-05,
+      "loss": 0.3395,
+      "step": 1208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117418354774762e-05,
+      "loss": 0.3373,
+      "step": 1209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8113934767987662e-05,
+      "loss": 0.345,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110448296671975e-05,
+      "loss": 0.339,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810695894206716e-05,
+      "loss": 0.3412,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103466705413678e-05,
+      "loss": 0.3253,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099971587953043e-05,
+      "loss": 0.3431,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096473590927773e-05,
+      "loss": 0.3359,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092972715581424e-05,
+      "loss": 0.3605,
+      "step": 1216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089468963158555e-05,
+      "loss": 0.3312,
+      "step": 1217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808596233490477e-05,
+      "loss": 0.3496,
+      "step": 1218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808245283206669e-05,
+      "loss": 0.3472,
+      "step": 1219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807894045589194e-05,
+      "loss": 0.3391,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8075425207629193e-05,
+      "loss": 0.3368,
+      "step": 1221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8071907088528124e-05,
+      "loss": 0.3279,
+      "step": 1222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068386099839438e-05,
+      "loss": 0.3514,
+      "step": 1223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806486224281486e-05,
+      "loss": 0.3364,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061335518707122e-05,
+      "loss": 0.3445,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057805928769996e-05,
+      "loss": 0.344,
+      "step": 1226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054273474258257e-05,
+      "loss": 0.3385,
+      "step": 1227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050738156427704e-05,
+      "loss": 0.3431,
+      "step": 1228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047199976535153e-05,
+      "loss": 0.3367,
+      "step": 1229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043658935838436e-05,
+      "loss": 0.3475,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804011503559641e-05,
+      "loss": 0.3342,
+      "step": 1231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036568277068934e-05,
+      "loss": 0.3347,
+      "step": 1232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033018661516897e-05,
+      "loss": 0.3275,
+      "step": 1233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029466190202195e-05,
+      "loss": 0.3494,
+      "step": 1234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025910864387745e-05,
+      "loss": 0.3397,
+      "step": 1235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802235268533748e-05,
+      "loss": 0.3669,
+      "step": 1236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018791654316333e-05,
+      "loss": 0.3496,
+      "step": 1237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015227772590268e-05,
+      "loss": 0.3479,
+      "step": 1238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801166104142626e-05,
+      "loss": 0.351,
+      "step": 1239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008091462092285e-05,
+      "loss": 0.3124,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800451903585734e-05,
+      "loss": 0.3263,
+      "step": 1241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800094376399144e-05,
+      "loss": 0.3369,
+      "step": 1242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997365647765596e-05,
+      "loss": 0.3562,
+      "step": 1243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993784688451845e-05,
+      "loss": 0.3435,
+      "step": 1244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799020088732323e-05,
+      "loss": 0.3496,
+      "step": 1245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79866142456538e-05,
+      "loss": 0.3402,
+      "step": 1246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798302476471862e-05,
+      "loss": 0.3268,
+      "step": 1247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979432445793762e-05,
+      "loss": 0.3408,
+      "step": 1248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975837290156296e-05,
+      "loss": 0.3318,
+      "step": 1249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972239299084326e-05,
+      "loss": 0.3544,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796863847385694e-05,
+      "loss": 0.349,
+      "step": 1251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796503481575424e-05,
+      "loss": 0.3517,
+      "step": 1252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961428326057348e-05,
+      "loss": 0.3369,
+      "step": 1253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795781900604837e-05,
+      "loss": 0.3518,
+      "step": 1254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795420685701044e-05,
+      "loss": 0.3296,
+      "step": 1255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795059188022768e-05,
+      "loss": 0.3342,
+      "step": 1256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946974076985234e-05,
+      "loss": 0.3395,
+      "step": 1257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943353448569235e-05,
+      "loss": 0.3416,
+      "step": 1258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793972999626683e-05,
+      "loss": 0.3325,
+      "step": 1259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936103721366164e-05,
+      "loss": 0.3328,
+      "step": 1260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932474625156397e-05,
+      "loss": 0.3376,
+      "step": 1261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928842708927673e-05,
+      "loss": 0.3448,
+      "step": 1262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925207973971163e-05,
+      "loss": 0.3539,
+      "step": 1263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921570421579014e-05,
+      "loss": 0.3634,
+      "step": 1264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791793005304439e-05,
+      "loss": 0.325,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914286869661453e-05,
+      "loss": 0.3444,
+      "step": 1266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910640872725375e-05,
+      "loss": 0.3322,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906992063532304e-05,
+      "loss": 0.3403,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790334044337941e-05,
+      "loss": 0.3403,
+      "step": 1269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899686013564862e-05,
+      "loss": 0.3435,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896028775387807e-05,
+      "loss": 0.338,
+      "step": 1271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892368730148415e-05,
+      "loss": 0.3481,
+      "step": 1272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788870587914784e-05,
+      "loss": 0.3463,
+      "step": 1273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885040223688234e-05,
+      "loss": 0.3369,
+      "step": 1274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881371765072754e-05,
+      "loss": 0.3505,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877700504605546e-05,
+      "loss": 0.3352,
+      "step": 1276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787402644359175e-05,
+      "loss": 0.3291,
+      "step": 1277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870349583337514e-05,
+      "loss": 0.3544,
+      "step": 1278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786666992514997e-05,
+      "loss": 0.327,
+      "step": 1279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862987470337242e-05,
+      "loss": 0.3419,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785930222020846e-05,
+      "loss": 0.343,
+      "step": 1281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855614176073742e-05,
+      "loss": 0.3259,
+      "step": 1282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851923339244194e-05,
+      "loss": 0.3271,
+      "step": 1283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848229711031923e-05,
+      "loss": 0.3264,
+      "step": 1284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844533292750025e-05,
+      "loss": 0.3283,
+      "step": 1285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840834085712592e-05,
+      "loss": 0.3356,
+      "step": 1286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837132091234695e-05,
+      "loss": 0.3379,
+      "step": 1287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783342731063241e-05,
+      "loss": 0.3405,
+      "step": 1288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829719745222797e-05,
+      "loss": 0.3365,
+      "step": 1289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7826009396323907e-05,
+      "loss": 0.3413,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822296265254777e-05,
+      "loss": 0.3374,
+      "step": 1291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818580353335438e-05,
+      "loss": 0.3397,
+      "step": 1292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814861661886915e-05,
+      "loss": 0.333,
+      "step": 1293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811140192231202e-05,
+      "loss": 0.3441,
+      "step": 1294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807415945691306e-05,
+      "loss": 0.3352,
+      "step": 1295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803688923591194e-05,
+      "loss": 0.3251,
+      "step": 1296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799959127255845e-05,
+      "loss": 0.3511,
+      "step": 1297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796226558011206e-05,
+      "loss": 0.3124,
+      "step": 1298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792491217184223e-05,
+      "loss": 0.3408,
+      "step": 1299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788753106102813e-05,
+      "loss": 0.3455,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785012226095895e-05,
+      "loss": 0.3381,
+      "step": 1301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781268578493357e-05,
+      "loss": 0.3356,
+      "step": 1302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777752216462608e-05,
+      "loss": 0.3605,
+      "step": 1303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773772985825925e-05,
+      "loss": 0.3481,
+      "step": 1304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770021043425734e-05,
+      "loss": 0.3447,
+      "step": 1305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766266338759338e-05,
+      "loss": 0.334,
+      "step": 1306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762508873161542e-05,
+      "loss": 0.3348,
+      "step": 1307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758748647968144e-05,
+      "loss": 0.3307,
+      "step": 1308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77549856645159e-05,
+      "loss": 0.3367,
+      "step": 1309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775121992414258e-05,
+      "loss": 0.3608,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747451428186907e-05,
+      "loss": 0.3281,
+      "step": 1311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743680177988595e-05,
+      "loss": 0.3273,
+      "step": 1312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773990617488833e-05,
+      "loss": 0.3411,
+      "step": 1313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773612942022779e-05,
+      "loss": 0.3467,
+      "step": 1314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732349915349617e-05,
+      "loss": 0.3479,
+      "step": 1315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772856766159744e-05,
+      "loss": 0.3312,
+      "step": 1316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7724782660315858e-05,
+      "loss": 0.3416,
+      "step": 1317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720994912850452e-05,
+      "loss": 0.3478,
+      "step": 1318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771720442054778e-05,
+      "loss": 0.3228,
+      "step": 1319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771341118475537e-05,
+      "loss": 0.3433,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770961520682173e-05,
+      "loss": 0.3451,
+      "step": 1321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705816488096345e-05,
+      "loss": 0.3379,
+      "step": 1322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702015029929667e-05,
+      "loss": 0.3438,
+      "step": 1323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698210833673127e-05,
+      "loss": 0.3494,
+      "step": 1324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769440390067913e-05,
+      "loss": 0.3345,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690594232301052e-05,
+      "loss": 0.3406,
+      "step": 1326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768678182989324e-05,
+      "loss": 0.3514,
+      "step": 1327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682966694811017e-05,
+      "loss": 0.3356,
+      "step": 1328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767914882841067e-05,
+      "loss": 0.3311,
+      "step": 1329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767532823204947e-05,
+      "loss": 0.3521,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671504907085645e-05,
+      "loss": 0.3457,
+      "step": 1331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667678854878397e-05,
+      "loss": 0.3587,
+      "step": 1332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663850076787906e-05,
+      "loss": 0.3452,
+      "step": 1333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660018574175316e-05,
+      "loss": 0.3361,
+      "step": 1334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656184348402725e-05,
+      "loss": 0.3277,
+      "step": 1335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765234740083322e-05,
+      "loss": 0.339,
+      "step": 1336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648507732830854e-05,
+      "loss": 0.3533,
+      "step": 1337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644665345760626e-05,
+      "loss": 0.3287,
+      "step": 1338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764082024098853e-05,
+      "loss": 0.3422,
+      "step": 1339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636972419881507e-05,
+      "loss": 0.3596,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7633121883807465e-05,
+      "loss": 0.3326,
+      "step": 1341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762926863413529e-05,
+      "loss": 0.3381,
+      "step": 1342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625412672234815e-05,
+      "loss": 0.342,
+      "step": 1343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621553999476854e-05,
+      "loss": 0.3398,
+      "step": 1344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761769261723317e-05,
+      "loss": 0.3502,
+      "step": 1345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613828526876503e-05,
+      "loss": 0.3616,
+      "step": 1346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609961729780544e-05,
+      "loss": 0.3335,
+      "step": 1347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606092227319952e-05,
+      "loss": 0.343,
+      "step": 1348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760222002087035e-05,
+      "loss": 0.3395,
+      "step": 1349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598345111808315e-05,
+      "loss": 0.3345,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759446750151139e-05,
+      "loss": 0.3279,
+      "step": 1351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759058719135808e-05,
+      "loss": 0.3328,
+      "step": 1352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586704182727846e-05,
+      "loss": 0.3295,
+      "step": 1353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758281847700111e-05,
+      "loss": 0.3447,
+      "step": 1354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757893007555925e-05,
+      "loss": 0.3405,
+      "step": 1355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575038979784604e-05,
+      "loss": 0.3408,
+      "step": 1356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757114519106047e-05,
+      "loss": 0.3557,
+      "step": 1357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567248710771104e-05,
+      "loss": 0.34,
+      "step": 1358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563349540301713e-05,
+      "loss": 0.3217,
+      "step": 1359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559447681038466e-05,
+      "loss": 0.3396,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555543134368485e-05,
+      "loss": 0.3318,
+      "step": 1361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755163590167985e-05,
+      "loss": 0.3376,
+      "step": 1362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754772598436159e-05,
+      "loss": 0.34,
+      "step": 1363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543813383803695e-05,
+      "loss": 0.3431,
+      "step": 1364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539898101397114e-05,
+      "loss": 0.3141,
+      "step": 1365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535980138533727e-05,
+      "loss": 0.3496,
+      "step": 1366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753205949660639e-05,
+      "loss": 0.3484,
+      "step": 1367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528136177008908e-05,
+      "loss": 0.3293,
+      "step": 1368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524210181136022e-05,
+      "loss": 0.3463,
+      "step": 1369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520281510383443e-05,
+      "loss": 0.3223,
+      "step": 1370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516350166147827e-05,
+      "loss": 0.3412,
+      "step": 1371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751241614982677e-05,
+      "loss": 0.334,
+      "step": 1372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508479462818836e-05,
+      "loss": 0.3331,
+      "step": 1373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504540106523525e-05,
+      "loss": 0.3424,
+      "step": 1374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750059808234129e-05,
+      "loss": 0.3413,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496653391673533e-05,
+      "loss": 0.3431,
+      "step": 1376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492706035922605e-05,
+      "loss": 0.3452,
+      "step": 1377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488756016491803e-05,
+      "loss": 0.3353,
+      "step": 1378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748480333478537e-05,
+      "loss": 0.3367,
+      "step": 1379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480847992208495e-05,
+      "loss": 0.3354,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476889990167314e-05,
+      "loss": 0.3307,
+      "step": 1381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747292933006891e-05,
+      "loss": 0.3458,
+      "step": 1382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7468966013321316e-05,
+      "loss": 0.3469,
+      "step": 1383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465000041333496e-05,
+      "loss": 0.3217,
+      "step": 1384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7461031415515366e-05,
+      "loss": 0.3357,
+      "step": 1385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457060137277783e-05,
+      "loss": 0.3311,
+      "step": 1386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745308620803255e-05,
+      "loss": 0.3513,
+      "step": 1387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449109629192418e-05,
+      "loss": 0.3427,
+      "step": 1388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445130402171063e-05,
+      "loss": 0.3215,
+      "step": 1389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441148528383113e-05,
+      "loss": 0.3331,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437164009244143e-05,
+      "loss": 0.328,
+      "step": 1391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743317684617066e-05,
+      "loss": 0.3351,
+      "step": 1392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429187040580105e-05,
+      "loss": 0.3141,
+      "step": 1393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425194593890873e-05,
+      "loss": 0.3302,
+      "step": 1394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421199507522297e-05,
+      "loss": 0.3463,
+      "step": 1395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741720178289463e-05,
+      "loss": 0.3365,
+      "step": 1396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7413201421429077e-05,
+      "loss": 0.3448,
+      "step": 1397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409198424547788e-05,
+      "loss": 0.3373,
+      "step": 1398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405192793673834e-05,
+      "loss": 0.3445,
+      "step": 1399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740118453023123e-05,
+      "loss": 0.3497,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739717363564492e-05,
+      "loss": 0.3136,
+      "step": 1401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393160111340802e-05,
+      "loss": 0.3489,
+      "step": 1402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389143958745687e-05,
+      "loss": 0.3215,
+      "step": 1403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738512517928733e-05,
+      "loss": 0.3362,
+      "step": 1404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381103774394424e-05,
+      "loss": 0.3293,
+      "step": 1405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377079745496586e-05,
+      "loss": 0.3522,
+      "step": 1406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7373053094024373e-05,
+      "loss": 0.3442,
+      "step": 1407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736902382140927e-05,
+      "loss": 0.3389,
+      "step": 1408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364991929083697e-05,
+      "loss": 0.3318,
+      "step": 1409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360957418481006e-05,
+      "loss": 0.3347,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356920291035474e-05,
+      "loss": 0.3317,
+      "step": 1411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352880548182315e-05,
+      "loss": 0.344,
+      "step": 1412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348838191357668e-05,
+      "loss": 0.3302,
+      "step": 1413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344793221998605e-05,
+      "loss": 0.3491,
+      "step": 1414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734074564154312e-05,
+      "loss": 0.3331,
+      "step": 1415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336695451430152e-05,
+      "loss": 0.3203,
+      "step": 1416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332642653099542e-05,
+      "loss": 0.3323,
+      "step": 1417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732858724799208e-05,
+      "loss": 0.3309,
+      "step": 1418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7324529237549474e-05,
+      "loss": 0.34,
+      "step": 1419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320468623214358e-05,
+      "loss": 0.3353,
+      "step": 1420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316405406430294e-05,
+      "loss": 0.3385,
+      "step": 1421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312339588641766e-05,
+      "loss": 0.3237,
+      "step": 1422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308271171294184e-05,
+      "loss": 0.3324,
+      "step": 1423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730420015583389e-05,
+      "loss": 0.3345,
+      "step": 1424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730012654370813e-05,
+      "loss": 0.3419,
+      "step": 1425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72960503363651e-05,
+      "loss": 0.3181,
+      "step": 1426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729197153525389e-05,
+      "loss": 0.3214,
+      "step": 1427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287890141824537e-05,
+      "loss": 0.3285,
+      "step": 1428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283806157527984e-05,
+      "loss": 0.3395,
+      "step": 1429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279719583816102e-05,
+      "loss": 0.3212,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275630422141675e-05,
+      "loss": 0.339,
+      "step": 1431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727153867395842e-05,
+      "loss": 0.322,
+      "step": 1432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267444340720962e-05,
+      "loss": 0.3351,
+      "step": 1433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263347423884848e-05,
+      "loss": 0.3236,
+      "step": 1434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725924792490655e-05,
+      "loss": 0.3411,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255145845243444e-05,
+      "loss": 0.3296,
+      "step": 1436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725104118635384e-05,
+      "loss": 0.323,
+      "step": 1437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7246933949696948e-05,
+      "loss": 0.3339,
+      "step": 1438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724282413673291e-05,
+      "loss": 0.3298,
+      "step": 1439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238711748922776e-05,
+      "loss": 0.3275,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234596787728513e-05,
+      "loss": 0.3461,
+      "step": 1441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230479254613e-05,
+      "loss": 0.3282,
+      "step": 1442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722635915104003e-05,
+      "loss": 0.3348,
+      "step": 1443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722223647847432e-05,
+      "loss": 0.3264,
+      "step": 1444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721811123838149e-05,
+      "loss": 0.3287,
+      "step": 1445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721398343222807e-05,
+      "loss": 0.3224,
+      "step": 1446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720985306148151e-05,
+      "loss": 0.3381,
+      "step": 1447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720572012761018e-05,
+      "loss": 0.3324,
+      "step": 1448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201584632083332e-05,
+      "loss": 0.3216,
+      "step": 1449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197446576371163e-05,
+      "loss": 0.3324,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193305961944758e-05,
+      "loss": 0.3453,
+      "step": 1451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189162790276116e-05,
+      "loss": 0.3381,
+      "step": 1452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718501706283815e-05,
+      "loss": 0.3289,
+      "step": 1453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718086878110468e-05,
+      "loss": 0.3171,
+      "step": 1454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176717946550432e-05,
+      "loss": 0.33,
+      "step": 1455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717256456065104e-05,
+      "loss": 0.3218,
+      "step": 1456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716840862488304e-05,
+      "loss": 0.3257,
+      "step": 1457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716425014072389e-05,
+      "loss": 0.3281,
+      "step": 1458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160089109651937e-05,
+      "loss": 0.3522,
+      "step": 1459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715592553314644e-05,
+      "loss": 0.3181,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151759412687573e-05,
+      "loss": 0.3324,
+      "step": 1461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147590749756388e-05,
+      "loss": 0.3286,
+      "step": 1462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7143419545834866e-05,
+      "loss": 0.335,
+      "step": 1463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713924580240589e-05,
+      "loss": 0.3159,
+      "step": 1464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135069520953225e-05,
+      "loss": 0.3337,
+      "step": 1465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130890702961562e-05,
+      "loss": 0.3168,
+      "step": 1466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7126709349916478e-05,
+      "loss": 0.327,
+      "step": 1467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122525463304456e-05,
+      "loss": 0.334,
+      "step": 1468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118339044612888e-05,
+      "loss": 0.3351,
+      "step": 1469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114150095330048e-05,
+      "loss": 0.3311,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109958616945133e-05,
+      "loss": 0.3342,
+      "step": 1471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105764610948212e-05,
+      "loss": 0.3468,
+      "step": 1472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101568078830278e-05,
+      "loss": 0.3242,
+      "step": 1473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7097369022083208e-05,
+      "loss": 0.3339,
+      "step": 1474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7093167442199775e-05,
+      "loss": 0.3304,
+      "step": 1475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708896334067366e-05,
+      "loss": 0.3211,
+      "step": 1476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708475671899943e-05,
+      "loss": 0.3414,
+      "step": 1477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080547578672556e-05,
+      "loss": 0.3219,
+      "step": 1478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7076335921189395e-05,
+      "loss": 0.3473,
+      "step": 1479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7072121748047203e-05,
+      "loss": 0.3148,
+      "step": 1480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067905060744134e-05,
+      "loss": 0.3257,
+      "step": 1481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706368586077924e-05,
+      "loss": 0.3213,
+      "step": 1482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705946414965245e-05,
+      "loss": 0.3293,
+      "step": 1483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055239928864596e-05,
+      "loss": 0.3326,
+      "step": 1484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051013199917406e-05,
+      "loss": 0.3278,
+      "step": 1485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046783964313496e-05,
+      "loss": 0.3417,
+      "step": 1486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042552223556366e-05,
+      "loss": 0.3325,
+      "step": 1487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7038317979150418e-05,
+      "loss": 0.3342,
+      "step": 1488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034081232600934e-05,
+      "loss": 0.3295,
+      "step": 1489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029841985414095e-05,
+      "loss": 0.3298,
+      "step": 1490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025600239096962e-05,
+      "loss": 0.3268,
+      "step": 1491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021355995157495e-05,
+      "loss": 0.3256,
+      "step": 1492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017109255104526e-05,
+      "loss": 0.3241,
+      "step": 1493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012860020447797e-05,
+      "loss": 0.3295,
+      "step": 1494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7008608292697914e-05,
+      "loss": 0.3433,
+      "step": 1495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700435407336638e-05,
+      "loss": 0.3263,
+      "step": 1496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000097363965584e-05,
+      "loss": 0.3331,
+      "step": 1497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6995838166008803e-05,
+      "loss": 0.3342,
+      "step": 1498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991576481010193e-05,
+      "loss": 0.3608,
+      "step": 1499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698731231048479e-05,
+      "loss": 0.3315,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983045655948528e-05,
+      "loss": 0.3354,
+      "step": 1501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978776518918213e-05,
+      "loss": 0.3422,
+      "step": 1502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6974504900911536e-05,
+      "loss": 0.3184,
+      "step": 1503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697023080344707e-05,
+      "loss": 0.3307,
+      "step": 1504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965954228044265e-05,
+      "loss": 0.3246,
+      "step": 1505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961675176223465e-05,
+      "loss": 0.3492,
+      "step": 1506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957393649505883e-05,
+      "loss": 0.329,
+      "step": 1507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695310964941361e-05,
+      "loss": 0.3278,
+      "step": 1508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948823177469627e-05,
+      "loss": 0.3346,
+      "step": 1509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694453423519779e-05,
+      "loss": 0.3572,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940242824122823e-05,
+      "loss": 0.3505,
+      "step": 1511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693594894577034e-05,
+      "loss": 0.3308,
+      "step": 1512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693165260166683e-05,
+      "loss": 0.3298,
+      "step": 1513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927353793339652e-05,
+      "loss": 0.334,
+      "step": 1514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692305252231705e-05,
+      "loss": 0.3463,
+      "step": 1515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691874879012813e-05,
+      "loss": 0.3091,
+      "step": 1516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691444259830289e-05,
+      "loss": 0.3243,
+      "step": 1517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69101339483722e-05,
+      "loss": 0.3348,
+      "step": 1518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905822841867788e-05,
+      "loss": 0.3403,
+      "step": 1519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901509280322263e-05,
+      "loss": 0.3242,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897193265269122e-05,
+      "loss": 0.3433,
+      "step": 1521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689287479824271e-05,
+      "loss": 0.3309,
+      "step": 1522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888553880778263e-05,
+      "loss": 0.3295,
+      "step": 1523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6884230514411876e-05,
+      "loss": 0.3245,
+      "step": 1524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6879904700680516e-05,
+      "loss": 0.3357,
+      "step": 1525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6875576441122032e-05,
+      "loss": 0.326,
+      "step": 1526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871245737275124e-05,
+      "loss": 0.3326,
+      "step": 1527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686691259067938e-05,
+      "loss": 0.3359,
+      "step": 1528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862577002875237e-05,
+      "loss": 0.3245,
+      "step": 1529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858238975404017e-05,
+      "loss": 0.3363,
+      "step": 1530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853898509807897e-05,
+      "loss": 0.3367,
+      "step": 1531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849555607629932e-05,
+      "loss": 0.3204,
+      "step": 1532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684521027041403e-05,
+      "loss": 0.3253,
+      "step": 1533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840862499704976e-05,
+      "loss": 0.342,
+      "step": 1534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836512297048417e-05,
+      "loss": 0.3551,
+      "step": 1535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832159663990858e-05,
+      "loss": 0.332,
+      "step": 1536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827804602079675e-05,
+      "loss": 0.3575,
+      "step": 1537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682344711286311e-05,
+      "loss": 0.3298,
+      "step": 1538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681908719789026e-05,
+      "loss": 0.3468,
+      "step": 1539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814724858711088e-05,
+      "loss": 0.3245,
+      "step": 1540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681036009687642e-05,
+      "loss": 0.326,
+      "step": 1541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680599291393794e-05,
+      "loss": 0.3355,
+      "step": 1542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6801623311448202e-05,
+      "loss": 0.3439,
+      "step": 1543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797251290960603e-05,
+      "loss": 0.3238,
+      "step": 1544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6792876854029416e-05,
+      "loss": 0.3307,
+      "step": 1545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6788500002209764e-05,
+      "loss": 0.3381,
+      "step": 1546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784120737057637e-05,
+      "loss": 0.3285,
+      "step": 1547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6779739060129867e-05,
+      "loss": 0.3364,
+      "step": 1548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677535497298416e-05,
+      "loss": 0.3397,
+      "step": 1549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770968477179072e-05,
+      "loss": 0.3384,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676657957427402e-05,
+      "loss": 0.3278,
+      "step": 1551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762188265829257e-05,
+      "loss": 0.3383,
+      "step": 1552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6757794553405924e-05,
+      "loss": 0.3226,
+      "step": 1553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675339843856599e-05,
+      "loss": 0.3229,
+      "step": 1554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6748999922872292e-05,
+      "loss": 0.3258,
+      "step": 1555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744599007888512e-05,
+      "loss": 0.3406,
+      "step": 1556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674019569517919e-05,
+      "loss": 0.338,
+      "step": 1557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735789986309714e-05,
+      "loss": 0.3171,
+      "step": 1558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731381882846332e-05,
+      "loss": 0.3536,
+      "step": 1559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726971386356137e-05,
+      "loss": 0.3506,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722558498407075e-05,
+      "loss": 0.3319,
+      "step": 1561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718143220567934e-05,
+      "loss": 0.3326,
+      "step": 1562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713725554408367e-05,
+      "loss": 0.3392,
+      "step": 1563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709305501498863e-05,
+      "loss": 0.3372,
+      "step": 1564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704883063410767e-05,
+      "loss": 0.3252,
+      "step": 1565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670045824171627e-05,
+      "loss": 0.3274,
+      "step": 1566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669603103798841e-05,
+      "loss": 0.3381,
+      "step": 1567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691601453801067e-05,
+      "loss": 0.3184,
+      "step": 1568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687169490728975e-05,
+      "loss": 0.3302,
+      "step": 1569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6682735150347706e-05,
+      "loss": 0.3086,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6678298434233687e-05,
+      "loss": 0.3246,
+      "step": 1571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6673859343964184e-05,
+      "loss": 0.3461,
+      "step": 1572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669417881117302e-05,
+      "loss": 0.3298,
+      "step": 1573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6664974047272004e-05,
+      "loss": 0.3446,
+      "step": 1574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660527844008075e-05,
+      "loss": 0.3406,
+      "step": 1575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665607927290616e-05,
+      "loss": 0.3456,
+      "step": 1576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651628335547743e-05,
+      "loss": 0.3295,
+      "step": 1577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647175033515136e-05,
+      "loss": 0.3318,
+      "step": 1578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664271936839151e-05,
+      "loss": 0.3113,
+      "step": 1579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638261341760867e-05,
+      "loss": 0.3168,
+      "step": 1580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633800955208043e-05,
+      "loss": 0.3232,
+      "step": 1581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629338210318725e-05,
+      "loss": 0.314,
+      "step": 1582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662487310867943e-05,
+      "loss": 0.3229,
+      "step": 1583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620405651877516e-05,
+      "loss": 0.3192,
+      "step": 1584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615935841501175e-05,
+      "loss": 0.3407,
+      "step": 1585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611463679139444e-05,
+      "loss": 0.323,
+      "step": 1586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606989166382182e-05,
+      "loss": 0.3364,
+      "step": 1587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6602512304820102e-05,
+      "loss": 0.3356,
+      "step": 1588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6598033096044733e-05,
+      "loss": 0.3405,
+      "step": 1589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659355154164845e-05,
+      "loss": 0.3334,
+      "step": 1590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589067643224463e-05,
+      "loss": 0.3346,
+      "step": 1591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658458140236681e-05,
+      "loss": 0.3301,
+      "step": 1592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658009282067036e-05,
+      "loss": 0.3326,
+      "step": 1593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657560189973082e-05,
+      "loss": 0.325,
+      "step": 1594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657110864114473e-05,
+      "loss": 0.3285,
+      "step": 1595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566613046509448e-05,
+      "loss": 0.3319,
+      "step": 1596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6562115117423178e-05,
+      "loss": 0.3386,
+      "step": 1597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557614855484944e-05,
+      "loss": 0.3342,
+      "step": 1598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655311226229461e-05,
+      "loss": 0.3138,
+      "step": 1599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.3446,
+      "step": 1600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654410008856119e-05,
+      "loss": 0.321,
+      "step": 1601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6539590511221964e-05,
+      "loss": 0.3256,
+      "step": 1602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6535078609038343e-05,
+      "loss": 0.328,
+      "step": 1603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653056438361432e-05,
+      "loss": 0.3419,
+      "step": 1604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652604783655472e-05,
+      "loss": 0.3198,
+      "step": 1605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6521528969465188e-05,
+      "loss": 0.3416,
+      "step": 1606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517007783952185e-05,
+      "loss": 0.3242,
+      "step": 1607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6512484281623023e-05,
+      "loss": 0.3281,
+      "step": 1608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650795846408581e-05,
+      "loss": 0.3269,
+      "step": 1609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650343033294949e-05,
+      "loss": 0.3297,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498899889823828e-05,
+      "loss": 0.3269,
+      "step": 1611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494367136319415e-05,
+      "loss": 0.3336,
+      "step": 1612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648983207404765e-05,
+      "loss": 0.332,
+      "step": 1613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648529470462077e-05,
+      "loss": 0.3295,
+      "step": 1614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648075502965181e-05,
+      "loss": 0.328,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647621305075466e-05,
+      "loss": 0.3293,
+      "step": 1616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647166876954399e-05,
+      "loss": 0.3335,
+      "step": 1617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467122187635308e-05,
+      "loss": 0.3247,
+      "step": 1618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646257330664494e-05,
+      "loss": 0.3287,
+      "step": 1619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458022128190035e-05,
+      "loss": 0.3228,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453468653888537e-05,
+      "loss": 0.3131,
+      "step": 1621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644891288535923e-05,
+      "loss": 0.3306,
+      "step": 1622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444354824221698e-05,
+      "loss": 0.3252,
+      "step": 1623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439794472096353e-05,
+      "loss": 0.3413,
+      "step": 1624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435231830604404e-05,
+      "loss": 0.3176,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643066690136789e-05,
+      "loss": 0.3354,
+      "step": 1626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642609968600966e-05,
+      "loss": 0.3297,
+      "step": 1627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421530186153367e-05,
+      "loss": 0.3337,
+      "step": 1628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641695840342348e-05,
+      "loss": 0.3142,
+      "step": 1629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641238433944529e-05,
+      "loss": 0.3269,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407807995844888e-05,
+      "loss": 0.3253,
+      "step": 1631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403229374249177e-05,
+      "loss": 0.3289,
+      "step": 1632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639864847628587e-05,
+      "loss": 0.3196,
+      "step": 1633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639406530358349e-05,
+      "loss": 0.3157,
+      "step": 1634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6389479857771372e-05,
+      "loss": 0.3433,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638489214047965e-05,
+      "loss": 0.3363,
+      "step": 1636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638030215333928e-05,
+      "loss": 0.3295,
+      "step": 1637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375709897982014e-05,
+      "loss": 0.3185,
+      "step": 1638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6371115376040407e-05,
+      "loss": 0.3425,
+      "step": 1639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636651858914783e-05,
+      "loss": 0.3163,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6361919538938457e-05,
+      "loss": 0.3258,
+      "step": 1641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357318227047255e-05,
+      "loss": 0.3531,
+      "step": 1642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6352714655110018e-05,
+      "loss": 0.3356,
+      "step": 1643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348108824763318e-05,
+      "loss": 0.3212,
+      "step": 1644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634350073764455e-05,
+      "loss": 0.3402,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6338890395391896e-05,
+      "loss": 0.3344,
+      "step": 1646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6334277799644355e-05,
+      "loss": 0.3324,
+      "step": 1647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329662952041712e-05,
+      "loss": 0.3269,
+      "step": 1648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325045854224565e-05,
+      "loss": 0.3395,
+      "step": 1649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63204265078343e-05,
+      "loss": 0.3341,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631580491451311e-05,
+      "loss": 0.3206,
+      "step": 1651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311181075903986e-05,
+      "loss": 0.3245,
+      "step": 1652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306554993650727e-05,
+      "loss": 0.334,
+      "step": 1653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6301926669397902e-05,
+      "loss": 0.3352,
+      "step": 1654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297296104790904e-05,
+      "loss": 0.3359,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6292663301475915e-05,
+      "loss": 0.3154,
+      "step": 1656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288028261099905e-05,
+      "loss": 0.3307,
+      "step": 1657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628339098531065e-05,
+      "loss": 0.3322,
+      "step": 1658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278751475756712e-05,
+      "loss": 0.3289,
+      "step": 1659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627410973408745e-05,
+      "loss": 0.342,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269465761953023e-05,
+      "loss": 0.3098,
+      "step": 1661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6264819561004374e-05,
+      "loss": 0.3195,
+      "step": 1662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6260171132893233e-05,
+      "loss": 0.3324,
+      "step": 1663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6255520479272148e-05,
+      "loss": 0.3193,
+      "step": 1664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250867601794426e-05,
+      "loss": 0.3326,
+      "step": 1665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246212502114185e-05,
+      "loss": 0.3479,
+      "step": 1666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624155518188633e-05,
+      "loss": 0.325,
+      "step": 1667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236895642766547e-05,
+      "loss": 0.3308,
+      "step": 1668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232233886411316e-05,
+      "loss": 0.3341,
+      "step": 1669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6227569914477914e-05,
+      "loss": 0.3213,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6222903728624387e-05,
+      "loss": 0.3239,
+      "step": 1671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218235330509582e-05,
+      "loss": 0.3188,
+      "step": 1672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621356472179313e-05,
+      "loss": 0.3312,
+      "step": 1673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208891904135445e-05,
+      "loss": 0.3353,
+      "step": 1674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6204216879197733e-05,
+      "loss": 0.3309,
+      "step": 1675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199539648641974e-05,
+      "loss": 0.3271,
+      "step": 1676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619486021413094e-05,
+      "loss": 0.3121,
+      "step": 1677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190178577328183e-05,
+      "loss": 0.3197,
+      "step": 1678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185494739898038e-05,
+      "loss": 0.3343,
+      "step": 1679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180808703505628e-05,
+      "loss": 0.3535,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617612046981685e-05,
+      "loss": 0.3409,
+      "step": 1681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6171430040498384e-05,
+      "loss": 0.3339,
+      "step": 1682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166737417217696e-05,
+      "loss": 0.3331,
+      "step": 1683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6162042601643024e-05,
+      "loss": 0.3418,
+      "step": 1684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615734559544339e-05,
+      "loss": 0.3151,
+      "step": 1685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152646400288602e-05,
+      "loss": 0.3271,
+      "step": 1686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6147945017849225e-05,
+      "loss": 0.3231,
+      "step": 1687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6143241449796627e-05,
+      "loss": 0.3127,
+      "step": 1688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6138535697802927e-05,
+      "loss": 0.3185,
+      "step": 1689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133827763541047e-05,
+      "loss": 0.3466,
+      "step": 1690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612911764868467e-05,
+      "loss": 0.3169,
+      "step": 1691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124405354908246e-05,
+      "loss": 0.337,
+      "step": 1692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119690883887022e-05,
+      "loss": 0.3231,
+      "step": 1693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114974237297e-05,
+      "loss": 0.3245,
+      "step": 1694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6110255416814967e-05,
+      "loss": 0.3337,
+      "step": 1695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105534424118477e-05,
+      "loss": 0.3134,
+      "step": 1696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610081126088585e-05,
+      "loss": 0.3282,
+      "step": 1697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096085928796196e-05,
+      "loss": 0.3372,
+      "step": 1698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609135842952938e-05,
+      "loss": 0.3248,
+      "step": 1699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086628764766037e-05,
+      "loss": 0.3317,
+      "step": 1700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081896936187588e-05,
+      "loss": 0.3199,
+      "step": 1701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077162945476202e-05,
+      "loss": 0.3335,
+      "step": 1702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6072426794314836e-05,
+      "loss": 0.3206,
+      "step": 1703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60676884843872e-05,
+      "loss": 0.3348,
+      "step": 1704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606294801737778e-05,
+      "loss": 0.3291,
+      "step": 1705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6058205394971828e-05,
+      "loss": 0.3264,
+      "step": 1706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053460618855358e-05,
+      "loss": 0.3275,
+      "step": 1707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604871369071515e-05,
+      "loss": 0.3223,
+      "step": 1708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604396461223876e-05,
+      "loss": 0.3408,
+      "step": 1709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039213385114496e-05,
+      "loss": 0.3243,
+      "step": 1710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603446001103143e-05,
+      "loss": 0.3158,
+      "step": 1711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029704491679405e-05,
+      "loss": 0.327,
+      "step": 1712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024946828749018e-05,
+      "loss": 0.324,
+      "step": 1713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602018702393164e-05,
+      "loss": 0.3136,
+      "step": 1714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015425078919392e-05,
+      "loss": 0.3352,
+      "step": 1715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6010660995405157e-05,
+      "loss": 0.3167,
+      "step": 1716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005894775082586e-05,
+      "loss": 0.3301,
+      "step": 1717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001126419646085e-05,
+      "loss": 0.3229,
+      "step": 1718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996355930790813e-05,
+      "loss": 0.313,
+      "step": 1719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5991583310212703e-05,
+      "loss": 0.3097,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986808559608423e-05,
+      "loss": 0.3188,
+      "step": 1721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982031680675424e-05,
+      "loss": 0.3367,
+      "step": 1722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59772526751119e-05,
+      "loss": 0.334,
+      "step": 1723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597247154461679e-05,
+      "loss": 0.3195,
+      "step": 1724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5967688290889817e-05,
+      "loss": 0.3319,
+      "step": 1725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596290291563143e-05,
+      "loss": 0.3384,
+      "step": 1726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595811542054285e-05,
+      "loss": 0.3378,
+      "step": 1727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953325807326042e-05,
+      "loss": 0.3265,
+      "step": 1728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594853407768373e-05,
+      "loss": 0.3477,
+      "step": 1729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943740233319393e-05,
+      "loss": 0.3384,
+      "step": 1730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938944275937257e-05,
+      "loss": 0.3156,
+      "step": 1731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934146207242287e-05,
+      "loss": 0.3353,
+      "step": 1732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5929346028940225e-05,
+      "loss": 0.3037,
+      "step": 1733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924543742737547e-05,
+      "loss": 0.3256,
+      "step": 1734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5919739350341473e-05,
+      "loss": 0.3229,
+      "step": 1735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591493285345999e-05,
+      "loss": 0.3118,
+      "step": 1736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910124253801806e-05,
+      "loss": 0.3232,
+      "step": 1737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590531355307641e-05,
+      "loss": 0.3462,
+      "step": 1738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900500752994016e-05,
+      "loss": 0.3275,
+      "step": 1739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895685855265582e-05,
+      "loss": 0.329,
+      "step": 1740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890868861602825e-05,
+      "loss": 0.3219,
+      "step": 1741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58860497737182e-05,
+      "loss": 0.3207,
+      "step": 1742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588122859332491e-05,
+      "loss": 0.344,
+      "step": 1743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876405322136892e-05,
+      "loss": 0.3231,
+      "step": 1744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587157996186884e-05,
+      "loss": 0.3379,
+      "step": 1745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586675251423618e-05,
+      "loss": 0.3198,
+      "step": 1746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5861922980955092e-05,
+      "loss": 0.3409,
+      "step": 1747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5857091363742483e-05,
+      "loss": 0.3265,
+      "step": 1748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852257664316007e-05,
+      "loss": 0.324,
+      "step": 1749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5847421884394064e-05,
+      "loss": 0.3232,
+      "step": 1750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842584025695784e-05,
+      "loss": 0.3231,
+      "step": 1751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5837744089941044e-05,
+      "loss": 0.3364,
+      "step": 1752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583290207885045e-05,
+      "loss": 0.3265,
+      "step": 1753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828057994145363e-05,
+      "loss": 0.3207,
+      "step": 1754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582321183754786e-05,
+      "loss": 0.3147,
+      "step": 1755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5818363610780767e-05,
+      "loss": 0.3434,
+      "step": 1756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5813513315567643e-05,
+      "loss": 0.3375,
+      "step": 1757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5808660953632783e-05,
+      "loss": 0.3326,
+      "step": 1758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580380652670122e-05,
+      "loss": 0.3199,
+      "step": 1759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798950036498713e-05,
+      "loss": 0.3339,
+      "step": 1760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579409148475176e-05,
+      "loss": 0.3414,
+      "step": 1761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789230873187593e-05,
+      "loss": 0.317,
+      "step": 1762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5784368203534174e-05,
+      "loss": 0.3386,
+      "step": 1763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5779503477520193e-05,
+      "loss": 0.3276,
+      "step": 1764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5774636696875078e-05,
+      "loss": 0.3114,
+      "step": 1765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769767863328986e-05,
+      "loss": 0.3195,
+      "step": 1766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57648969786128e-05,
+      "loss": 0.3436,
+      "step": 1767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760024044458134e-05,
+      "loss": 0.3363,
+      "step": 1768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755149062597332e-05,
+      "loss": 0.3289,
+      "step": 1769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5750272034763466e-05,
+      "loss": 0.3213,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745392962690334e-05,
+      "loss": 0.3235,
+      "step": 1771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574051184811246e-05,
+      "loss": 0.3201,
+      "step": 1772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735628692765092e-05,
+      "loss": 0.3204,
+      "step": 1773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5730743498384207e-05,
+      "loss": 0.3253,
+      "step": 1774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5725856266706515e-05,
+      "loss": 0.3329,
+      "step": 1775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572096699946944e-05,
+      "loss": 0.3207,
+      "step": 1776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571607569841112e-05,
+      "loss": 0.3136,
+      "step": 1777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5711182365270436e-05,
+      "loss": 0.3218,
+      "step": 1778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5706287001786986e-05,
+      "loss": 0.3292,
+      "step": 1779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5701389609701076e-05,
+      "loss": 0.3181,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5696490190753755e-05,
+      "loss": 0.3386,
+      "step": 1781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691588746686775e-05,
+      "loss": 0.3271,
+      "step": 1782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686685279242618e-05,
+      "loss": 0.3134,
+      "step": 1783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568177979016448e-05,
+      "loss": 0.3229,
+      "step": 1784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676872281196272e-05,
+      "loss": 0.3244,
+      "step": 1785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567196275408264e-05,
+      "loss": 0.3262,
+      "step": 1786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667051210568925e-05,
+      "loss": 0.3286,
+      "step": 1787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56621376524012e-05,
+      "loss": 0.3206,
+      "step": 1788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657222081326256e-05,
+      "loss": 0.3295,
+      "step": 1789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5652304499091582e-05,
+      "loss": 0.3455,
+      "step": 1790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647384907445402e-05,
+      "loss": 0.3207,
+      "step": 1791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564246330813664e-05,
+      "loss": 0.3109,
+      "step": 1792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637539702914947e-05,
+      "loss": 0.327,
+      "step": 1793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563261409353067e-05,
+      "loss": 0.3328,
+      "step": 1794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627686481734886e-05,
+      "loss": 0.3251,
+      "step": 1795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622756869279375e-05,
+      "loss": 0.3219,
+      "step": 1796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617825257916623e-05,
+      "loss": 0.3179,
+      "step": 1797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5612891649399838e-05,
+      "loss": 0.3308,
+      "step": 1798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560795604548294e-05,
+      "loss": 0.3262,
+      "step": 1799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560301844792054e-05,
+      "loss": 0.3133,
+      "step": 1800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559807885846797e-05,
+      "loss": 0.3289,
+      "step": 1801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559313727888127e-05,
+      "loss": 0.3042,
+      "step": 1802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5588193710917194e-05,
+      "loss": 0.3137,
+      "step": 1803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558324815633319e-05,
+      "loss": 0.3248,
+      "step": 1804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578300616887414e-05,
+      "loss": 0.3135,
+      "step": 1805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573351094338737e-05,
+      "loss": 0.3274,
+      "step": 1806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568399590446727e-05,
+      "loss": 0.3301,
+      "step": 1807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556344610697166e-05,
+      "loss": 0.3187,
+      "step": 1808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558490645674516e-05,
+      "loss": 0.3352,
+      "step": 1809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553533208316972e-05,
+      "loss": 0.3406,
+      "step": 1810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5548573796661414e-05,
+      "loss": 0.3357,
+      "step": 1811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554361241247093e-05,
+      "loss": 0.3212,
+      "step": 1812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538649057509307e-05,
+      "loss": 0.33,
+      "step": 1813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553368373354103e-05,
+      "loss": 0.3153,
+      "step": 1814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528716442331284e-05,
+      "loss": 0.3265,
+      "step": 1815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5523747185645966e-05,
+      "loss": 0.3289,
+      "step": 1816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518775965251655e-05,
+      "loss": 0.3251,
+      "step": 1817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5513802782915636e-05,
+      "loss": 0.3127,
+      "step": 1818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5508827640405894e-05,
+      "loss": 0.3119,
+      "step": 1819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5503850539491102e-05,
+      "loss": 0.3331,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498871481940638e-05,
+      "loss": 0.3164,
+      "step": 1821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493890469524573e-05,
+      "loss": 0.3257,
+      "step": 1822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5488907504013666e-05,
+      "loss": 0.3315,
+      "step": 1823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483922587179386e-05,
+      "loss": 0.3302,
+      "step": 1824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547893572079388e-05,
+      "loss": 0.3226,
+      "step": 1825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547394690662999e-05,
+      "loss": 0.3186,
+      "step": 1826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5468956146461263e-05,
+      "loss": 0.3051,
+      "step": 1827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463963442061926e-05,
+      "loss": 0.3132,
+      "step": 1828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5458968795206902e-05,
+      "loss": 0.3252,
+      "step": 1829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54539722076718e-05,
+      "loss": 0.3204,
+      "step": 1830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448973681232923e-05,
+      "loss": 0.3317,
+      "step": 1831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443973217667267e-05,
+      "loss": 0.3375,
+      "step": 1832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543897081875251e-05,
+      "loss": 0.3116,
+      "step": 1833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433966486267014e-05,
+      "loss": 0.3059,
+      "step": 1834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428960221989844e-05,
+      "loss": 0.3148,
+      "step": 1835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423952027700737e-05,
+      "loss": 0.3402,
+      "step": 1836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541894190518012e-05,
+      "loss": 0.3159,
+      "step": 1837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541392985620911e-05,
+      "loss": 0.3186,
+      "step": 1838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408915882569505e-05,
+      "loss": 0.329,
+      "step": 1839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403899986043787e-05,
+      "loss": 0.3428,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398882168415124e-05,
+      "loss": 0.311,
+      "step": 1841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539386243146736e-05,
+      "loss": 0.3267,
+      "step": 1842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5388840776985034e-05,
+      "loss": 0.3291,
+      "step": 1843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383817206753356e-05,
+      "loss": 0.3097,
+      "step": 1844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378791722558225e-05,
+      "loss": 0.3278,
+      "step": 1845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537376432618621e-05,
+      "loss": 0.3224,
+      "step": 1846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536873501942457e-05,
+      "loss": 0.3121,
+      "step": 1847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363703804061233e-05,
+      "loss": 0.3309,
+      "step": 1848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358670681884816e-05,
+      "loss": 0.3134,
+      "step": 1849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5353635654684613e-05,
+      "loss": 0.3269,
+      "step": 1850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5348598724250584e-05,
+      "loss": 0.3118,
+      "step": 1851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343559892373375e-05,
+      "loss": 0.3253,
+      "step": 1852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533851916084431e-05,
+      "loss": 0.3254,
+      "step": 1853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333476531455382e-05,
+      "loss": 0.3245,
+      "step": 1854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328432005999265e-05,
+      "loss": 0.2957,
+      "step": 1855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532338558626929e-05,
+      "loss": 0.3259,
+      "step": 1856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5318337274059492e-05,
+      "loss": 0.322,
+      "step": 1857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313287071164546e-05,
+      "loss": 0.3203,
+      "step": 1858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530823497937983e-05,
+      "loss": 0.3248,
+      "step": 1859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303181000501366e-05,
+      "loss": 0.3295,
+      "step": 1860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298125136325864e-05,
+      "loss": 0.3297,
+      "step": 1861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5293067388650694e-05,
+      "loss": 0.3303,
+      "step": 1862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528800775927391e-05,
+      "loss": 0.3259,
+      "step": 1863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282946249994214e-05,
+      "loss": 0.3173,
+      "step": 1864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277882862611004e-05,
+      "loss": 0.3221,
+      "step": 1865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272817598924312e-05,
+      "loss": 0.3132,
+      "step": 1866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5267750460734867e-05,
+      "loss": 0.3351,
+      "step": 1867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262681449844047e-05,
+      "loss": 0.3331,
+      "step": 1868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257610568053901e-05,
+      "loss": 0.3063,
+      "step": 1869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252537817167143e-05,
+      "loss": 0.3192,
+      "step": 1870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5247463198987157e-05,
+      "loss": 0.3116,
+      "step": 1871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242386715317974e-05,
+      "loss": 0.3235,
+      "step": 1872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237308367964312e-05,
+      "loss": 0.329,
+      "step": 1873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5232228158731528e-05,
+      "loss": 0.3252,
+      "step": 1874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227146089425659e-05,
+      "loss": 0.3234,
+      "step": 1875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222062161853396e-05,
+      "loss": 0.3115,
+      "step": 1876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216976377822086e-05,
+      "loss": 0.3193,
+      "step": 1877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211888739139743e-05,
+      "loss": 0.3156,
+      "step": 1878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206799247615038e-05,
+      "loss": 0.3376,
+      "step": 1879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201707905057304e-05,
+      "loss": 0.3037,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196614713276526e-05,
+      "loss": 0.3295,
+      "step": 1881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191519674083348e-05,
+      "loss": 0.3031,
+      "step": 1882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186422789289074e-05,
+      "loss": 0.3354,
+      "step": 1883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181324060705663e-05,
+      "loss": 0.3289,
+      "step": 1884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176223490145725e-05,
+      "loss": 0.3246,
+      "step": 1885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517112107942253e-05,
+      "loss": 0.3391,
+      "step": 1886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166016830350004e-05,
+      "loss": 0.3278,
+      "step": 1887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5160910744742716e-05,
+      "loss": 0.3075,
+      "step": 1888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155802824415902e-05,
+      "loss": 0.3264,
+      "step": 1889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150693071185441e-05,
+      "loss": 0.3101,
+      "step": 1890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145581486867863e-05,
+      "loss": 0.3307,
+      "step": 1891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140468073280357e-05,
+      "loss": 0.3153,
+      "step": 1892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135352832240755e-05,
+      "loss": 0.3131,
+      "step": 1893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5130235765567536e-05,
+      "loss": 0.3401,
+      "step": 1894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512511687507984e-05,
+      "loss": 0.3167,
+      "step": 1895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119996162597444e-05,
+      "loss": 0.3132,
+      "step": 1896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114873629940779e-05,
+      "loss": 0.3212,
+      "step": 1897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510974927893092e-05,
+      "loss": 0.3383,
+      "step": 1898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104623111389588e-05,
+      "loss": 0.3138,
+      "step": 1899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099495129139154e-05,
+      "loss": 0.3243,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094365334002631e-05,
+      "loss": 0.3148,
+      "step": 1901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089233727803676e-05,
+      "loss": 0.3291,
+      "step": 1902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5084100312366591e-05,
+      "loss": 0.3129,
+      "step": 1903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078965089516317e-05,
+      "loss": 0.309,
+      "step": 1904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073828061078448e-05,
+      "loss": 0.3282,
+      "step": 1905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068689228879209e-05,
+      "loss": 0.3111,
+      "step": 1906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5063548594745472e-05,
+      "loss": 0.3391,
+      "step": 1907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5058406160504749e-05,
+      "loss": 0.3247,
+      "step": 1908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5053261927985187e-05,
+      "loss": 0.3192,
+      "step": 1909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048115899015585e-05,
+      "loss": 0.3237,
+      "step": 1910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042968075425362e-05,
+      "loss": 0.3085,
+      "step": 1911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037818459044589e-05,
+      "loss": 0.3353,
+      "step": 1912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5032667051703973e-05,
+      "loss": 0.3161,
+      "step": 1913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027513855234852e-05,
+      "loss": 0.3403,
+      "step": 1914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022358871469203e-05,
+      "loss": 0.3263,
+      "step": 1915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501720210223964e-05,
+      "loss": 0.3203,
+      "step": 1916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012043549379405e-05,
+      "loss": 0.3235,
+      "step": 1917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500688321472239e-05,
+      "loss": 0.318,
+      "step": 1918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001721100103104e-05,
+      "loss": 0.3306,
+      "step": 1919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996557207356688e-05,
+      "loss": 0.3171,
+      "step": 1920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499139153831893e-05,
+      "loss": 0.3072,
+      "step": 1921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986224094826241e-05,
+      "loss": 0.3287,
+      "step": 1922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498105487871566e-05,
+      "loss": 0.329,
+      "step": 1923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4975883891824858e-05,
+      "loss": 0.3389,
+      "step": 1924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497071113599214e-05,
+      "loss": 0.3264,
+      "step": 1925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4965536613056438e-05,
+      "loss": 0.3267,
+      "step": 1926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960360324857302e-05,
+      "loss": 0.3265,
+      "step": 1927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4955182273234925e-05,
+      "loss": 0.3221,
+      "step": 1928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950002460030119e-05,
+      "loss": 0.3055,
+      "step": 1929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4944820887084329e-05,
+      "loss": 0.3452,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939637556239607e-05,
+      "loss": 0.3204,
+      "step": 1931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934452469338653e-05,
+      "loss": 0.3242,
+      "step": 1932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929265628224778e-05,
+      "loss": 0.306,
+      "step": 1933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924077034741924e-05,
+      "loss": 0.3232,
+      "step": 1934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918886690734647e-05,
+      "loss": 0.3242,
+      "step": 1935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491369459804813e-05,
+      "loss": 0.3376,
+      "step": 1936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490850075852818e-05,
+      "loss": 0.319,
+      "step": 1937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903305174021224e-05,
+      "loss": 0.3025,
+      "step": 1938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898107846374301e-05,
+      "loss": 0.33,
+      "step": 1939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4892908777435087e-05,
+      "loss": 0.3213,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887707969051859e-05,
+      "loss": 0.3302,
+      "step": 1941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882505423073525e-05,
+      "loss": 0.3192,
+      "step": 1942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877301141349602e-05,
+      "loss": 0.319,
+      "step": 1943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872095125730232e-05,
+      "loss": 0.3359,
+      "step": 1944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4866887378066169e-05,
+      "loss": 0.3264,
+      "step": 1945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861677900208782e-05,
+      "loss": 0.3339,
+      "step": 1946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856466694010055e-05,
+      "loss": 0.3339,
+      "step": 1947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851253761322589e-05,
+      "loss": 0.3264,
+      "step": 1948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4846039103999594e-05,
+      "loss": 0.3104,
+      "step": 1949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840822723894906e-05,
+      "loss": 0.3334,
+      "step": 1950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835604622862954e-05,
+      "loss": 0.3315,
+      "step": 1951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830384802758793e-05,
+      "loss": 0.3306,
+      "step": 1952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825163265438085e-05,
+      "loss": 0.3055,
+      "step": 1953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4819940012757105e-05,
+      "loss": 0.3149,
+      "step": 1954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481471504657273e-05,
+      "loss": 0.3369,
+      "step": 1955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809488368742458e-05,
+      "loss": 0.3165,
+      "step": 1956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804259981124381e-05,
+      "loss": 0.3259,
+      "step": 1957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479902988557721e-05,
+      "loss": 0.3218,
+      "step": 1958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793798083960264e-05,
+      "loss": 0.325,
+      "step": 1959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478856457813346e-05,
+      "loss": 0.312,
+      "step": 1960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783329369957327e-05,
+      "loss": 0.3209,
+      "step": 1961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778092461292994e-05,
+      "loss": 0.3354,
+      "step": 1962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772853854002204e-05,
+      "loss": 0.3311,
+      "step": 1963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4767613549947293e-05,
+      "loss": 0.3303,
+      "step": 1964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4762371550991205e-05,
+      "loss": 0.3191,
+      "step": 1965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757127858997495e-05,
+      "loss": 0.3149,
+      "step": 1966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751882475830302e-05,
+      "loss": 0.3137,
+      "step": 1967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4746635403354377e-05,
+      "loss": 0.3115,
+      "step": 1968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741386643435074e-05,
+      "loss": 0.3202,
+      "step": 1969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736136197938338e-05,
+      "loss": 0.3015,
+      "step": 1970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730884068730726e-05,
+      "loss": 0.3286,
+      "step": 1971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725630257679381e-05,
+      "loss": 0.3153,
+      "step": 1972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4720374766652048e-05,
+      "loss": 0.3336,
+      "step": 1973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471511759751707e-05,
+      "loss": 0.3145,
+      "step": 1974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709858752143393e-05,
+      "loss": 0.3269,
+      "step": 1975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704598232400544e-05,
+      "loss": 0.3027,
+      "step": 1976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469933604015866e-05,
+      "loss": 0.3168,
+      "step": 1977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694072177288464e-05,
+      "loss": 0.3345,
+      "step": 1978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688806645661276e-05,
+      "loss": 0.3063,
+      "step": 1979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683539447149007e-05,
+      "loss": 0.3209,
+      "step": 1980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678270583624162e-05,
+      "loss": 0.3047,
+      "step": 1981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467300005695984e-05,
+      "loss": 0.3169,
+      "step": 1982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4667727869029731e-05,
+      "loss": 0.3217,
+      "step": 1983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662454021708107e-05,
+      "loss": 0.3325,
+      "step": 1984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657178516869844e-05,
+      "loss": 0.3243,
+      "step": 1985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651901356390397e-05,
+      "loss": 0.3365,
+      "step": 1986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646622542145815e-05,
+      "loss": 0.3179,
+      "step": 1987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464134207601273e-05,
+      "loss": 0.3254,
+      "step": 1988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636059959868363e-05,
+      "loss": 0.314,
+      "step": 1989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630776195590524e-05,
+      "loss": 0.3369,
+      "step": 1990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4625490785057608e-05,
+      "loss": 0.3252,
+      "step": 1991
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620203730148592e-05,
+      "loss": 0.3193,
+      "step": 1992
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614915032743045e-05,
+      "loss": 0.3046,
+      "step": 1993
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460962469472111e-05,
+      "loss": 0.2923,
+      "step": 1994
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604332717963523e-05,
+      "loss": 0.3183,
+      "step": 1995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4599039104351591e-05,
+      "loss": 0.3291,
+      "step": 1996
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593743855767217e-05,
+      "loss": 0.326,
+      "step": 1997
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4588446974092873e-05,
+      "loss": 0.3213,
+      "step": 1998
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583148461211622e-05,
+      "loss": 0.3192,
+      "step": 1999
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577848319007098e-05,
+      "loss": 0.3289,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572546549363515e-05,
+      "loss": 0.3108,
+      "step": 2001
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567243154165671e-05,
+      "loss": 0.314,
+      "step": 2002
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561938135298944e-05,
+      "loss": 0.3049,
+      "step": 2003
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556631494649278e-05,
+      "loss": 0.3169,
+      "step": 2004
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45513232341032e-05,
+      "loss": 0.3164,
+      "step": 2005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546013355547814e-05,
+      "loss": 0.3309,
+      "step": 2006
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540701860870803e-05,
+      "loss": 0.3123,
+      "step": 2007
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535388751960411e-05,
+      "loss": 0.323,
+      "step": 2008
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530074030705472e-05,
+      "loss": 0.333,
+      "step": 2009
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452475769899538e-05,
+      "loss": 0.3042,
+      "step": 2010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519439758720108e-05,
+      "loss": 0.306,
+      "step": 2011
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514120211770206e-05,
+      "loss": 0.3264,
+      "step": 2012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508799060036782e-05,
+      "loss": 0.3232,
+      "step": 2013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503476305411526e-05,
+      "loss": 0.3246,
+      "step": 2014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498151949786689e-05,
+      "loss": 0.3291,
+      "step": 2015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4492825995055104e-05,
+      "loss": 0.3192,
+      "step": 2016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487498443110153e-05,
+      "loss": 0.3145,
+      "step": 2017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482169295845805e-05,
+      "loss": 0.312,
+      "step": 2018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476838555156587e-05,
+      "loss": 0.3171,
+      "step": 2019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471506222937593e-05,
+      "loss": 0.3167,
+      "step": 2020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446617230108448e-05,
+      "loss": 0.3027,
+      "step": 2021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460836791493477e-05,
+      "loss": 0.3196,
+      "step": 2022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455499696061374e-05,
+      "loss": 0.3118,
+      "step": 2023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450161016685525e-05,
+      "loss": 0.3083,
+      "step": 2024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444820755263847e-05,
+      "loss": 0.3241,
+      "step": 2025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4439478913694814e-05,
+      "loss": 0.3256,
+      "step": 2026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434135493877473e-05,
+      "loss": 0.3281,
+      "step": 2027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428790497711424e-05,
+      "loss": 0.3331,
+      "step": 2028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442344392709683e-05,
+      "loss": 0.3139,
+      "step": 2029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418095783934413e-05,
+      "loss": 0.3065,
+      "step": 2030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412746070125454e-05,
+      "loss": 0.3109,
+      "step": 2031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407394787571794e-05,
+      "loss": 0.2982,
+      "step": 2032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4402041938175826e-05,
+      "loss": 0.3174,
+      "step": 2033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396687523840506e-05,
+      "loss": 0.3134,
+      "step": 2034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439133154646935e-05,
+      "loss": 0.3235,
+      "step": 2035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438597400796642e-05,
+      "loss": 0.3041,
+      "step": 2036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4380614910236338e-05,
+      "loss": 0.3242,
+      "step": 2037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375254255184275e-05,
+      "loss": 0.2995,
+      "step": 2038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369892044715965e-05,
+      "loss": 0.312,
+      "step": 2039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364528280737692e-05,
+      "loss": 0.3072,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359162965156288e-05,
+      "loss": 0.3112,
+      "step": 2041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353796099879139e-05,
+      "loss": 0.3146,
+      "step": 2042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348427686814179e-05,
+      "loss": 0.3242,
+      "step": 2043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43430577278699e-05,
+      "loss": 0.3186,
+      "step": 2044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337686224955335e-05,
+      "loss": 0.312,
+      "step": 2045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332313179980077e-05,
+      "loss": 0.3323,
+      "step": 2046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326938594854251e-05,
+      "loss": 0.327,
+      "step": 2047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321562471488543e-05,
+      "loss": 0.3091,
+      "step": 2048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316184811794181e-05,
+      "loss": 0.3293,
+      "step": 2049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310805617682936e-05,
+      "loss": 0.3199,
+      "step": 2050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305424891067135e-05,
+      "loss": 0.321,
+      "step": 2051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300042633859637e-05,
+      "loss": 0.3114,
+      "step": 2052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429465884797385e-05,
+      "loss": 0.3247,
+      "step": 2053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4289273535323732e-05,
+      "loss": 0.3145,
+      "step": 2054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428388669782377e-05,
+      "loss": 0.3207,
+      "step": 2055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427849833738901e-05,
+      "loss": 0.3176,
+      "step": 2056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273108455935027e-05,
+      "loss": 0.3,
+      "step": 2057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4267717055377939e-05,
+      "loss": 0.3118,
+      "step": 2058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262324137634402e-05,
+      "loss": 0.3164,
+      "step": 2059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4256929704621625e-05,
+      "loss": 0.3224,
+      "step": 2060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251533758257336e-05,
+      "loss": 0.3087,
+      "step": 2061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4246136300459816e-05,
+      "loss": 0.3112,
+      "step": 2062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240737333147874e-05,
+      "loss": 0.3187,
+      "step": 2063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235336858240863e-05,
+      "loss": 0.3214,
+      "step": 2064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229934877658662e-05,
+      "loss": 0.3148,
+      "step": 2065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224531393321696e-05,
+      "loss": 0.3129,
+      "step": 2066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219126407150925e-05,
+      "loss": 0.3203,
+      "step": 2067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213719921067828e-05,
+      "loss": 0.3134,
+      "step": 2068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208311936994434e-05,
+      "loss": 0.3254,
+      "step": 2069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202902456853296e-05,
+      "loss": 0.3164,
+      "step": 2070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41974914825675e-05,
+      "loss": 0.3145,
+      "step": 2071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419207901606067e-05,
+      "loss": 0.3069,
+      "step": 2072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186665059256946e-05,
+      "loss": 0.3304,
+      "step": 2073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181249614081007e-05,
+      "loss": 0.3296,
+      "step": 2074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175832682458064e-05,
+      "loss": 0.3137,
+      "step": 2075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417041426631385e-05,
+      "loss": 0.3274,
+      "step": 2076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4164994367574631e-05,
+      "loss": 0.335,
+      "step": 2077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41595729881672e-05,
+      "loss": 0.3334,
+      "step": 2078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.3181,
+      "step": 2079
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4148725795057477e-05,
+      "loss": 0.3126,
+      "step": 2080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41432999852114e-05,
+      "loss": 0.3331,
+      "step": 2081
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137872702409526e-05,
+      "loss": 0.3169,
+      "step": 2082
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413244394858127e-05,
+      "loss": 0.3214,
+      "step": 2083
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127013725656573e-05,
+      "loss": 0.3214,
+      "step": 2084
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121582035565897e-05,
+      "loss": 0.3184,
+      "step": 2085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411614888024022e-05,
+      "loss": 0.3171,
+      "step": 2086
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110714261611046e-05,
+      "loss": 0.3328,
+      "step": 2087
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41052781816104e-05,
+      "loss": 0.3329,
+      "step": 2088
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4099840642170827e-05,
+      "loss": 0.3196,
+      "step": 2089
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094401645225384e-05,
+      "loss": 0.3127,
+      "step": 2090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4088961192707654e-05,
+      "loss": 0.3052,
+      "step": 2091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083519286551734e-05,
+      "loss": 0.3086,
+      "step": 2092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078075928692243e-05,
+      "loss": 0.323,
+      "step": 2093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072631121064304e-05,
+      "loss": 0.3184,
+      "step": 2094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067184865603565e-05,
+      "loss": 0.3018,
+      "step": 2095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061737164246188e-05,
+      "loss": 0.3199,
+      "step": 2096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4056288018928845e-05,
+      "loss": 0.3282,
+      "step": 2097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050837431588727e-05,
+      "loss": 0.3124,
+      "step": 2098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045385404163533e-05,
+      "loss": 0.3162,
+      "step": 2099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039931938591472e-05,
+      "loss": 0.3073,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034477036811272e-05,
+      "loss": 0.3179,
+      "step": 2101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029020700762165e-05,
+      "loss": 0.3071,
+      "step": 2102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023562932383899e-05,
+      "loss": 0.3164,
+      "step": 2103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018103733616723e-05,
+      "loss": 0.3086,
+      "step": 2104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012643106401402e-05,
+      "loss": 0.3046,
+      "step": 2105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4007181052679202e-05,
+      "loss": 0.3172,
+      "step": 2106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001717574391902e-05,
+      "loss": 0.329,
+      "step": 2107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996252673481782e-05,
+      "loss": 0.324,
+      "step": 2108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3990786351891639e-05,
+      "loss": 0.3226,
+      "step": 2109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985318611564761e-05,
+      "loss": 0.3301,
+      "step": 2110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979849454444947e-05,
+      "loss": 0.3133,
+      "step": 2111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974378882476498e-05,
+      "loss": 0.3156,
+      "step": 2112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3968906897604222e-05,
+      "loss": 0.3264,
+      "step": 2113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3963433501773422e-05,
+      "loss": 0.3234,
+      "step": 2114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957958696929913e-05,
+      "loss": 0.3218,
+      "step": 2115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3952482485019998e-05,
+      "loss": 0.3281,
+      "step": 2116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947004867990495e-05,
+      "loss": 0.2896,
+      "step": 2117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3941525847788704e-05,
+      "loss": 0.3264,
+      "step": 2118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936045426362441e-05,
+      "loss": 0.3158,
+      "step": 2119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930563605660013e-05,
+      "loss": 0.3219,
+      "step": 2120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925080387630217e-05,
+      "loss": 0.3134,
+      "step": 2121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919595774222359e-05,
+      "loss": 0.2883,
+      "step": 2122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914109767386235e-05,
+      "loss": 0.3016,
+      "step": 2123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908622369072133e-05,
+      "loss": 0.3189,
+      "step": 2124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390313358123085e-05,
+      "loss": 0.3149,
+      "step": 2125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389764340581366e-05,
+      "loss": 0.3258,
+      "step": 2126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892151844772335e-05,
+      "loss": 0.3223,
+      "step": 2127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3886658900059141e-05,
+      "loss": 0.3236,
+      "step": 2128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881164573626845e-05,
+      "loss": 0.3087,
+      "step": 2129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875668867428693e-05,
+      "loss": 0.3119,
+      "step": 2130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870171783418422e-05,
+      "loss": 0.326,
+      "step": 2131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864673323550264e-05,
+      "loss": 0.3167,
+      "step": 2132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3859173489778937e-05,
+      "loss": 0.3243,
+      "step": 2133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853672284059654e-05,
+      "loss": 0.3201,
+      "step": 2134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3848169708348103e-05,
+      "loss": 0.3102,
+      "step": 2135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3842665764600473e-05,
+      "loss": 0.319,
+      "step": 2136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3837160454773428e-05,
+      "loss": 0.3248,
+      "step": 2137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831653780824125e-05,
+      "loss": 0.3262,
+      "step": 2138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826145744710204e-05,
+      "loss": 0.2993,
+      "step": 2139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820636348389787e-05,
+      "loss": 0.321,
+      "step": 2140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3815125593821481e-05,
+      "loss": 0.3182,
+      "step": 2141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380961348296438e-05,
+      "loss": 0.3135,
+      "step": 2142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3804100017778048e-05,
+      "loss": 0.3141,
+      "step": 2143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3798585200222547e-05,
+      "loss": 0.3135,
+      "step": 2144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793069032258406e-05,
+      "loss": 0.3197,
+      "step": 2145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787551515846644e-05,
+      "loss": 0.3488,
+      "step": 2146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3782032652948751e-05,
+      "loss": 0.3265,
+      "step": 2147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776512445526705e-05,
+      "loss": 0.3141,
+      "step": 2148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770990895542946e-05,
+      "loss": 0.3192,
+      "step": 2149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765468004960414e-05,
+      "loss": 0.3401,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375994377574251e-05,
+      "loss": 0.3145,
+      "step": 2151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754418209853108e-05,
+      "loss": 0.3047,
+      "step": 2152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3748891309256569e-05,
+      "loss": 0.3201,
+      "step": 2153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3743363075917723e-05,
+      "loss": 0.3215,
+      "step": 2154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737833511801872e-05,
+      "loss": 0.3156,
+      "step": 2155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732302618874798e-05,
+      "loss": 0.3116,
+      "step": 2156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726770399102743e-05,
+      "loss": 0.307,
+      "step": 2157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721236854452431e-05,
+      "loss": 0.3165,
+      "step": 2158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715701986891054e-05,
+      "loss": 0.318,
+      "step": 2159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3710165798386277e-05,
+      "loss": 0.3121,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704628290906228e-05,
+      "loss": 0.3193,
+      "step": 2161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699089466419513e-05,
+      "loss": 0.3198,
+      "step": 2162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693549326895193e-05,
+      "loss": 0.3226,
+      "step": 2163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368800787430281e-05,
+      "loss": 0.3021,
+      "step": 2164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682465110612361e-05,
+      "loss": 0.3108,
+      "step": 2165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676921037794329e-05,
+      "loss": 0.2981,
+      "step": 2166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3671375657819638e-05,
+      "loss": 0.299,
+      "step": 2167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665828972659689e-05,
+      "loss": 0.3132,
+      "step": 2168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3660280984286348e-05,
+      "loss": 0.317,
+      "step": 2169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654731694671938e-05,
+      "loss": 0.3224,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649181105789251e-05,
+      "loss": 0.3015,
+      "step": 2171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364362921961154e-05,
+      "loss": 0.3033,
+      "step": 2172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638076038112514e-05,
+      "loss": 0.3158,
+      "step": 2173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632521563266347e-05,
+      "loss": 0.311,
+      "step": 2174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3626965797047676e-05,
+      "loss": 0.3101,
+      "step": 2175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621408741431583e-05,
+      "loss": 0.3035,
+      "step": 2176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3615850398393628e-05,
+      "loss": 0.325,
+      "step": 2177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610290769909815e-05,
+      "loss": 0.332,
+      "step": 2178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360472985795661e-05,
+      "loss": 0.3079,
+      "step": 2179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599167664510933e-05,
+      "loss": 0.3008,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593604191550157e-05,
+      "loss": 0.3179,
+      "step": 2181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588039441052123e-05,
+      "loss": 0.3066,
+      "step": 2182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3582473414995112e-05,
+      "loss": 0.3238,
+      "step": 2183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357690611535786e-05,
+      "loss": 0.3204,
+      "step": 2184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357133754411956e-05,
+      "loss": 0.323,
+      "step": 2185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356576770325986e-05,
+      "loss": 0.313,
+      "step": 2186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560196594758855e-05,
+      "loss": 0.3309,
+      "step": 2187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554624220597088e-05,
+      "loss": 0.3109,
+      "step": 2188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549050582755556e-05,
+      "loss": 0.3216,
+      "step": 2189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3543475683215703e-05,
+      "loss": 0.3229,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537899523959428e-05,
+      "loss": 0.3126,
+      "step": 2191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532322106969066e-05,
+      "loss": 0.3204,
+      "step": 2192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526743434227413e-05,
+      "loss": 0.3146,
+      "step": 2193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521163507717695e-05,
+      "loss": 0.3066,
+      "step": 2194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515582329423604e-05,
+      "loss": 0.3241,
+      "step": 2195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509999901329257e-05,
+      "loss": 0.3032,
+      "step": 2196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3504416225419226e-05,
+      "loss": 0.3165,
+      "step": 2197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498831303678525e-05,
+      "loss": 0.3119,
+      "step": 2198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493245138092616e-05,
+      "loss": 0.3203,
+      "step": 2199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348765773064739e-05,
+      "loss": 0.316,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3482069083329192e-05,
+      "loss": 0.3104,
+      "step": 2201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34764791981248e-05,
+      "loss": 0.3164,
+      "step": 2202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470888077021442e-05,
+      "loss": 0.3146,
+      "step": 2203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465295722006775e-05,
+      "loss": 0.3197,
+      "step": 2204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459702135068895e-05,
+      "loss": 0.3091,
+      "step": 2205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454107318196339e-05,
+      "loss": 0.3138,
+      "step": 2206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448511273378088e-05,
+      "loss": 0.3223,
+      "step": 2207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442914002603547e-05,
+      "loss": 0.3202,
+      "step": 2208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437315507862568e-05,
+      "loss": 0.3148,
+      "step": 2209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3431715791145425e-05,
+      "loss": 0.3339,
+      "step": 2210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426114854442844e-05,
+      "loss": 0.2991,
+      "step": 2211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420512699745968e-05,
+      "loss": 0.3196,
+      "step": 2212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414909329046383e-05,
+      "loss": 0.3315,
+      "step": 2213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3409304744336104e-05,
+      "loss": 0.3097,
+      "step": 2214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403698947607579e-05,
+      "loss": 0.3042,
+      "step": 2215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398091940853686e-05,
+      "loss": 0.3093,
+      "step": 2216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339248372606773e-05,
+      "loss": 0.3188,
+      "step": 2217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3386874305243447e-05,
+      "loss": 0.3036,
+      "step": 2218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381263680375012e-05,
+      "loss": 0.3206,
+      "step": 2219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375651853457012e-05,
+      "loss": 0.3229,
+      "step": 2220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337003882648447e-05,
+      "loss": 0.3101,
+      "step": 2221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364424601452835e-05,
+      "loss": 0.3167,
+      "step": 2222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358809180357981e-05,
+      "loss": 0.3198,
+      "step": 2223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3353192565196205e-05,
+      "loss": 0.3309,
+      "step": 2224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347574757964237e-05,
+      "loss": 0.3221,
+      "step": 2225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341955760659218e-05,
+      "loss": 0.3281,
+      "step": 2226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336335575278723e-05,
+      "loss": 0.3061,
+      "step": 2227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330714203820741e-05,
+      "loss": 0.3241,
+      "step": 2228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3325091648283687e-05,
+      "loss": 0.3192,
+      "step": 2229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319467910666406e-05,
+      "loss": 0.3175,
+      "step": 2230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331384299296814e-05,
+      "loss": 0.3376,
+      "step": 2231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308216897188575e-05,
+      "loss": 0.3053,
+      "step": 2232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33025896253278e-05,
+      "loss": 0.3246,
+      "step": 2233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296961179386328e-05,
+      "loss": 0.3141,
+      "step": 2234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291331561365091e-05,
+      "loss": 0.3235,
+      "step": 2235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285700773265435e-05,
+      "loss": 0.3164,
+      "step": 2236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280068817089117e-05,
+      "loss": 0.3085,
+      "step": 2237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3274435694838318e-05,
+      "loss": 0.3173,
+      "step": 2238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268801408515635e-05,
+      "loss": 0.3285,
+      "step": 2239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326316596012407e-05,
+      "loss": 0.3108,
+      "step": 2240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257529351667042e-05,
+      "loss": 0.3065,
+      "step": 2241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325189158514838e-05,
+      "loss": 0.3231,
+      "step": 2242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324625266257233e-05,
+      "loss": 0.3119,
+      "step": 2243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3240612585943547e-05,
+      "loss": 0.3182,
+      "step": 2244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3234971357267093e-05,
+      "loss": 0.3159,
+      "step": 2245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3229328978548445e-05,
+      "loss": 0.3004,
+      "step": 2246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223685451793483e-05,
+      "loss": 0.3218,
+      "step": 2247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218040779008501e-05,
+      "loss": 0.2939,
+      "step": 2248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212394962200195e-05,
+      "loss": 0.3125,
+      "step": 2249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3206748003375668e-05,
+      "loss": 0.3094,
+      "step": 2250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3201099904542438e-05,
+      "loss": 0.3011,
+      "step": 2251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3195450667708418e-05,
+      "loss": 0.3223,
+      "step": 2252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3189800294881926e-05,
+      "loss": 0.3047,
+      "step": 2253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3184148788071686e-05,
+      "loss": 0.3185,
+      "step": 2254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317849614928683e-05,
+      "loss": 0.2922,
+      "step": 2255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172842380536892e-05,
+      "loss": 0.3042,
+      "step": 2256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167187483831799e-05,
+      "loss": 0.309,
+      "step": 2257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316153146118188e-05,
+      "loss": 0.3225,
+      "step": 2258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155874314597873e-05,
+      "loss": 0.3034,
+      "step": 2259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150216046090914e-05,
+      "loss": 0.3369,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144556657672527e-05,
+      "loss": 0.3169,
+      "step": 2261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313889615135465e-05,
+      "loss": 0.3164,
+      "step": 2262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3133234529149605e-05,
+      "loss": 0.3134,
+      "step": 2263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312757179307012e-05,
+      "loss": 0.3151,
+      "step": 2264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121907945129308e-05,
+      "loss": 0.3392,
+      "step": 2265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311624298734069e-05,
+      "loss": 0.3279,
+      "step": 2266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110576921718177e-05,
+      "loss": 0.3385,
+      "step": 2267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104909750276069e-05,
+      "loss": 0.3173,
+      "step": 2268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3099241475029064e-05,
+      "loss": 0.3042,
+      "step": 2269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093572097992247e-05,
+      "loss": 0.3221,
+      "step": 2270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087901621181107e-05,
+      "loss": 0.3184,
+      "step": 2271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308223004661151e-05,
+      "loss": 0.308,
+      "step": 2272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076557376299723e-05,
+      "loss": 0.3259,
+      "step": 2273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070883612262391e-05,
+      "loss": 0.3152,
+      "step": 2274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065208756516557e-05,
+      "loss": 0.2991,
+      "step": 2275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059532811079653e-05,
+      "loss": 0.3279,
+      "step": 2276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053855777969492e-05,
+      "loss": 0.296,
+      "step": 2277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048177659204276e-05,
+      "loss": 0.2987,
+      "step": 2278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042498456802592e-05,
+      "loss": 0.3123,
+      "step": 2279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036818172783421e-05,
+      "loss": 0.3131,
+      "step": 2280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031136809166114e-05,
+      "loss": 0.309,
+      "step": 2281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025454367970416e-05,
+      "loss": 0.3298,
+      "step": 2282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3019770851216453e-05,
+      "loss": 0.312,
+      "step": 2283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301408626092473e-05,
+      "loss": 0.3114,
+      "step": 2284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008400599116138e-05,
+      "loss": 0.2881,
+      "step": 2285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3002713867811947e-05,
+      "loss": 0.3239,
+      "step": 2286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2997026069033804e-05,
+      "loss": 0.3151,
+      "step": 2287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2991337204803747e-05,
+      "loss": 0.3285,
+      "step": 2288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298564727714418e-05,
+      "loss": 0.3066,
+      "step": 2289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979956288077885e-05,
+      "loss": 0.3196,
+      "step": 2290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297426423962803e-05,
+      "loss": 0.297,
+      "step": 2291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2968571133818157e-05,
+      "loss": 0.3184,
+      "step": 2292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962876972672182e-05,
+      "loss": 0.301,
+      "step": 2293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957181758214399e-05,
+      "loss": 0.3074,
+      "step": 2294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951485492469467e-05,
+      "loss": 0.3304,
+      "step": 2295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945788177462428e-05,
+      "loss": 0.3081,
+      "step": 2296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940089815218703e-05,
+      "loss": 0.3041,
+      "step": 2297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934390407764065e-05,
+      "loss": 0.3156,
+      "step": 2298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928689957124683e-05,
+      "loss": 0.3127,
+      "step": 2299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922988465327074e-05,
+      "loss": 0.3142,
+      "step": 2300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917285934398143e-05,
+      "loss": 0.3073,
+      "step": 2301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911582366365152e-05,
+      "loss": 0.3148,
+      "step": 2302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905877763255736e-05,
+      "loss": 0.3214,
+      "step": 2303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290017212709791e-05,
+      "loss": 0.3047,
+      "step": 2304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2894465459920031e-05,
+      "loss": 0.3079,
+      "step": 2305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888757763750845e-05,
+      "loss": 0.3231,
+      "step": 2306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288304904061945e-05,
+      "loss": 0.3331,
+      "step": 2307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287733929255532e-05,
+      "loss": 0.332,
+      "step": 2308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871628521588288e-05,
+      "loss": 0.2987,
+      "step": 2309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865916729748546e-05,
+      "loss": 0.3079,
+      "step": 2310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860203919066654e-05,
+      "loss": 0.3108,
+      "step": 2311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854490091573535e-05,
+      "loss": 0.3041,
+      "step": 2312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2848775249300473e-05,
+      "loss": 0.3167,
+      "step": 2313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284305939427911e-05,
+      "loss": 0.3185,
+      "step": 2314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837342528541454e-05,
+      "loss": 0.3036,
+      "step": 2315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831624654119864e-05,
+      "loss": 0.3132,
+      "step": 2316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825905773047063e-05,
+      "loss": 0.3019,
+      "step": 2317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2820185887356129e-05,
+      "loss": 0.3156,
+      "step": 2318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28144649990805e-05,
+      "loss": 0.3324,
+      "step": 2319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808743110253973e-05,
+      "loss": 0.3157,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2803020222910693e-05,
+      "loss": 0.3107,
+      "step": 2321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797296339085164e-05,
+      "loss": 0.3022,
+      "step": 2322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791571460812243e-05,
+      "loss": 0.3122,
+      "step": 2323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785845590127143e-05,
+      "loss": 0.3069,
+      "step": 2324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780118729065434e-05,
+      "loss": 0.3193,
+      "step": 2325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774390879663025e-05,
+      "loss": 0.3087,
+      "step": 2326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768662043956187e-05,
+      "loss": 0.3018,
+      "step": 2327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762932223981537e-05,
+      "loss": 0.3136,
+      "step": 2328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757201421776044e-05,
+      "loss": 0.3032,
+      "step": 2329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751469639377028e-05,
+      "loss": 0.3016,
+      "step": 2330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745736878822156e-05,
+      "loss": 0.3062,
+      "step": 2331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740003142149433e-05,
+      "loss": 0.3309,
+      "step": 2332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2734268431397233e-05,
+      "loss": 0.3121,
+      "step": 2333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2728532748604252e-05,
+      "loss": 0.2853,
+      "step": 2334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722796095809547e-05,
+      "loss": 0.2935,
+      "step": 2335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717058475052517e-05,
+      "loss": 0.3228,
+      "step": 2336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711319888372902e-05,
+      "loss": 0.3302,
+      "step": 2337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270558033781079e-05,
+      "loss": 0.3267,
+      "step": 2338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699839825406604e-05,
+      "loss": 0.304,
+      "step": 2339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694098353201115e-05,
+      "loss": 0.3015,
+      "step": 2340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688355923235439e-05,
+      "loss": 0.3112,
+      "step": 2341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682612537551022e-05,
+      "loss": 0.3176,
+      "step": 2342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676868198189653e-05,
+      "loss": 0.2935,
+      "step": 2343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671122907193469e-05,
+      "loss": 0.2797,
+      "step": 2344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665376666604931e-05,
+      "loss": 0.3224,
+      "step": 2345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2659629478466854e-05,
+      "loss": 0.3186,
+      "step": 2346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653881344822377e-05,
+      "loss": 0.3005,
+      "step": 2347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648132267714974e-05,
+      "loss": 0.3101,
+      "step": 2348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642382249188465e-05,
+      "loss": 0.299,
+      "step": 2349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636631291287e-05,
+      "loss": 0.3087,
+      "step": 2350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630879396055056e-05,
+      "loss": 0.3105,
+      "step": 2351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625126565537458e-05,
+      "loss": 0.3067,
+      "step": 2352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2619372801779346e-05,
+      "loss": 0.3032,
+      "step": 2353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613618106826206e-05,
+      "loss": 0.3136,
+      "step": 2354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607862482723849e-05,
+      "loss": 0.3027,
+      "step": 2355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602105931518413e-05,
+      "loss": 0.3021,
+      "step": 2356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596348455256377e-05,
+      "loss": 0.3188,
+      "step": 2357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590590055984538e-05,
+      "loss": 0.299,
+      "step": 2358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258483073575002e-05,
+      "loss": 0.319,
+      "step": 2359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2579070496600283e-05,
+      "loss": 0.3005,
+      "step": 2360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573309340583108e-05,
+      "loss": 0.3199,
+      "step": 2361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567547269746607e-05,
+      "loss": 0.3173,
+      "step": 2362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561784286139214e-05,
+      "loss": 0.314,
+      "step": 2363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255602039180968e-05,
+      "loss": 0.3005,
+      "step": 2364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550255588807092e-05,
+      "loss": 0.3113,
+      "step": 2365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254448987918086e-05,
+      "loss": 0.3208,
+      "step": 2366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2538723264980702e-05,
+      "loss": 0.3214,
+      "step": 2367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2532955748256671e-05,
+      "loss": 0.3139,
+      "step": 2368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2527187331059138e-05,
+      "loss": 0.3085,
+      "step": 2369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521418015438793e-05,
+      "loss": 0.3091,
+      "step": 2370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515647803446642e-05,
+      "loss": 0.3156,
+      "step": 2371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2509876697134016e-05,
+      "loss": 0.2977,
+      "step": 2372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2504104698552565e-05,
+      "loss": 0.3239,
+      "step": 2373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498331809754243e-05,
+      "loss": 0.3085,
+      "step": 2374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492558032791334e-05,
+      "loss": 0.3132,
+      "step": 2375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486783369716431e-05,
+      "loss": 0.3233,
+      "step": 2376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481007822582447e-05,
+      "loss": 0.2881,
+      "step": 2377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2475231393442611e-05,
+      "loss": 0.3221,
+      "step": 2378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469454084350454e-05,
+      "loss": 0.315,
+      "step": 2379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463675897359829e-05,
+      "loss": 0.3102,
+      "step": 2380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457896834524899e-05,
+      "loss": 0.3029,
+      "step": 2381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245211689790014e-05,
+      "loss": 0.3008,
+      "step": 2382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446336089540334e-05,
+      "loss": 0.3008,
+      "step": 2383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440554411500583e-05,
+      "loss": 0.3042,
+      "step": 2384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434771865836281e-05,
+      "loss": 0.3206,
+      "step": 2385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242898845460315e-05,
+      "loss": 0.2986,
+      "step": 2386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423204179857205e-05,
+      "loss": 0.308,
+      "step": 2387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417419043654773e-05,
+      "loss": 0.3112,
+      "step": 2388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411633048052494e-05,
+      "loss": 0.3134,
+      "step": 2389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24058461951073e-05,
+      "loss": 0.3102,
+      "step": 2390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400058486876434e-05,
+      "loss": 0.3153,
+      "step": 2391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239426992541745e-05,
+      "loss": 0.3035,
+      "step": 2392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388480512788197e-05,
+      "loss": 0.312,
+      "step": 2393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382690251046827e-05,
+      "loss": 0.2974,
+      "step": 2394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376899142251801e-05,
+      "loss": 0.3022,
+      "step": 2395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371107188461868e-05,
+      "loss": 0.323,
+      "step": 2396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365314391736089e-05,
+      "loss": 0.3039,
+      "step": 2397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359520754133823e-05,
+      "loss": 0.3143,
+      "step": 2398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353726277714723e-05,
+      "loss": 0.2955,
+      "step": 2399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347930964538746e-05,
+      "loss": 0.31,
+      "step": 2400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234213481666614e-05,
+      "loss": 0.3229,
+      "step": 2401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336337836157455e-05,
+      "loss": 0.3134,
+      "step": 2402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2330540025073537e-05,
+      "loss": 0.3066,
+      "step": 2403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232474138547552e-05,
+      "loss": 0.3235,
+      "step": 2404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318941919424846e-05,
+      "loss": 0.3008,
+      "step": 2405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2313141628983234e-05,
+      "loss": 0.3014,
+      "step": 2406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307340516212713e-05,
+      "loss": 0.3083,
+      "step": 2407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2301538583175592e-05,
+      "loss": 0.309,
+      "step": 2408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2295735831934477e-05,
+      "loss": 0.3045,
+      "step": 2409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289932264552264e-05,
+      "loss": 0.2957,
+      "step": 2410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284127883092138e-05,
+      "loss": 0.2789,
+      "step": 2411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278322689617577e-05,
+      "loss": 0.318,
+      "step": 2412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227251668619234e-05,
+      "loss": 0.3208,
+      "step": 2413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.226670987488048e-05,
+      "loss": 0.3258,
+      "step": 2414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2260902257746343e-05,
+      "loss": 0.3109,
+      "step": 2415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2255093836854549e-05,
+      "loss": 0.3006,
+      "step": 2416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2249284614270007e-05,
+      "loss": 0.3118,
+      "step": 2417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243474592057918e-05,
+      "loss": 0.3046,
+      "step": 2418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223766377228376e-05,
+      "loss": 0.3075,
+      "step": 2419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231852157013299e-05,
+      "loss": 0.3173,
+      "step": 2420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226039748312583e-05,
+      "loss": 0.3182,
+      "step": 2421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2220226548247938e-05,
+      "loss": 0.3177,
+      "step": 2422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214412558885975e-05,
+      "loss": 0.2982,
+      "step": 2423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208597782293586e-05,
+      "loss": 0.3088,
+      "step": 2424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2202782220537939e-05,
+      "loss": 0.2881,
+      "step": 2425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2196965875686489e-05,
+      "loss": 0.3083,
+      "step": 2426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2191148749806962e-05,
+      "loss": 0.307,
+      "step": 2427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185330844967363e-05,
+      "loss": 0.3156,
+      "step": 2428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2179512163235973e-05,
+      "loss": 0.3034,
+      "step": 2429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173692706681356e-05,
+      "loss": 0.3217,
+      "step": 2430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167872477372344e-05,
+      "loss": 0.3164,
+      "step": 2431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216205147737805e-05,
+      "loss": 0.3103,
+      "step": 2432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156229708767851e-05,
+      "loss": 0.3145,
+      "step": 2433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150407173611406e-05,
+      "loss": 0.3187,
+      "step": 2434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2144583873978643e-05,
+      "loss": 0.3091,
+      "step": 2435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2138759811939767e-05,
+      "loss": 0.3158,
+      "step": 2436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132934989565248e-05,
+      "loss": 0.3157,
+      "step": 2437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127109408925824e-05,
+      "loss": 0.3066,
+      "step": 2438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121283072092513e-05,
+      "loss": 0.3164,
+      "step": 2439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115455981136587e-05,
+      "loss": 0.3196,
+      "step": 2440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109628138129605e-05,
+      "loss": 0.2825,
+      "step": 2441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2103799545143378e-05,
+      "loss": 0.3093,
+      "step": 2442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097970204249989e-05,
+      "loss": 0.3078,
+      "step": 2443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092140117521785e-05,
+      "loss": 0.2996,
+      "step": 2444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208630928703138e-05,
+      "loss": 0.3096,
+      "step": 2445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2080477714851651e-05,
+      "loss": 0.3089,
+      "step": 2446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2074645403055745e-05,
+      "loss": 0.3024,
+      "step": 2447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2068812353717064e-05,
+      "loss": 0.3152,
+      "step": 2448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062978568909272e-05,
+      "loss": 0.3111,
+      "step": 2449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20571440507063e-05,
+      "loss": 0.3019,
+      "step": 2450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051308801182333e-05,
+      "loss": 0.3175,
+      "step": 2451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045472822411827e-05,
+      "loss": 0.3021,
+      "step": 2452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203963611646949e-05,
+      "loss": 0.303,
+      "step": 2453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203379868543028e-05,
+      "loss": 0.2964,
+      "step": 2454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2027960531369429e-05,
+      "loss": 0.3064,
+      "step": 2455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022121656362418e-05,
+      "loss": 0.2946,
+      "step": 2456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016282062484981e-05,
+      "loss": 0.2992,
+      "step": 2457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010441751813115e-05,
+      "loss": 0.3199,
+      "step": 2458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2004600726423067e-05,
+      "loss": 0.3105,
+      "step": 2459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1998758988391341e-05,
+      "loss": 0.3225,
+      "step": 2460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199291653979469e-05,
+      "loss": 0.3222,
+      "step": 2461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987073382710123e-05,
+      "loss": 0.3044,
+      "step": 2462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981229519214904e-05,
+      "loss": 0.2935,
+      "step": 2463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975384951386539e-05,
+      "loss": 0.317,
+      "step": 2464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969539681302794e-05,
+      "loss": 0.3002,
+      "step": 2465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1963693711041673e-05,
+      "loss": 0.3016,
+      "step": 2466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957847042681446e-05,
+      "loss": 0.3116,
+      "step": 2467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1951999678300617e-05,
+      "loss": 0.3039,
+      "step": 2468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946151619977941e-05,
+      "loss": 0.3119,
+      "step": 2469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194030286979242e-05,
+      "loss": 0.3033,
+      "step": 2470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1934453429823304e-05,
+      "loss": 0.3063,
+      "step": 2471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1928603302150087e-05,
+      "loss": 0.3058,
+      "step": 2472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922752488852512e-05,
+      "loss": 0.3021,
+      "step": 2473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191690099201055e-05,
+      "loss": 0.3169,
+      "step": 2474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911048813704433e-05,
+      "loss": 0.3088,
+      "step": 2475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905195956014631e-05,
+      "loss": 0.3242,
+      "step": 2476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899342421021844e-05,
+      "loss": 0.321,
+      "step": 2477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189348821080703e-05,
+      "loss": 0.296,
+      "step": 2478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887633327451373e-05,
+      "loss": 0.3134,
+      "step": 2479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881777773036307e-05,
+      "loss": 0.3076,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1875921549643495e-05,
+      "loss": 0.31,
+      "step": 2481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870064659354842e-05,
+      "loss": 0.3049,
+      "step": 2482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864207104252495e-05,
+      "loss": 0.3142,
+      "step": 2483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1858348886418832e-05,
+      "loss": 0.3163,
+      "step": 2484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852490007936461e-05,
+      "loss": 0.3153,
+      "step": 2485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846630470888236e-05,
+      "loss": 0.3076,
+      "step": 2486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184077027735724e-05,
+      "loss": 0.3193,
+      "step": 2487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834909429426794e-05,
+      "loss": 0.3026,
+      "step": 2488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1829047929180439e-05,
+      "loss": 0.3032,
+      "step": 2489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182318577870196e-05,
+      "loss": 0.312,
+      "step": 2490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181732298007537e-05,
+      "loss": 0.3229,
+      "step": 2491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811459535384913e-05,
+      "loss": 0.3155,
+      "step": 2492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805595446715055e-05,
+      "loss": 0.3035,
+      "step": 2493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799730716150507e-05,
+      "loss": 0.3099,
+      "step": 2494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1793865345776193e-05,
+      "loss": 0.3191,
+      "step": 2495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1787999337677269e-05,
+      "loss": 0.3179,
+      "step": 2496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1782132693939124e-05,
+      "loss": 0.3118,
+      "step": 2497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177626541664736e-05,
+      "loss": 0.3223,
+      "step": 2498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770397507887821e-05,
+      "loss": 0.3004,
+      "step": 2499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764528969746558e-05,
+      "loss": 0.3143,
+      "step": 2500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758659804309862e-05,
+      "loss": 0.3107,
+      "step": 2501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752790013664232e-05,
+      "loss": 0.3165,
+      "step": 2502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746919599896402e-05,
+      "loss": 0.3213,
+      "step": 2503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174104856509332e-05,
+      "loss": 0.2994,
+      "step": 2504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735176911342159e-05,
+      "loss": 0.3015,
+      "step": 2505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729304640730303e-05,
+      "loss": 0.2872,
+      "step": 2506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172343175534537e-05,
+      "loss": 0.3068,
+      "step": 2507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717558257275181e-05,
+      "loss": 0.3169,
+      "step": 2508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711684148607792e-05,
+      "loss": 0.2978,
+      "step": 2509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705809431431461e-05,
+      "loss": 0.2968,
+      "step": 2510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699934107834664e-05,
+      "loss": 0.3074,
+      "step": 2511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694058179906105e-05,
+      "loss": 0.3148,
+      "step": 2512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688181649734692e-05,
+      "loss": 0.3031,
+      "step": 2513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682304519409542e-05,
+      "loss": 0.3073,
+      "step": 2514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1676426791020002e-05,
+      "loss": 0.3143,
+      "step": 2515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670548466655615e-05,
+      "loss": 0.314,
+      "step": 2516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664669548406153e-05,
+      "loss": 0.2999,
+      "step": 2517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165879003836158e-05,
+      "loss": 0.3081,
+      "step": 2518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652909938612082e-05,
+      "loss": 0.3061,
+      "step": 2519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1647029251248056e-05,
+      "loss": 0.3085,
+      "step": 2520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641147978360103e-05,
+      "loss": 0.2815,
+      "step": 2521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1635266122039029e-05,
+      "loss": 0.2994,
+      "step": 2522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1629383684375854e-05,
+      "loss": 0.3108,
+      "step": 2523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623500667461806e-05,
+      "loss": 0.3124,
+      "step": 2524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1617617073388317e-05,
+      "loss": 0.3076,
+      "step": 2525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611732904247014e-05,
+      "loss": 0.3048,
+      "step": 2526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605848162129744e-05,
+      "loss": 0.3087,
+      "step": 2527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159996284912855e-05,
+      "loss": 0.302,
+      "step": 2528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594076967335677e-05,
+      "loss": 0.3011,
+      "step": 2529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588190518843572e-05,
+      "loss": 0.306,
+      "step": 2530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582303505744891e-05,
+      "loss": 0.3042,
+      "step": 2531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157641593013248e-05,
+      "loss": 0.3107,
+      "step": 2532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570527794099396e-05,
+      "loss": 0.3098,
+      "step": 2533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1564639099738881e-05,
+      "loss": 0.3173,
+      "step": 2534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1558749849144388e-05,
+      "loss": 0.3021,
+      "step": 2535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1552860044409568e-05,
+      "loss": 0.3096,
+      "step": 2536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154696968762826e-05,
+      "loss": 0.3068,
+      "step": 2537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15410787808945e-05,
+      "loss": 0.3093,
+      "step": 2538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153518732630253e-05,
+      "loss": 0.3234,
+      "step": 2539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529295325946778e-05,
+      "loss": 0.3196,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523402781921868e-05,
+      "loss": 0.3241,
+      "step": 2541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151750969632262e-05,
+      "loss": 0.3043,
+      "step": 2542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1511616071244038e-05,
+      "loss": 0.2951,
+      "step": 2543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505721908781327e-05,
+      "loss": 0.2985,
+      "step": 2544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1499827211029882e-05,
+      "loss": 0.2997,
+      "step": 2545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149393198008528e-05,
+      "loss": 0.3135,
+      "step": 2546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488036218043298e-05,
+      "loss": 0.298,
+      "step": 2547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14821399269999e-05,
+      "loss": 0.3273,
+      "step": 2548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476243109051231e-05,
+      "loss": 0.3223,
+      "step": 2549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470345766293627e-05,
+      "loss": 0.2948,
+      "step": 2550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464447900823612e-05,
+      "loss": 0.3017,
+      "step": 2551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458549514737902e-05,
+      "loss": 0.3027,
+      "step": 2552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452650610133383e-05,
+      "loss": 0.3047,
+      "step": 2553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1446751189107135e-05,
+      "loss": 0.3253,
+      "step": 2554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1440851253756423e-05,
+      "loss": 0.3132,
+      "step": 2555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434950806178688e-05,
+      "loss": 0.3091,
+      "step": 2556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429049848471562e-05,
+      "loss": 0.3017,
+      "step": 2557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.3118,
+      "step": 2558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417246411060548e-05,
+      "loss": 0.3159,
+      "step": 2559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411343935552813e-05,
+      "loss": 0.3083,
+      "step": 2560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405440958308001e-05,
+      "loss": 0.3074,
+      "step": 2561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399537481424636e-05,
+      "loss": 0.3085,
+      "step": 2562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393633507001427e-05,
+      "loss": 0.3098,
+      "step": 2563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1387729037137245e-05,
+      "loss": 0.303,
+      "step": 2564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381824073931156e-05,
+      "loss": 0.3085,
+      "step": 2565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137591861948239e-05,
+      "loss": 0.3157,
+      "step": 2566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137001267589035e-05,
+      "loss": 0.3151,
+      "step": 2567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1364106245254622e-05,
+      "loss": 0.3057,
+      "step": 2568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358199329674959e-05,
+      "loss": 0.3142,
+      "step": 2569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352291931251289e-05,
+      "loss": 0.2993,
+      "step": 2570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346384052083703e-05,
+      "loss": 0.3045,
+      "step": 2571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340475694272479e-05,
+      "loss": 0.3121,
+      "step": 2572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1334566859918053e-05,
+      "loss": 0.3173,
+      "step": 2573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328657551121035e-05,
+      "loss": 0.3173,
+      "step": 2574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1322747769982195e-05,
+      "loss": 0.3185,
+      "step": 2575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316837518602484e-05,
+      "loss": 0.2941,
+      "step": 2576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1310926799083015e-05,
+      "loss": 0.2755,
+      "step": 2577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130501561352507e-05,
+      "loss": 0.3077,
+      "step": 2578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1299103964030088e-05,
+      "loss": 0.3018,
+      "step": 2579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293191852699677e-05,
+      "loss": 0.3142,
+      "step": 2580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287279281635614e-05,
+      "loss": 0.3196,
+      "step": 2581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281366252939839e-05,
+      "loss": 0.3164,
+      "step": 2582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275452768714444e-05,
+      "loss": 0.3093,
+      "step": 2583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269538831061699e-05,
+      "loss": 0.2947,
+      "step": 2584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263624442084021e-05,
+      "loss": 0.3085,
+      "step": 2585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257709603883995e-05,
+      "loss": 0.2949,
+      "step": 2586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1251794318564365e-05,
+      "loss": 0.2976,
+      "step": 2587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245878588228032e-05,
+      "loss": 0.2983,
+      "step": 2588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239962414978058e-05,
+      "loss": 0.2936,
+      "step": 2589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234045800917659e-05,
+      "loss": 0.2783,
+      "step": 2590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228128748150206e-05,
+      "loss": 0.303,
+      "step": 2591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222211258779237e-05,
+      "loss": 0.3127,
+      "step": 2592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121629333490843e-05,
+      "loss": 0.3076,
+      "step": 2593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121037497864163e-05,
+      "loss": 0.3066,
+      "step": 2594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1204456192082833e-05,
+      "loss": 0.3132,
+      "step": 2595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198536977336176e-05,
+      "loss": 0.3146,
+      "step": 2596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192617336505966e-05,
+      "loss": 0.3091,
+      "step": 2597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1186697271696654e-05,
+      "loss": 0.2986,
+      "step": 2598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180776785012835e-05,
+      "loss": 0.3087,
+      "step": 2599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1174855878559265e-05,
+      "loss": 0.2985,
+      "step": 2600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116893455444084e-05,
+      "loss": 0.2921,
+      "step": 2601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163012814762616e-05,
+      "loss": 0.2973,
+      "step": 2602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157090661629781e-05,
+      "loss": 0.299,
+      "step": 2603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151168097147683e-05,
+      "loss": 0.3185,
+      "step": 2604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145245123421812e-05,
+      "loss": 0.3018,
+      "step": 2605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139321742557806e-05,
+      "loss": 0.3013,
+      "step": 2606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133397956661438e-05,
+      "loss": 0.3066,
+      "step": 2607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127473767838635e-05,
+      "loss": 0.3291,
+      "step": 2608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121549178195465e-05,
+      "loss": 0.3188,
+      "step": 2609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1115624189838139e-05,
+      "loss": 0.3129,
+      "step": 2610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1109698804873008e-05,
+      "loss": 0.3184,
+      "step": 2611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103773025406558e-05,
+      "loss": 0.3204,
+      "step": 2612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097846853545431e-05,
+      "loss": 0.3105,
+      "step": 2613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091920291396394e-05,
+      "loss": 0.3069,
+      "step": 2614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108599334106636e-05,
+      "loss": 0.2975,
+      "step": 2615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108006600466238e-05,
+      "loss": 0.311,
+      "step": 2616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074138284291633e-05,
+      "loss": 0.2993,
+      "step": 2617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1068210182061447e-05,
+      "loss": 0.2974,
+      "step": 2618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1062281700079282e-05,
+      "loss": 0.3001,
+      "step": 2619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1056352840452726e-05,
+      "loss": 0.2948,
+      "step": 2620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050423605289513e-05,
+      "loss": 0.3145,
+      "step": 2621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1044493996697496e-05,
+      "loss": 0.3147,
+      "step": 2622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038564016784679e-05,
+      "loss": 0.3129,
+      "step": 2623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1032633667659179e-05,
+      "loss": 0.3095,
+      "step": 2624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026702951429256e-05,
+      "loss": 0.2964,
+      "step": 2625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020771870203302e-05,
+      "loss": 0.3126,
+      "step": 2626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101484042608983e-05,
+      "loss": 0.3064,
+      "step": 2627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1008908621197488e-05,
+      "loss": 0.3051,
+      "step": 2628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002976457635048e-05,
+      "loss": 0.2922,
+      "step": 2629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997043937511415e-05,
+      "loss": 0.2996,
+      "step": 2630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0991111062935623e-05,
+      "loss": 0.3131,
+      "step": 2631
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0985177836016818e-05,
+      "loss": 0.3042,
+      "step": 2632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979244258864284e-05,
+      "loss": 0.3005,
+      "step": 2633
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0973310333587424e-05,
+      "loss": 0.2985,
+      "step": 2634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0967376062295772e-05,
+      "loss": 0.3093,
+      "step": 2635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961441447098972e-05,
+      "loss": 0.2995,
+      "step": 2636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955506490106806e-05,
+      "loss": 0.2972,
+      "step": 2637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949571193429162e-05,
+      "loss": 0.3167,
+      "step": 2638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943635559176063e-05,
+      "loss": 0.3028,
+      "step": 2639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0937699589457635e-05,
+      "loss": 0.3033,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931763286384143e-05,
+      "loss": 0.2969,
+      "step": 2641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925826652065959e-05,
+      "loss": 0.3048,
+      "step": 2642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0919889688613575e-05,
+      "loss": 0.2927,
+      "step": 2643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913952398137592e-05,
+      "loss": 0.3113,
+      "step": 2644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908014782748743e-05,
+      "loss": 0.3055,
+      "step": 2645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902076844557866e-05,
+      "loss": 0.3063,
+      "step": 2646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896138585675916e-05,
+      "loss": 0.3109,
+      "step": 2647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890200008213964e-05,
+      "loss": 0.3089,
+      "step": 2648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088426111428319e-05,
+      "loss": 0.3153,
+      "step": 2649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878321905994888e-05,
+      "loss": 0.3026,
+      "step": 2650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872382385460471e-05,
+      "loss": 0.326,
+      "step": 2651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866442554791447e-05,
+      "loss": 0.3001,
+      "step": 2652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860502416099454e-05,
+      "loss": 0.3066,
+      "step": 2653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085456197149622e-05,
+      "loss": 0.2995,
+      "step": 2654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848621223093598e-05,
+      "loss": 0.2889,
+      "step": 2655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084268017300354e-05,
+      "loss": 0.321,
+      "step": 2656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836738823338105e-05,
+      "loss": 0.3347,
+      "step": 2657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830797176209467e-05,
+      "loss": 0.2875,
+      "step": 2658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0824855233729894e-05,
+      "loss": 0.3165,
+      "step": 2659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818912998011766e-05,
+      "loss": 0.3165,
+      "step": 2660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812970471167568e-05,
+      "loss": 0.2969,
+      "step": 2661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0807027655309883e-05,
+      "loss": 0.3065,
+      "step": 2662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801084552551405e-05,
+      "loss": 0.311,
+      "step": 2663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0795141165004922e-05,
+      "loss": 0.304,
+      "step": 2664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0789197494783325e-05,
+      "loss": 0.3076,
+      "step": 2665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783253543999608e-05,
+      "loss": 0.3062,
+      "step": 2666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0777309314766866e-05,
+      "loss": 0.3224,
+      "step": 2667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771364809198291e-05,
+      "loss": 0.2863,
+      "step": 2668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076542002940717e-05,
+      "loss": 0.298,
+      "step": 2669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075947497750689e-05,
+      "loss": 0.3005,
+      "step": 2670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753529655610935e-05,
+      "loss": 0.3171,
+      "step": 2671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074758406583289e-05,
+      "loss": 0.296,
+      "step": 2672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0741638210286425e-05,
+      "loss": 0.3134,
+      "step": 2673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735692091085313e-05,
+      "loss": 0.3121,
+      "step": 2674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729745710343413e-05,
+      "loss": 0.3131,
+      "step": 2675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723799070174687e-05,
+      "loss": 0.3181,
+      "step": 2676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0717852172693182e-05,
+      "loss": 0.302,
+      "step": 2677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0711905020013036e-05,
+      "loss": 0.2952,
+      "step": 2678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0705957614248484e-05,
+      "loss": 0.3018,
+      "step": 2679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700009957513844e-05,
+      "loss": 0.3087,
+      "step": 2680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0694062051923524e-05,
+      "loss": 0.3102,
+      "step": 2681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688113899592025e-05,
+      "loss": 0.2997,
+      "step": 2682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682165502633935e-05,
+      "loss": 0.3005,
+      "step": 2683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676216863163926e-05,
+      "loss": 0.2964,
+      "step": 2684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0670267983296759e-05,
+      "loss": 0.3174,
+      "step": 2685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0664318865147273e-05,
+      "loss": 0.3057,
+      "step": 2686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658369510830405e-05,
+      "loss": 0.3065,
+      "step": 2687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652419922461165e-05,
+      "loss": 0.3065,
+      "step": 2688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646470102154649e-05,
+      "loss": 0.2956,
+      "step": 2689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640520052026042e-05,
+      "loss": 0.3024,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634569774190596e-05,
+      "loss": 0.3017,
+      "step": 2691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062861927076366e-05,
+      "loss": 0.2997,
+      "step": 2692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622668543860653e-05,
+      "loss": 0.3015,
+      "step": 2693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616717595597074e-05,
+      "loss": 0.308,
+      "step": 2694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610766428088512e-05,
+      "loss": 0.3004,
+      "step": 2695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060481504345062e-05,
+      "loss": 0.3053,
+      "step": 2696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059886344379913e-05,
+      "loss": 0.3103,
+      "step": 2697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592911631249855e-05,
+      "loss": 0.311,
+      "step": 2698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586959607918687e-05,
+      "loss": 0.3027,
+      "step": 2699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058100737592159e-05,
+      "loss": 0.32,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575054937374594e-05,
+      "loss": 0.2929,
+      "step": 2701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569102294393811e-05,
+      "loss": 0.2826,
+      "step": 2702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0563149449095421e-05,
+      "loss": 0.3099,
+      "step": 2703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557196403595689e-05,
+      "loss": 0.3041,
+      "step": 2704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0551243160010932e-05,
+      "loss": 0.3114,
+      "step": 2705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545289720457548e-05,
+      "loss": 0.3086,
+      "step": 2706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0539336087052001e-05,
+      "loss": 0.313,
+      "step": 2707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533382261910831e-05,
+      "loss": 0.319,
+      "step": 2708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527428247150638e-05,
+      "loss": 0.3003,
+      "step": 2709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0521474044888089e-05,
+      "loss": 0.3124,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051551965723993e-05,
+      "loss": 0.3094,
+      "step": 2711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509565086322957e-05,
+      "loss": 0.2898,
+      "step": 2712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0503610334254039e-05,
+      "loss": 0.2965,
+      "step": 2713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0497655403150108e-05,
+      "loss": 0.2996,
+      "step": 2714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0491700295128163e-05,
+      "loss": 0.3151,
+      "step": 2715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485745012305261e-05,
+      "loss": 0.2806,
+      "step": 2716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0479789556798528e-05,
+      "loss": 0.3041,
+      "step": 2717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473833930725137e-05,
+      "loss": 0.2969,
+      "step": 2718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467878136202339e-05,
+      "loss": 0.2944,
+      "step": 2719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461922175347434e-05,
+      "loss": 0.3058,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455966050277787e-05,
+      "loss": 0.2886,
+      "step": 2721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450009763110816e-05,
+      "loss": 0.3121,
+      "step": 2722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444053315964e-05,
+      "loss": 0.3044,
+      "step": 2723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438096710954873e-05,
+      "loss": 0.3084,
+      "step": 2724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043213995020103e-05,
+      "loss": 0.2987,
+      "step": 2725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0426183035820112e-05,
+      "loss": 0.2801,
+      "step": 2726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420225969929827e-05,
+      "loss": 0.3063,
+      "step": 2727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414268754647923e-05,
+      "loss": 0.2951,
+      "step": 2728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408311392092217e-05,
+      "loss": 0.294,
+      "step": 2729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402353884380558e-05,
+      "loss": 0.3042,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0396396233630865e-05,
+      "loss": 0.3192,
+      "step": 2731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390438441961106e-05,
+      "loss": 0.3051,
+      "step": 2732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384480511489286e-05,
+      "loss": 0.3145,
+      "step": 2733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037852244433347e-05,
+      "loss": 0.3021,
+      "step": 2734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372564242611767e-05,
+      "loss": 0.3246,
+      "step": 2735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036660590844234e-05,
+      "loss": 0.3014,
+      "step": 2736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0360647443943392e-05,
+      "loss": 0.3125,
+      "step": 2737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0354688851233178e-05,
+      "loss": 0.2965,
+      "step": 2738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348730132429993e-05,
+      "loss": 0.3002,
+      "step": 2739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0342771289652182e-05,
+      "loss": 0.3134,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336812325018133e-05,
+      "loss": 0.3129,
+      "step": 2741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033085324064627e-05,
+      "loss": 0.3308,
+      "step": 2742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0324894038655074e-05,
+      "loss": 0.3036,
+      "step": 2743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031893472116305e-05,
+      "loss": 0.3226,
+      "step": 2744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0312975290288763e-05,
+      "loss": 0.3196,
+      "step": 2745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307015748150801e-05,
+      "loss": 0.3011,
+      "step": 2746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301056096867806e-05,
+      "loss": 0.3081,
+      "step": 2747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0295096338558447e-05,
+      "loss": 0.2971,
+      "step": 2748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289136475341444e-05,
+      "loss": 0.2938,
+      "step": 2749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283176509335535e-05,
+      "loss": 0.3154,
+      "step": 2750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0277216442659512e-05,
+      "loss": 0.2942,
+      "step": 2751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02712562774322e-05,
+      "loss": 0.3113,
+      "step": 2752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265296015772452e-05,
+      "loss": 0.312,
+      "step": 2753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0259335659799163e-05,
+      "loss": 0.2889,
+      "step": 2754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253375211631252e-05,
+      "loss": 0.304,
+      "step": 2755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247414673387679e-05,
+      "loss": 0.3052,
+      "step": 2756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241454047187437e-05,
+      "loss": 0.3021,
+      "step": 2757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0235493335149543e-05,
+      "loss": 0.2999,
+      "step": 2758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229532539393051e-05,
+      "loss": 0.3025,
+      "step": 2759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0223571662037035e-05,
+      "loss": 0.3038,
+      "step": 2760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0217610705200615e-05,
+      "loss": 0.3086,
+      "step": 2761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211649671002924e-05,
+      "loss": 0.2922,
+      "step": 2762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205688561563125e-05,
+      "loss": 0.3024,
+      "step": 2763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199727379000419e-05,
+      "loss": 0.3073,
+      "step": 2764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193766125434018e-05,
+      "loss": 0.3046,
+      "step": 2765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187804802983168e-05,
+      "loss": 0.2928,
+      "step": 2766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0181843413767137e-05,
+      "loss": 0.298,
+      "step": 2767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175881959905215e-05,
+      "loss": 0.2987,
+      "step": 2768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0169920443516723e-05,
+      "loss": 0.3027,
+      "step": 2769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163958866720996e-05,
+      "loss": 0.3029,
+      "step": 2770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157997231637387e-05,
+      "loss": 0.2917,
+      "step": 2771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152035540385278e-05,
+      "loss": 0.2975,
+      "step": 2772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146073795084073e-05,
+      "loss": 0.3119,
+      "step": 2773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0140111997853187e-05,
+      "loss": 0.3104,
+      "step": 2774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134150150812062e-05,
+      "loss": 0.2972,
+      "step": 2775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0128188256080145e-05,
+      "loss": 0.2941,
+      "step": 2776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012222631577691e-05,
+      "loss": 0.2858,
+      "step": 2777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011626433202185e-05,
+      "loss": 0.3016,
+      "step": 2778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110302306934463e-05,
+      "loss": 0.2855,
+      "step": 2779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010434024263427e-05,
+      "loss": 0.3102,
+      "step": 2780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00983781412408e-05,
+      "loss": 0.2941,
+      "step": 2781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0092416004873605e-05,
+      "loss": 0.2988,
+      "step": 2782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0086453835652232e-05,
+      "loss": 0.296,
+      "step": 2783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080491635696258e-05,
+      "loss": 0.301,
+      "step": 2784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074529407125267e-05,
+      "loss": 0.3193,
+      "step": 2785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068567152058842e-05,
+      "loss": 0.298,
+      "step": 2786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062604872616581e-05,
+      "loss": 0.3119,
+      "step": 2787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056642570918099e-05,
+      "loss": 0.309,
+      "step": 2788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050680249083008e-05,
+      "loss": 0.3009,
+      "step": 2789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0044717909230941e-05,
+      "loss": 0.306,
+      "step": 2790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038755553481514e-05,
+      "loss": 0.301,
+      "step": 2791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032793183954371e-05,
+      "loss": 0.3078,
+      "step": 2792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026830802769149e-05,
+      "loss": 0.3041,
+      "step": 2793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020868412045496e-05,
+      "loss": 0.3087,
+      "step": 2794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0014906013903063e-05,
+      "loss": 0.3142,
+      "step": 2795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0008943610461493e-05,
+      "loss": 0.3104,
+      "step": 2796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000298120384044e-05,
+      "loss": 0.3123,
+      "step": 2797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997018796159563e-06,
+      "loss": 0.3071,
+      "step": 2798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991056389538509e-06,
+      "loss": 0.3085,
+      "step": 2799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98509398609694e-06,
+      "loss": 0.2824,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979131587954507e-06,
+      "loss": 0.2935,
+      "step": 2801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973169197230853e-06,
+      "loss": 0.2988,
+      "step": 2802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967206816045634e-06,
+      "loss": 0.3236,
+      "step": 2803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961244446518491e-06,
+      "loss": 0.3114,
+      "step": 2804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.955282090769062e-06,
+      "loss": 0.3118,
+      "step": 2805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949319750916995e-06,
+      "loss": 0.288,
+      "step": 2806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943357429081903e-06,
+      "loss": 0.3068,
+      "step": 2807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937395127383422e-06,
+      "loss": 0.2922,
+      "step": 2808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931432847941165e-06,
+      "loss": 0.3102,
+      "step": 2809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925470592874734e-06,
+      "loss": 0.3101,
+      "step": 2810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919508364303743e-06,
+      "loss": 0.3041,
+      "step": 2811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.913546164347773e-06,
+      "loss": 0.3038,
+      "step": 2812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9075839951264e-06,
+      "loss": 0.2919,
+      "step": 2813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901621858759203e-06,
+      "loss": 0.304,
+      "step": 2814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895659757365731e-06,
+      "loss": 0.321,
+      "step": 2815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88969769306554e-06,
+      "loss": 0.3015,
+      "step": 2816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883735667978155e-06,
+      "loss": 0.3016,
+      "step": 2817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877773684223091e-06,
+      "loss": 0.3221,
+      "step": 2818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871811743919858e-06,
+      "loss": 0.3107,
+      "step": 2819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865849849187943e-06,
+      "loss": 0.2994,
+      "step": 2820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859888002146814e-06,
+      "loss": 0.3096,
+      "step": 2821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85392620491593e-06,
+      "loss": 0.2956,
+      "step": 2822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847964459614722e-06,
+      "loss": 0.2953,
+      "step": 2823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842002768362618e-06,
+      "loss": 0.2965,
+      "step": 2824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83604113327901e-06,
+      "loss": 0.3167,
+      "step": 2825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830079556483279e-06,
+      "loss": 0.3115,
+      "step": 2826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824118040094786e-06,
+      "loss": 0.3005,
+      "step": 2827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818156586232865e-06,
+      "loss": 0.2964,
+      "step": 2828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812195197016833e-06,
+      "loss": 0.305,
+      "step": 2829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806233874565985e-06,
+      "loss": 0.316,
+      "step": 2830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800272620999581e-06,
+      "loss": 0.2839,
+      "step": 2831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794311438436878e-06,
+      "loss": 0.3049,
+      "step": 2832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788350328997083e-06,
+      "loss": 0.2921,
+      "step": 2833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.782389294799388e-06,
+      "loss": 0.292,
+      "step": 2834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776428337962968e-06,
+      "loss": 0.2839,
+      "step": 2835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77046746060695e-06,
+      "loss": 0.2939,
+      "step": 2836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76450666485046e-06,
+      "loss": 0.2958,
+      "step": 2837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758545952812568e-06,
+      "loss": 0.2885,
+      "step": 2838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752585326612321e-06,
+      "loss": 0.2969,
+      "step": 2839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74662478836875e-06,
+      "loss": 0.2957,
+      "step": 2840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740664340200842e-06,
+      "loss": 0.3054,
+      "step": 2841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73470398422755e-06,
+      "loss": 0.2964,
+      "step": 2842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728743722567803e-06,
+      "loss": 0.304,
+      "step": 2843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.722783557340488e-06,
+      "loss": 0.291,
+      "step": 2844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716823490664468e-06,
+      "loss": 0.3092,
+      "step": 2845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710863524658563e-06,
+      "loss": 0.3119,
+      "step": 2846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704903661441555e-06,
+      "loss": 0.3146,
+      "step": 2847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698943903132197e-06,
+      "loss": 0.3018,
+      "step": 2848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692984251849202e-06,
+      "loss": 0.2845,
+      "step": 2849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.687024709711239e-06,
+      "loss": 0.2874,
+      "step": 2850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.681065278836952e-06,
+      "loss": 0.2969,
+      "step": 2851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.67510596134493e-06,
+      "loss": 0.296,
+      "step": 2852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669146759353732e-06,
+      "loss": 0.3018,
+      "step": 2853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663187674981872e-06,
+      "loss": 0.2969,
+      "step": 2854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657228710347818e-06,
+      "loss": 0.2966,
+      "step": 2855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651269867570009e-06,
+      "loss": 0.2805,
+      "step": 2856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.645311148766825e-06,
+      "loss": 0.2932,
+      "step": 2857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63935255605661e-06,
+      "loss": 0.2896,
+      "step": 2858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633394091557663e-06,
+      "loss": 0.2986,
+      "step": 2859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627435757388235e-06,
+      "loss": 0.3134,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621477555666534e-06,
+      "loss": 0.295,
+      "step": 2861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615519488510718e-06,
+      "loss": 0.3015,
+      "step": 2862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609561558038897e-06,
+      "loss": 0.3163,
+      "step": 2863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603603766369136e-06,
+      "loss": 0.3036,
+      "step": 2864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597646115619447e-06,
+      "loss": 0.2953,
+      "step": 2865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591688607907788e-06,
+      "loss": 0.2945,
+      "step": 2866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585731245352079e-06,
+      "loss": 0.2983,
+      "step": 2867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579774030070176e-06,
+      "loss": 0.2955,
+      "step": 2868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573816964179891e-06,
+      "loss": 0.2773,
+      "step": 2869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567860049798975e-06,
+      "loss": 0.3005,
+      "step": 2870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561903289045128e-06,
+      "loss": 0.29,
+      "step": 2871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555946684036004e-06,
+      "loss": 0.3271,
+      "step": 2872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549990236889189e-06,
+      "loss": 0.2961,
+      "step": 2873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544033949722217e-06,
+      "loss": 0.3022,
+      "step": 2874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53807782465257e-06,
+      "loss": 0.2936,
+      "step": 2875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532121863797663e-06,
+      "loss": 0.2997,
+      "step": 2876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526166069274866e-06,
+      "loss": 0.3087,
+      "step": 2877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520210443201477e-06,
+      "loss": 0.2926,
+      "step": 2878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51425498769474e-06,
+      "loss": 0.2861,
+      "step": 2879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50829970487184e-06,
+      "loss": 0.321,
+      "step": 2880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.502344596849893e-06,
+      "loss": 0.3029,
+      "step": 2881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496389665745964e-06,
+      "loss": 0.295,
+      "step": 2882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490434913677048e-06,
+      "loss": 0.3029,
+      "step": 2883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484480342760074e-06,
+      "loss": 0.3102,
+      "step": 2884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478525955111913e-06,
+      "loss": 0.2985,
+      "step": 2885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472571752849367e-06,
+      "loss": 0.2974,
+      "step": 2886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46661773808917e-06,
+      "loss": 0.3038,
+      "step": 2887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460663912948002e-06,
+      "loss": 0.2922,
+      "step": 2888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454710279542455e-06,
+      "loss": 0.3109,
+      "step": 2889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448756839989072e-06,
+      "loss": 0.3071,
+      "step": 2890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442803596404315e-06,
+      "loss": 0.3035,
+      "step": 2891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436850550904577e-06,
+      "loss": 0.3081,
+      "step": 2892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430897705606192e-06,
+      "loss": 0.3046,
+      "step": 2893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424945062625411e-06,
+      "loss": 0.3033,
+      "step": 2894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418992624078414e-06,
+      "loss": 0.2961,
+      "step": 2895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413040392081315e-06,
+      "loss": 0.3108,
+      "step": 2896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407088368750143e-06,
+      "loss": 0.3115,
+      "step": 2897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401136556200874e-06,
+      "loss": 0.3141,
+      "step": 2898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395184956549385e-06,
+      "loss": 0.3056,
+      "step": 2899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389233571911491e-06,
+      "loss": 0.2946,
+      "step": 2900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383282404402927e-06,
+      "loss": 0.3009,
+      "step": 2901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377331456139352e-06,
+      "loss": 0.2917,
+      "step": 2902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.371380729236343e-06,
+      "loss": 0.2934,
+      "step": 2903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365430225809406e-06,
+      "loss": 0.3068,
+      "step": 2904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359479947973962e-06,
+      "loss": 0.3,
+      "step": 2905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.353529897845353e-06,
+      "loss": 0.3053,
+      "step": 2906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34758007753884e-06,
+      "loss": 0.2856,
+      "step": 2907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.341630489169597e-06,
+      "loss": 0.2919,
+      "step": 2908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335681134852728e-06,
+      "loss": 0.3014,
+      "step": 2909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.329732016703246e-06,
+      "loss": 0.299,
+      "step": 2910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323783136836076e-06,
+      "loss": 0.2897,
+      "step": 2911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317834497366068e-06,
+      "loss": 0.3032,
+      "step": 2912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311886100407975e-06,
+      "loss": 0.302,
+      "step": 2913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30593794807648e-06,
+      "loss": 0.3031,
+      "step": 2914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299990042486163e-06,
+      "loss": 0.2936,
+      "step": 2915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.294042385751521e-06,
+      "loss": 0.2946,
+      "step": 2916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.288094979986968e-06,
+      "loss": 0.2957,
+      "step": 2917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.282147827306823e-06,
+      "loss": 0.3143,
+      "step": 2918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276200929825315e-06,
+      "loss": 0.3057,
+      "step": 2919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27025428965659e-06,
+      "loss": 0.2957,
+      "step": 2920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26430790891469e-06,
+      "loss": 0.3057,
+      "step": 2921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25836178971358e-06,
+      "loss": 0.2838,
+      "step": 2922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252415934167115e-06,
+      "loss": 0.2766,
+      "step": 2923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246470344389065e-06,
+      "loss": 0.2946,
+      "step": 2924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240525022493113e-06,
+      "loss": 0.3058,
+      "step": 2925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234579970592835e-06,
+      "loss": 0.2991,
+      "step": 2926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228635190801712e-06,
+      "loss": 0.3,
+      "step": 2927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.222690685233137e-06,
+      "loss": 0.2852,
+      "step": 2928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216746456000392e-06,
+      "loss": 0.293,
+      "step": 2929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210802505216678e-06,
+      "loss": 0.3093,
+      "step": 2930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204858834995083e-06,
+      "loss": 0.3104,
+      "step": 2931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198915447448599e-06,
+      "loss": 0.2981,
+      "step": 2932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19297234469012e-06,
+      "loss": 0.3005,
+      "step": 2933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187029528832433e-06,
+      "loss": 0.3011,
+      "step": 2934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181087001988236e-06,
+      "loss": 0.3066,
+      "step": 2935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175144766270109e-06,
+      "loss": 0.3226,
+      "step": 2936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169202823790537e-06,
+      "loss": 0.3007,
+      "step": 2937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163261176661897e-06,
+      "loss": 0.3048,
+      "step": 2938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.157319826996466e-06,
+      "loss": 0.3119,
+      "step": 2939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.151378776906403e-06,
+      "loss": 0.3116,
+      "step": 2940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145438028503783e-06,
+      "loss": 0.3154,
+      "step": 2941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139497583900549e-06,
+      "loss": 0.2952,
+      "step": 2942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133557445208555e-06,
+      "loss": 0.298,
+      "step": 2943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127617614539534e-06,
+      "loss": 0.3104,
+      "step": 2944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121678094005112e-06,
+      "loss": 0.296,
+      "step": 2945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115738885716812e-06,
+      "loss": 0.2953,
+      "step": 2946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109799991786039e-06,
+      "loss": 0.2819,
+      "step": 2947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103861414324085e-06,
+      "loss": 0.2919,
+      "step": 2948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097923155442138e-06,
+      "loss": 0.3068,
+      "step": 2949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091985217251258e-06,
+      "loss": 0.2818,
+      "step": 2950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08604760186241e-06,
+      "loss": 0.2954,
+      "step": 2951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.080110311386432e-06,
+      "loss": 0.3046,
+      "step": 2952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074173347934043e-06,
+      "loss": 0.2935,
+      "step": 2953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068236713615859e-06,
+      "loss": 0.3136,
+      "step": 2954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062300410542368e-06,
+      "loss": 0.3014,
+      "step": 2955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056364440823942e-06,
+      "loss": 0.3082,
+      "step": 2956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050428806570841e-06,
+      "loss": 0.2962,
+      "step": 2957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044493509893198e-06,
+      "loss": 0.3049,
+      "step": 2958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03855855290103e-06,
+      "loss": 0.3079,
+      "step": 2959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032623937704233e-06,
+      "loss": 0.295,
+      "step": 2960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.026689666412576e-06,
+      "loss": 0.3128,
+      "step": 2961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02075574113572e-06,
+      "loss": 0.2926,
+      "step": 2962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014822163983187e-06,
+      "loss": 0.3021,
+      "step": 2963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008888937064382e-06,
+      "loss": 0.3086,
+      "step": 2964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002956062488586e-06,
+      "loss": 0.297,
+      "step": 2965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997023542364952e-06,
+      "loss": 0.2816,
+      "step": 2966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991091378802515e-06,
+      "loss": 0.3018,
+      "step": 2967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985159573910173e-06,
+      "loss": 0.3076,
+      "step": 2968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979228129796702e-06,
+      "loss": 0.2902,
+      "step": 2969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973297048570747e-06,
+      "loss": 0.297,
+      "step": 2970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.967366332340826e-06,
+      "loss": 0.2865,
+      "step": 2971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961435983215326e-06,
+      "loss": 0.3174,
+      "step": 2972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.955506003302506e-06,
+      "loss": 0.3074,
+      "step": 2973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949576394710492e-06,
+      "loss": 0.2916,
+      "step": 2974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.943647159547275e-06,
+      "loss": 0.308,
+      "step": 2975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.937718299920723e-06,
+      "loss": 0.2759,
+      "step": 2976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.931789817938553e-06,
+      "loss": 0.2974,
+      "step": 2977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925861715708368e-06,
+      "loss": 0.302,
+      "step": 2978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919933995337624e-06,
+      "loss": 0.2966,
+      "step": 2979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.914006658933641e-06,
+      "loss": 0.2997,
+      "step": 2980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908079708603608e-06,
+      "loss": 0.3088,
+      "step": 2981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902153146454569e-06,
+      "loss": 0.2923,
+      "step": 2982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896226974593443e-06,
+      "loss": 0.2923,
+      "step": 2983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.890301195126999e-06,
+      "loss": 0.2967,
+      "step": 2984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.884375810161863e-06,
+      "loss": 0.3209,
+      "step": 2985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.878450821804538e-06,
+      "loss": 0.3154,
+      "step": 2986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87252623216137e-06,
+      "loss": 0.3136,
+      "step": 2987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866602043338566e-06,
+      "loss": 0.2982,
+      "step": 2988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8606782574422e-06,
+      "loss": 0.3051,
+      "step": 2989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85475487657819e-06,
+      "loss": 0.2914,
+      "step": 2990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848831902852319e-06,
+      "loss": 0.303,
+      "step": 2991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842909338370224e-06,
+      "loss": 0.2979,
+      "step": 2992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836987185237388e-06,
+      "loss": 0.3041,
+      "step": 2993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831065445559161e-06,
+      "loss": 0.2939,
+      "step": 2994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825144121440738e-06,
+      "loss": 0.3025,
+      "step": 2995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.819223214987169e-06,
+      "loss": 0.3051,
+      "step": 2996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.813302728303351e-06,
+      "loss": 0.2999,
+      "step": 2997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807382663494034e-06,
+      "loss": 0.3119,
+      "step": 2998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801463022663826e-06,
+      "loss": 0.2787,
+      "step": 2999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795543807917172e-06,
+      "loss": 0.3033,
+      "step": 3000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789625021358371e-06,
+      "loss": 0.3141,
+      "step": 3001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.783706665091573e-06,
+      "loss": 0.2855,
+      "step": 3002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777788741220764e-06,
+      "loss": 0.3049,
+      "step": 3003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771871251849796e-06,
+      "loss": 0.3056,
+      "step": 3004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765954199082346e-06,
+      "loss": 0.3109,
+      "step": 3005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760037585021945e-06,
+      "loss": 0.2953,
+      "step": 3006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754121411771973e-06,
+      "loss": 0.3052,
+      "step": 3007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74820568143564e-06,
+      "loss": 0.3,
+      "step": 3008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742290396116007e-06,
+      "loss": 0.279,
+      "step": 3009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736375557915982e-06,
+      "loss": 0.2992,
+      "step": 3010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730461168938304e-06,
+      "loss": 0.2899,
+      "step": 3011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724547231285558e-06,
+      "loss": 0.2953,
+      "step": 3012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718633747060166e-06,
+      "loss": 0.2892,
+      "step": 3013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712720718364387e-06,
+      "loss": 0.2879,
+      "step": 3014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706808147300327e-06,
+      "loss": 0.2971,
+      "step": 3015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.700896035969917e-06,
+      "loss": 0.3054,
+      "step": 3016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694984386474934e-06,
+      "loss": 0.3057,
+      "step": 3017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689073200916988e-06,
+      "loss": 0.2856,
+      "step": 3018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683162481397516e-06,
+      "loss": 0.2936,
+      "step": 3019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.677252230017809e-06,
+      "loss": 0.3094,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671342448878972e-06,
+      "loss": 0.3029,
+      "step": 3021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66543314008195e-06,
+      "loss": 0.3002,
+      "step": 3022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659524305727524e-06,
+      "loss": 0.3053,
+      "step": 3023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6536159479163e-06,
+      "loss": 0.2906,
+      "step": 3024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647708068748715e-06,
+      "loss": 0.2888,
+      "step": 3025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641800670325043e-06,
+      "loss": 0.3042,
+      "step": 3026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63589375474538e-06,
+      "loss": 0.2938,
+      "step": 3027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.629987324109654e-06,
+      "loss": 0.3004,
+      "step": 3028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624081380517616e-06,
+      "loss": 0.2949,
+      "step": 3029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618175926068846e-06,
+      "loss": 0.314,
+      "step": 3030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.612270962862756e-06,
+      "loss": 0.2934,
+      "step": 3031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60636649299858e-06,
+      "loss": 0.296,
+      "step": 3032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600462518575366e-06,
+      "loss": 0.2933,
+      "step": 3033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594559041692004e-06,
+      "loss": 0.2975,
+      "step": 3034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588656064447187e-06,
+      "loss": 0.2916,
+      "step": 3035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582753588939457e-06,
+      "loss": 0.3012,
+      "step": 3036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.3309,
+      "step": 3037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57095015152844e-06,
+      "loss": 0.2932,
+      "step": 3038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565049193821314e-06,
+      "loss": 0.2975,
+      "step": 3039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559148746243582e-06,
+      "loss": 0.2994,
+      "step": 3040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553248810892866e-06,
+      "loss": 0.2952,
+      "step": 3041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547349389866622e-06,
+      "loss": 0.2848,
+      "step": 3042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541450485262102e-06,
+      "loss": 0.3031,
+      "step": 3043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53555209917639e-06,
+      "loss": 0.3004,
+      "step": 3044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529654233706378e-06,
+      "loss": 0.2972,
+      "step": 3045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523756890948772e-06,
+      "loss": 0.291,
+      "step": 3046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517860073000104e-06,
+      "loss": 0.2987,
+      "step": 3047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511963781956703e-06,
+      "loss": 0.2801,
+      "step": 3048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506068019914723e-06,
+      "loss": 0.2775,
+      "step": 3049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500172788970125e-06,
+      "loss": 0.3018,
+      "step": 3050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494278091218675e-06,
+      "loss": 0.2932,
+      "step": 3051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.488383928755965e-06,
+      "loss": 0.2831,
+      "step": 3052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482490303677385e-06,
+      "loss": 0.2946,
+      "step": 3053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476597218078134e-06,
+      "loss": 0.2936,
+      "step": 3054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470704674053225e-06,
+      "loss": 0.3052,
+      "step": 3055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464812673697468e-06,
+      "loss": 0.304,
+      "step": 3056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458921219105501e-06,
+      "loss": 0.3055,
+      "step": 3057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453030312371747e-06,
+      "loss": 0.2825,
+      "step": 3058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447139955590436e-06,
+      "loss": 0.2985,
+      "step": 3059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441250150855614e-06,
+      "loss": 0.2858,
+      "step": 3060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.435360900261124e-06,
+      "loss": 0.3192,
+      "step": 3061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429472205900609e-06,
+      "loss": 0.2955,
+      "step": 3062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.423584069867522e-06,
+      "loss": 0.2923,
+      "step": 3063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417696494255112e-06,
+      "loss": 0.2969,
+      "step": 3064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41180948115643e-06,
+      "loss": 0.3089,
+      "step": 3065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405923032664328e-06,
+      "loss": 0.2982,
+      "step": 3066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.400037150871452e-06,
+      "loss": 0.3007,
+      "step": 3067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394151837870257e-06,
+      "loss": 0.2947,
+      "step": 3068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.388267095752989e-06,
+      "loss": 0.2993,
+      "step": 3069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382382926611686e-06,
+      "loss": 0.2961,
+      "step": 3070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.376499332538195e-06,
+      "loss": 0.3008,
+      "step": 3071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370616315624145e-06,
+      "loss": 0.3044,
+      "step": 3072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364733877960975e-06,
+      "loss": 0.2957,
+      "step": 3073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358852021639902e-06,
+      "loss": 0.2735,
+      "step": 3074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352970748751947e-06,
+      "loss": 0.3008,
+      "step": 3075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347090061387921e-06,
+      "loss": 0.2999,
+      "step": 3076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.341209961638426e-06,
+      "loss": 0.3052,
+      "step": 3077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33533045159385e-06,
+      "loss": 0.3008,
+      "step": 3078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329451533344387e-06,
+      "loss": 0.2957,
+      "step": 3079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32357320898e-06,
+      "loss": 0.3013,
+      "step": 3080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317695480590461e-06,
+      "loss": 0.3098,
+      "step": 3081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311818350265315e-06,
+      "loss": 0.3081,
+      "step": 3082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305941820093896e-06,
+      "loss": 0.3131,
+      "step": 3083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300065892165337e-06,
+      "loss": 0.314,
+      "step": 3084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294190568568544e-06,
+      "loss": 0.2887,
+      "step": 3085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28831585139221e-06,
+      "loss": 0.2969,
+      "step": 3086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28244174272482e-06,
+      "loss": 0.3024,
+      "step": 3087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276568244654632e-06,
+      "loss": 0.3014,
+      "step": 3088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270695359269699e-06,
+      "loss": 0.2914,
+      "step": 3089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264823088657846e-06,
+      "loss": 0.3032,
+      "step": 3090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258951434906683e-06,
+      "loss": 0.2924,
+      "step": 3091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253080400103601e-06,
+      "loss": 0.2928,
+      "step": 3092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247209986335771e-06,
+      "loss": 0.2921,
+      "step": 3093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24134019569014e-06,
+      "loss": 0.2952,
+      "step": 3094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235471030253444e-06,
+      "loss": 0.2909,
+      "step": 3095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229602492112182e-06,
+      "loss": 0.2842,
+      "step": 3096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.223734583352643e-06,
+      "loss": 0.288,
+      "step": 3097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.217867306060883e-06,
+      "loss": 0.28,
+      "step": 3098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.212000662322733e-06,
+      "loss": 0.298,
+      "step": 3099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20613465422381e-06,
+      "loss": 0.2995,
+      "step": 3100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200269283849496e-06,
+      "loss": 0.2955,
+      "step": 3101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194404553284947e-06,
+      "loss": 0.2782,
+      "step": 3102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188540464615094e-06,
+      "loss": 0.2947,
+      "step": 3103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182677019924632e-06,
+      "loss": 0.2994,
+      "step": 3104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176814221298045e-06,
+      "loss": 0.3075,
+      "step": 3105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.170952070819566e-06,
+      "loss": 0.2828,
+      "step": 3106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165090570573211e-06,
+      "loss": 0.2881,
+      "step": 3107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159229722642763e-06,
+      "loss": 0.2865,
+      "step": 3108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153369529111764e-06,
+      "loss": 0.3081,
+      "step": 3109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147509992063542e-06,
+      "loss": 0.3049,
+      "step": 3110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.141651113581173e-06,
+      "loss": 0.2942,
+      "step": 3111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135792895747507e-06,
+      "loss": 0.289,
+      "step": 3112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129935340645161e-06,
+      "loss": 0.311,
+      "step": 3113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12407845035651e-06,
+      "loss": 0.3072,
+      "step": 3114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118222226963696e-06,
+      "loss": 0.2941,
+      "step": 3115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11236667254863e-06,
+      "loss": 0.2993,
+      "step": 3116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106511789192972e-06,
+      "loss": 0.2955,
+      "step": 3117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100657578978159e-06,
+      "loss": 0.3035,
+      "step": 3118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094804043985376e-06,
+      "loss": 0.2949,
+      "step": 3119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088951186295567e-06,
+      "loss": 0.2881,
+      "step": 3120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083099007989451e-06,
+      "loss": 0.2961,
+      "step": 3121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077247511147495e-06,
+      "loss": 0.3049,
+      "step": 3122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071396697849914e-06,
+      "loss": 0.2865,
+      "step": 3123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065546570176698e-06,
+      "loss": 0.2869,
+      "step": 3124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05969713020758e-06,
+      "loss": 0.3118,
+      "step": 3125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053848380022062e-06,
+      "loss": 0.3019,
+      "step": 3126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048000321699388e-06,
+      "loss": 0.2981,
+      "step": 3127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042152957318556e-06,
+      "loss": 0.3084,
+      "step": 3128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036306288958328e-06,
+      "loss": 0.2839,
+      "step": 3129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030460318697213e-06,
+      "loss": 0.2907,
+      "step": 3130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.024615048613463e-06,
+      "loss": 0.2892,
+      "step": 3131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0187704807851e-06,
+      "loss": 0.2884,
+      "step": 3132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012926617289879e-06,
+      "loss": 0.3008,
+      "step": 3133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007083460205313e-06,
+      "loss": 0.2885,
+      "step": 3134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.001241011608664e-06,
+      "loss": 0.2927,
+      "step": 3135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995399273576934e-06,
+      "loss": 0.3043,
+      "step": 3136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989558248186887e-06,
+      "loss": 0.2889,
+      "step": 3137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983717937515024e-06,
+      "loss": 0.3077,
+      "step": 3138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977878343637586e-06,
+      "loss": 0.3054,
+      "step": 3139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972039468630574e-06,
+      "loss": 0.2914,
+      "step": 3140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96620131456972e-06,
+      "loss": 0.2894,
+      "step": 3141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960363883530512e-06,
+      "loss": 0.2963,
+      "step": 3142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954527177588174e-06,
+      "loss": 0.2825,
+      "step": 3143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948691198817667e-06,
+      "loss": 0.2828,
+      "step": 3144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942855949293704e-06,
+      "loss": 0.3015,
+      "step": 3145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937021431090733e-06,
+      "loss": 0.2907,
+      "step": 3146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931187646282938e-06,
+      "loss": 0.2874,
+      "step": 3147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.925354596944257e-06,
+      "loss": 0.2972,
+      "step": 3148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919522285148349e-06,
+      "loss": 0.308,
+      "step": 3149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913690712968624e-06,
+      "loss": 0.3091,
+      "step": 3150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90785988247822e-06,
+      "loss": 0.299,
+      "step": 3151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902029795750014e-06,
+      "loss": 0.2915,
+      "step": 3152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896200454856624e-06,
+      "loss": 0.2977,
+      "step": 3153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890371861870397e-06,
+      "loss": 0.2932,
+      "step": 3154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884544018863414e-06,
+      "loss": 0.2961,
+      "step": 3155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878716927907492e-06,
+      "loss": 0.2872,
+      "step": 3156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872890591074176e-06,
+      "loss": 0.298,
+      "step": 3157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.867065010434755e-06,
+      "loss": 0.2911,
+      "step": 3158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.861240188060237e-06,
+      "loss": 0.2881,
+      "step": 3159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.855416126021358e-06,
+      "loss": 0.284,
+      "step": 3160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8495928263886e-06,
+      "loss": 0.2873,
+      "step": 3161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84377029123215e-06,
+      "loss": 0.2839,
+      "step": 3162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837948522621954e-06,
+      "loss": 0.297,
+      "step": 3163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832127522627658e-06,
+      "loss": 0.3018,
+      "step": 3164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826307293318645e-06,
+      "loss": 0.2927,
+      "step": 3165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82048783676403e-06,
+      "loss": 0.3105,
+      "step": 3166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814669155032642e-06,
+      "loss": 0.3024,
+      "step": 3167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80885125019304e-06,
+      "loss": 0.301,
+      "step": 3168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.803034124313513e-06,
+      "loss": 0.2944,
+      "step": 3169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.797217779462061e-06,
+      "loss": 0.3014,
+      "step": 3170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.791402217706419e-06,
+      "loss": 0.2897,
+      "step": 3171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78558744111403e-06,
+      "loss": 0.2993,
+      "step": 3172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779773451752064e-06,
+      "loss": 0.303,
+      "step": 3173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77396025168742e-06,
+      "loss": 0.2877,
+      "step": 3174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768147842986703e-06,
+      "loss": 0.2869,
+      "step": 3175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.762336227716241e-06,
+      "loss": 0.2866,
+      "step": 3176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756525407942087e-06,
+      "loss": 0.3047,
+      "step": 3177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750715385729993e-06,
+      "loss": 0.2759,
+      "step": 3178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.744906163145453e-06,
+      "loss": 0.2907,
+      "step": 3179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73909774225366e-06,
+      "loss": 0.2953,
+      "step": 3180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73329012511952e-06,
+      "loss": 0.2932,
+      "step": 3181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727483313807664e-06,
+      "loss": 0.2883,
+      "step": 3182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72167731038243e-06,
+      "loss": 0.2918,
+      "step": 3183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715872116907864e-06,
+      "loss": 0.2805,
+      "step": 3184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71006773544774e-06,
+      "loss": 0.3083,
+      "step": 3185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704264168065524e-06,
+      "loss": 0.2926,
+      "step": 3186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698461416824411e-06,
+      "loss": 0.2916,
+      "step": 3187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692659483787292e-06,
+      "loss": 0.3036,
+      "step": 3188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686858371016766e-06,
+      "loss": 0.3008,
+      "step": 3189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681058080575158e-06,
+      "loss": 0.2911,
+      "step": 3190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675258614524483e-06,
+      "loss": 0.2916,
+      "step": 3191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.669459974926467e-06,
+      "loss": 0.301,
+      "step": 3192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663662163842549e-06,
+      "loss": 0.2988,
+      "step": 3193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657865183333862e-06,
+      "loss": 0.3088,
+      "step": 3194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652069035461256e-06,
+      "loss": 0.3147,
+      "step": 3195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64627372228528e-06,
+      "loss": 0.2963,
+      "step": 3196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640479245866178e-06,
+      "loss": 0.3014,
+      "step": 3197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634685608263915e-06,
+      "loss": 0.2924,
+      "step": 3198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628892811538138e-06,
+      "loss": 0.3029,
+      "step": 3199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623100857748203e-06,
+      "loss": 0.2888,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617309748953175e-06,
+      "loss": 0.2906,
+      "step": 3201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611519487211805e-06,
+      "loss": 0.2918,
+      "step": 3202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.605730074582552e-06,
+      "loss": 0.3038,
+      "step": 3203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.59994151312357e-06,
+      "loss": 0.2828,
+      "step": 3204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.594153804892703e-06,
+      "loss": 0.2957,
+      "step": 3205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58836695194751e-06,
+      "loss": 0.2957,
+      "step": 3206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582580956345227e-06,
+      "loss": 0.3026,
+      "step": 3207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5767958201427984e-06,
+      "loss": 0.2985,
+      "step": 3208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571011545396854e-06,
+      "loss": 0.2933,
+      "step": 3209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565228134163719e-06,
+      "loss": 0.3,
+      "step": 3210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559445588499421e-06,
+      "loss": 0.2925,
+      "step": 3211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553663910459668e-06,
+      "loss": 0.3072,
+      "step": 3212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547883102099862e-06,
+      "loss": 0.2795,
+      "step": 3213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5421031654751056e-06,
+      "loss": 0.2938,
+      "step": 3214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536324102640172e-06,
+      "loss": 0.2987,
+      "step": 3215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530545915649549e-06,
+      "loss": 0.3016,
+      "step": 3216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524768606557392e-06,
+      "loss": 0.2833,
+      "step": 3217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.518992177417552e-06,
+      "loss": 0.3126,
+      "step": 3218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513216630283572e-06,
+      "loss": 0.3016,
+      "step": 3219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507441967208673e-06,
+      "loss": 0.2892,
+      "step": 3220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50166819024576e-06,
+      "loss": 0.2936,
+      "step": 3221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49589530144744e-06,
+      "loss": 0.2899,
+      "step": 3222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.490123302865983e-06,
+      "loss": 0.3042,
+      "step": 3223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4843521965533595e-06,
+      "loss": 0.2941,
+      "step": 3224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478581984561211e-06,
+      "loss": 0.2792,
+      "step": 3225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.472812668940862e-06,
+      "loss": 0.2945,
+      "step": 3226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46704425174333e-06,
+      "loss": 0.2935,
+      "step": 3227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461276735019303e-06,
+      "loss": 0.3011,
+      "step": 3228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455510120819144e-06,
+      "loss": 0.2912,
+      "step": 3229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4497444111929095e-06,
+      "loss": 0.2862,
+      "step": 3230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.443979608190321e-06,
+      "loss": 0.3011,
+      "step": 3231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.43821571386079e-06,
+      "loss": 0.299,
+      "step": 3232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432452730253395e-06,
+      "loss": 0.3027,
+      "step": 3233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426690659416893e-06,
+      "loss": 0.2928,
+      "step": 3234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420929503399721e-06,
+      "loss": 0.2996,
+      "step": 3235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415169264249987e-06,
+      "loss": 0.3014,
+      "step": 3236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409409944015465e-06,
+      "loss": 0.298,
+      "step": 3237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.403651544743625e-06,
+      "loss": 0.2859,
+      "step": 3238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.397894068481586e-06,
+      "loss": 0.2978,
+      "step": 3239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392137517276154e-06,
+      "loss": 0.2761,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386381893173797e-06,
+      "loss": 0.3068,
+      "step": 3241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3806271982206545e-06,
+      "loss": 0.2939,
+      "step": 3242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.374873434462546e-06,
+      "loss": 0.2874,
+      "step": 3243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3691206039449455e-06,
+      "loss": 0.3054,
+      "step": 3244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363368708713004e-06,
+      "loss": 0.3066,
+      "step": 3245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357617750811538e-06,
+      "loss": 0.2966,
+      "step": 3246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351867732285026e-06,
+      "loss": 0.288,
+      "step": 3247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346118655177626e-06,
+      "loss": 0.2844,
+      "step": 3248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340370521533148e-06,
+      "loss": 0.2981,
+      "step": 3249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334623333395068e-06,
+      "loss": 0.2921,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328877092806534e-06,
+      "loss": 0.2948,
+      "step": 3251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32313180181035e-06,
+      "loss": 0.3041,
+      "step": 3252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.317387462448982e-06,
+      "loss": 0.2975,
+      "step": 3253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311644076764565e-06,
+      "loss": 0.2889,
+      "step": 3254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305901646798885e-06,
+      "loss": 0.2851,
+      "step": 3255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300160174593399e-06,
+      "loss": 0.3029,
+      "step": 3256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294419662189215e-06,
+      "loss": 0.2941,
+      "step": 3257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288680111627098e-06,
+      "loss": 0.2864,
+      "step": 3258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282941524947485e-06,
+      "loss": 0.2976,
+      "step": 3259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2772039041904555e-06,
+      "loss": 0.3005,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.271467251395752e-06,
+      "loss": 0.2936,
+      "step": 3261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.265731568602774e-06,
+      "loss": 0.2867,
+      "step": 3262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2599968578505675e-06,
+      "loss": 0.2896,
+      "step": 3263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2542631211778495e-06,
+      "loss": 0.3059,
+      "step": 3264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.248530360622975e-06,
+      "loss": 0.295,
+      "step": 3265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242798578223957e-06,
+      "loss": 0.2959,
+      "step": 3266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237067776018467e-06,
+      "loss": 0.2776,
+      "step": 3267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231337956043814e-06,
+      "loss": 0.282,
+      "step": 3268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225609120336977e-06,
+      "loss": 0.2976,
+      "step": 3269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21988127093457e-06,
+      "loss": 0.2843,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214154409872857e-06,
+      "loss": 0.3042,
+      "step": 3271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20842853918776e-06,
+      "loss": 0.3071,
+      "step": 3272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.202703660914841e-06,
+      "loss": 0.3,
+      "step": 3273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.196979777089309e-06,
+      "loss": 0.2972,
+      "step": 3274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.19125688974603e-06,
+      "loss": 0.2928,
+      "step": 3275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1855350009195e-06,
+      "loss": 0.2751,
+      "step": 3276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.179814112643874e-06,
+      "loss": 0.28,
+      "step": 3277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174094226952941e-06,
+      "loss": 0.2981,
+      "step": 3278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168375345880137e-06,
+      "loss": 0.2911,
+      "step": 3279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162657471458548e-06,
+      "loss": 0.2917,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156940605720891e-06,
+      "loss": 0.2923,
+      "step": 3281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.151224750699528e-06,
+      "loss": 0.275,
+      "step": 3282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145509908426468e-06,
+      "loss": 0.2928,
+      "step": 3283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1397960809333464e-06,
+      "loss": 0.287,
+      "step": 3284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1340832702514575e-06,
+      "loss": 0.2866,
+      "step": 3285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128371478411715e-06,
+      "loss": 0.3054,
+      "step": 3286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.12266070744468e-06,
+      "loss": 0.2928,
+      "step": 3287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116950959380551e-06,
+      "loss": 0.298,
+      "step": 3288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111242236249161e-06,
+      "loss": 0.2914,
+      "step": 3289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10553454007997e-06,
+      "loss": 0.2948,
+      "step": 3290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099827872902094e-06,
+      "loss": 0.3047,
+      "step": 3291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0941222367442635e-06,
+      "loss": 0.2769,
+      "step": 3292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088417633634852e-06,
+      "loss": 0.3088,
+      "step": 3293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082714065601863e-06,
+      "loss": 0.2913,
+      "step": 3294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.077011534672926e-06,
+      "loss": 0.2863,
+      "step": 3295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07131004287532e-06,
+      "loss": 0.2916,
+      "step": 3296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065609592235936e-06,
+      "loss": 0.3005,
+      "step": 3297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0599101847813e-06,
+      "loss": 0.2969,
+      "step": 3298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054211822537573e-06,
+      "loss": 0.295,
+      "step": 3299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048514507530534e-06,
+      "loss": 0.28,
+      "step": 3300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042818241785605e-06,
+      "loss": 0.2915,
+      "step": 3301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037123027327819e-06,
+      "loss": 0.2978,
+      "step": 3302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.031428866181844e-06,
+      "loss": 0.3069,
+      "step": 3303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0257357603719724e-06,
+      "loss": 0.3048,
+      "step": 3304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02004371192212e-06,
+      "loss": 0.297,
+      "step": 3305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014352722855824e-06,
+      "loss": 0.2925,
+      "step": 3306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.008662795196255e-06,
+      "loss": 0.301,
+      "step": 3307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.002973930966195e-06,
+      "loss": 0.2936,
+      "step": 3308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9972861321880566e-06,
+      "loss": 0.3206,
+      "step": 3309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991599400883866e-06,
+      "loss": 0.293,
+      "step": 3310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985913739075272e-06,
+      "loss": 0.3021,
+      "step": 3311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98022914878355e-06,
+      "loss": 0.2867,
+      "step": 3312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974545632029588e-06,
+      "loss": 0.3022,
+      "step": 3313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968863190833889e-06,
+      "loss": 0.2897,
+      "step": 3314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963181827216583e-06,
+      "loss": 0.2872,
+      "step": 3315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957501543197407e-06,
+      "loss": 0.2823,
+      "step": 3316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951822340795727e-06,
+      "loss": 0.289,
+      "step": 3317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.946144222030512e-06,
+      "loss": 0.2949,
+      "step": 3318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94046718892035e-06,
+      "loss": 0.2886,
+      "step": 3319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9347912434834455e-06,
+      "loss": 0.3014,
+      "step": 3320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929116387737611e-06,
+      "loss": 0.291,
+      "step": 3321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923442623700281e-06,
+      "loss": 0.3055,
+      "step": 3322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9177699533884915e-06,
+      "loss": 0.3018,
+      "step": 3323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912098378818895e-06,
+      "loss": 0.3066,
+      "step": 3324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906427902007755e-06,
+      "loss": 0.312,
+      "step": 3325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.900758524970943e-06,
+      "loss": 0.291,
+      "step": 3326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895090249723934e-06,
+      "loss": 0.2858,
+      "step": 3327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889423078281826e-06,
+      "loss": 0.3063,
+      "step": 3328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.883757012659311e-06,
+      "loss": 0.3008,
+      "step": 3329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878092054870694e-06,
+      "loss": 0.2786,
+      "step": 3330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872428206929887e-06,
+      "loss": 0.2665,
+      "step": 3331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.866765470850396e-06,
+      "loss": 0.2903,
+      "step": 3332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861103848645352e-06,
+      "loss": 0.2894,
+      "step": 3333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855443342327474e-06,
+      "loss": 0.2898,
+      "step": 3334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84978395390909e-06,
+      "loss": 0.2862,
+      "step": 3335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84412568540213e-06,
+      "loss": 0.296,
+      "step": 3336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838468538818122e-06,
+      "loss": 0.2878,
+      "step": 3337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832812516168206e-06,
+      "loss": 0.2821,
+      "step": 3338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.827157619463112e-06,
+      "loss": 0.2861,
+      "step": 3339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82150385071317e-06,
+      "loss": 0.2852,
+      "step": 3340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.815851211928317e-06,
+      "loss": 0.2938,
+      "step": 3341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.810199705118081e-06,
+      "loss": 0.2867,
+      "step": 3342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.804549332291586e-06,
+      "loss": 0.2809,
+      "step": 3343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.798900095457563e-06,
+      "loss": 0.2802,
+      "step": 3344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793251996624332e-06,
+      "loss": 0.2945,
+      "step": 3345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.787605037799808e-06,
+      "loss": 0.3114,
+      "step": 3346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781959220991503e-06,
+      "loss": 0.2744,
+      "step": 3347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776314548206517e-06,
+      "loss": 0.2917,
+      "step": 3348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770671021451557e-06,
+      "loss": 0.291,
+      "step": 3349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76502864273291e-06,
+      "loss": 0.2906,
+      "step": 3350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.759387414056455e-06,
+      "loss": 0.2832,
+      "step": 3351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753747337427674e-06,
+      "loss": 0.3007,
+      "step": 3352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748108414851622e-06,
+      "loss": 0.2865,
+      "step": 3353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7424706483329615e-06,
+      "loss": 0.301,
+      "step": 3354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.736834039875934e-06,
+      "loss": 0.304,
+      "step": 3355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731198591484366e-06,
+      "loss": 0.2975,
+      "step": 3356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7255643051616825e-06,
+      "loss": 0.2905,
+      "step": 3357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719931182910888e-06,
+      "loss": 0.2813,
+      "step": 3358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714299226734569e-06,
+      "loss": 0.2963,
+      "step": 3359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708668438634911e-06,
+      "loss": 0.2899,
+      "step": 3360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7030388206136735e-06,
+      "loss": 0.2945,
+      "step": 3361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697410374672203e-06,
+      "loss": 0.2978,
+      "step": 3362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691783102811429e-06,
+      "loss": 0.2989,
+      "step": 3363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6861570070318595e-06,
+      "loss": 0.2824,
+      "step": 3364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680532089333598e-06,
+      "loss": 0.2764,
+      "step": 3365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6749083517163135e-06,
+      "loss": 0.3176,
+      "step": 3366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669285796179262e-06,
+      "loss": 0.2981,
+      "step": 3367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663664424721283e-06,
+      "loss": 0.2858,
+      "step": 3368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658044239340784e-06,
+      "loss": 0.2856,
+      "step": 3369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652425242035766e-06,
+      "loss": 0.2811,
+      "step": 3370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646807434803797e-06,
+      "loss": 0.2844,
+      "step": 3371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641190819642023e-06,
+      "loss": 0.2869,
+      "step": 3372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635575398547169e-06,
+      "loss": 0.2867,
+      "step": 3373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629961173515531e-06,
+      "loss": 0.2916,
+      "step": 3374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624348146542991e-06,
+      "loss": 0.2781,
+      "step": 3375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618736319624991e-06,
+      "loss": 0.3041,
+      "step": 3376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6131256947565525e-06,
+      "loss": 0.2925,
+      "step": 3377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.607516273932275e-06,
+      "loss": 0.2946,
+      "step": 3378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601908059146321e-06,
+      "loss": 0.2896,
+      "step": 3379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596301052392422e-06,
+      "loss": 0.3016,
+      "step": 3380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590695255663897e-06,
+      "loss": 0.2953,
+      "step": 3381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585090670953618e-06,
+      "loss": 0.2816,
+      "step": 3382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579487300254035e-06,
+      "loss": 0.287,
+      "step": 3383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573885145557161e-06,
+      "loss": 0.2799,
+      "step": 3384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568284208854574e-06,
+      "loss": 0.2985,
+      "step": 3385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562684492137435e-06,
+      "loss": 0.2869,
+      "step": 3386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557085997396455e-06,
+      "loss": 0.2755,
+      "step": 3387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551488726621915e-06,
+      "loss": 0.2888,
+      "step": 3388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545892681803663e-06,
+      "loss": 0.2903,
+      "step": 3389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540297864931107e-06,
+      "loss": 0.2923,
+      "step": 3390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534704277993228e-06,
+      "loss": 0.2784,
+      "step": 3391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529111922978559e-06,
+      "loss": 0.2961,
+      "step": 3392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523520801875198e-06,
+      "loss": 0.2924,
+      "step": 3393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.51793091667081e-06,
+      "loss": 0.293,
+      "step": 3394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.512342269352614e-06,
+      "loss": 0.2737,
+      "step": 3395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506754861907387e-06,
+      "loss": 0.2863,
+      "step": 3396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501168696321476e-06,
+      "loss": 0.2898,
+      "step": 3397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495583774580776e-06,
+      "loss": 0.2872,
+      "step": 3398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490000098670748e-06,
+      "loss": 0.2864,
+      "step": 3399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484417670576402e-06,
+      "loss": 0.2994,
+      "step": 3400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.478836492282303e-06,
+      "loss": 0.2925,
+      "step": 3401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47325656577259e-06,
+      "loss": 0.2795,
+      "step": 3402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467677893030935e-06,
+      "loss": 0.2813,
+      "step": 3403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462100476040574e-06,
+      "loss": 0.2946,
+      "step": 3404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456524316784299e-06,
+      "loss": 0.2863,
+      "step": 3405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450949417244446e-06,
+      "loss": 0.2985,
+      "step": 3406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445375779402914e-06,
+      "loss": 0.287,
+      "step": 3407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439803405241149e-06,
+      "loss": 0.2971,
+      "step": 3408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434232296740141e-06,
+      "loss": 0.2817,
+      "step": 3409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428662455880442e-06,
+      "loss": 0.3068,
+      "step": 3410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423093884642146e-06,
+      "loss": 0.2923,
+      "step": 3411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.417526585004892e-06,
+      "loss": 0.296,
+      "step": 3412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411960558947878e-06,
+      "loss": 0.3145,
+      "step": 3413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4063958084498425e-06,
+      "loss": 0.2986,
+      "step": 3414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400832335489071e-06,
+      "loss": 0.2808,
+      "step": 3415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395270142043395e-06,
+      "loss": 0.2866,
+      "step": 3416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389709230090185e-06,
+      "loss": 0.2721,
+      "step": 3417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.384149601606373e-06,
+      "loss": 0.2867,
+      "step": 3418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.378591258568421e-06,
+      "loss": 0.2814,
+      "step": 3419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3730342029523284e-06,
+      "loss": 0.2986,
+      "step": 3420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3674784367336565e-06,
+      "loss": 0.2883,
+      "step": 3421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.361923961887486e-06,
+      "loss": 0.3055,
+      "step": 3422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356370780388462e-06,
+      "loss": 0.2871,
+      "step": 3423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350818894210753e-06,
+      "loss": 0.2991,
+      "step": 3424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345268305328064e-06,
+      "loss": 0.2875,
+      "step": 3425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.339719015713657e-06,
+      "loss": 0.2867,
+      "step": 3426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334171027340311e-06,
+      "loss": 0.2785,
+      "step": 3427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.328624342180364e-06,
+      "loss": 0.2976,
+      "step": 3428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323078962205673e-06,
+      "loss": 0.2955,
+      "step": 3429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317534889387638e-06,
+      "loss": 0.2836,
+      "step": 3430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311992125697195e-06,
+      "loss": 0.2922,
+      "step": 3431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306450673104813e-06,
+      "loss": 0.3078,
+      "step": 3432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300910533580491e-06,
+      "loss": 0.2935,
+      "step": 3433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295371709093774e-06,
+      "loss": 0.2966,
+      "step": 3434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.289834201613725e-06,
+      "loss": 0.2745,
+      "step": 3435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284298013108947e-06,
+      "loss": 0.2886,
+      "step": 3436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2787631455475725e-06,
+      "loss": 0.2938,
+      "step": 3437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273229600897258e-06,
+      "loss": 0.3047,
+      "step": 3438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267697381125205e-06,
+      "loss": 0.2983,
+      "step": 3439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2621664881981315e-06,
+      "loss": 0.2836,
+      "step": 3440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2566369240822775e-06,
+      "loss": 0.2878,
+      "step": 3441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.251108690743433e-06,
+      "loss": 0.2773,
+      "step": 3442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245581790146893e-06,
+      "loss": 0.2776,
+      "step": 3443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240056224257494e-06,
+      "loss": 0.2968,
+      "step": 3444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23453199503959e-06,
+      "loss": 0.2965,
+      "step": 3445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.229009104457053e-06,
+      "loss": 0.3063,
+      "step": 3446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2234875544733e-06,
+      "loss": 0.2955,
+      "step": 3447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217967347051252e-06,
+      "loss": 0.2875,
+      "step": 3448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212448484153357e-06,
+      "loss": 0.2941,
+      "step": 3449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206930967741598e-06,
+      "loss": 0.3019,
+      "step": 3450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201414799777454e-06,
+      "loss": 0.295,
+      "step": 3451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1958999822219535e-06,
+      "loss": 0.2863,
+      "step": 3452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.190386517035625e-06,
+      "loss": 0.2809,
+      "step": 3453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1848744061785185e-06,
+      "loss": 0.296,
+      "step": 3454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179363651610216e-06,
+      "loss": 0.282,
+      "step": 3455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1738542552898e-06,
+      "loss": 0.2874,
+      "step": 3456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168346219175876e-06,
+      "loss": 0.2754,
+      "step": 3457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.162839545226574e-06,
+      "loss": 0.2991,
+      "step": 3458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.157334235399528e-06,
+      "loss": 0.2721,
+      "step": 3459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151830291651899e-06,
+      "loss": 0.2783,
+      "step": 3460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146327715940351e-06,
+      "loss": 0.2832,
+      "step": 3461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140826510221063e-06,
+      "loss": 0.2868,
+      "step": 3462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13532667644974e-06,
+      "loss": 0.2968,
+      "step": 3463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129828216581584e-06,
+      "loss": 0.2812,
+      "step": 3464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.124331132571308e-06,
+      "loss": 0.2877,
+      "step": 3465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118835426373157e-06,
+      "loss": 0.2816,
+      "step": 3466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113341099940857e-06,
+      "loss": 0.2868,
+      "step": 3467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.107848155227669e-06,
+      "loss": 0.2819,
+      "step": 3468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102356594186345e-06,
+      "loss": 0.2823,
+      "step": 3469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096866418769151e-06,
+      "loss": 0.2997,
+      "step": 3470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091377630927868e-06,
+      "loss": 0.295,
+      "step": 3471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085890232613772e-06,
+      "loss": 0.287,
+      "step": 3472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080404225777643e-06,
+      "loss": 0.2959,
+      "step": 3473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.074919612369787e-06,
+      "loss": 0.2924,
+      "step": 3474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069436394339989e-06,
+      "loss": 0.286,
+      "step": 3475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063954573637562e-06,
+      "loss": 0.2662,
+      "step": 3476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0584741522113e-06,
+      "loss": 0.29,
+      "step": 3477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.05299513200951e-06,
+      "loss": 0.287,
+      "step": 3478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0475175149800035e-06,
+      "loss": 0.2886,
+      "step": 3479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042041303070087e-06,
+      "loss": 0.2745,
+      "step": 3480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0365664982265815e-06,
+      "loss": 0.2974,
+      "step": 3481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031093102395784e-06,
+      "loss": 0.2839,
+      "step": 3482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0256211175235034e-06,
+      "loss": 0.2953,
+      "step": 3483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0201505455550565e-06,
+      "loss": 0.2847,
+      "step": 3484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014681388435243e-06,
+      "loss": 0.283,
+      "step": 3485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.009213648108363e-06,
+      "loss": 0.2809,
+      "step": 3486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003747326518219e-06,
+      "loss": 0.2928,
+      "step": 3487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.998282425608099e-06,
+      "loss": 0.3014,
+      "step": 3488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992818947320801e-06,
+      "loss": 0.2839,
+      "step": 3489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.987356893598602e-06,
+      "loss": 0.3084,
+      "step": 3490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981896266383279e-06,
+      "loss": 0.3049,
+      "step": 3491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976437067616105e-06,
+      "loss": 0.2983,
+      "step": 3492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.970979299237838e-06,
+      "loss": 0.2886,
+      "step": 3493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96552296318873e-06,
+      "loss": 0.282,
+      "step": 3494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960068061408532e-06,
+      "loss": 0.263,
+      "step": 3495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95461459583647e-06,
+      "loss": 0.2999,
+      "step": 3496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949162568411277e-06,
+      "loss": 0.2971,
+      "step": 3497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94371198107116e-06,
+      "loss": 0.2855,
+      "step": 3498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938262835753814e-06,
+      "loss": 0.2938,
+      "step": 3499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932815134396438e-06,
+      "loss": 0.2829,
+      "step": 3500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9273688789357e-06,
+      "loss": 0.2964,
+      "step": 3501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.92192407130776e-06,
+      "loss": 0.3004,
+      "step": 3502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9164807134482675e-06,
+      "loss": 0.2808,
+      "step": 3503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9110388072923455e-06,
+      "loss": 0.2881,
+      "step": 3504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905598354774618e-06,
+      "loss": 0.2847,
+      "step": 3505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900159357829177e-06,
+      "loss": 0.2838,
+      "step": 3506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894721818389602e-06,
+      "loss": 0.2823,
+      "step": 3507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.889285738388958e-06,
+      "loss": 0.2939,
+      "step": 3508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883851119759787e-06,
+      "loss": 0.2763,
+      "step": 3509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878417964434107e-06,
+      "loss": 0.2954,
+      "step": 3510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872986274343428e-06,
+      "loss": 0.3102,
+      "step": 3511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.867556051418731e-06,
+      "loss": 0.2701,
+      "step": 3512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.862127297590477e-06,
+      "loss": 0.2868,
+      "step": 3513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856700014788606e-06,
+      "loss": 0.2927,
+      "step": 3514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851274204942525e-06,
+      "loss": 0.2848,
+      "step": 3515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.2936,
+      "step": 3516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840427011832805e-06,
+      "loss": 0.2993,
+      "step": 3517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83500563242537e-06,
+      "loss": 0.3038,
+      "step": 3518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8295857336861525e-06,
+      "loss": 0.2892,
+      "step": 3519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824167317541938e-06,
+      "loss": 0.2935,
+      "step": 3520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818750385918996e-06,
+      "loss": 0.2972,
+      "step": 3521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81333494074306e-06,
+      "loss": 0.2882,
+      "step": 3522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807920983939333e-06,
+      "loss": 0.2811,
+      "step": 3523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8025085174324995e-06,
+      "loss": 0.2936,
+      "step": 3524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797097543146706e-06,
+      "loss": 0.287,
+      "step": 3525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79168806300557e-06,
+      "loss": 0.2836,
+      "step": 3526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786280078932176e-06,
+      "loss": 0.28,
+      "step": 3527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780873592849079e-06,
+      "loss": 0.2862,
+      "step": 3528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775468606678305e-06,
+      "loss": 0.2928,
+      "step": 3529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770065122341343e-06,
+      "loss": 0.2922,
+      "step": 3530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7646631417591415e-06,
+      "loss": 0.2874,
+      "step": 3531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.759262666852129e-06,
+      "loss": 0.2996,
+      "step": 3532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753863699540185e-06,
+      "loss": 0.2879,
+      "step": 3533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748466241742665e-06,
+      "loss": 0.2845,
+      "step": 3534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.743070295378377e-06,
+      "loss": 0.3001,
+      "step": 3535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.737675862365599e-06,
+      "loss": 0.2775,
+      "step": 3536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732282944622067e-06,
+      "loss": 0.2722,
+      "step": 3537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7268915440649784e-06,
+      "loss": 0.2998,
+      "step": 3538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721501662610992e-06,
+      "loss": 0.293,
+      "step": 3539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7161133021762315e-06,
+      "loss": 0.2981,
+      "step": 3540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710726464676271e-06,
+      "loss": 0.2819,
+      "step": 3541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705341152026152e-06,
+      "loss": 0.2923,
+      "step": 3542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699957366140367e-06,
+      "loss": 0.2958,
+      "step": 3543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694575108932866e-06,
+      "loss": 0.2841,
+      "step": 3544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689194382317065e-06,
+      "loss": 0.296,
+      "step": 3545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683815188205822e-06,
+      "loss": 0.2919,
+      "step": 3546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67843752851146e-06,
+      "loss": 0.2895,
+      "step": 3547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6730614051457545e-06,
+      "loss": 0.2801,
+      "step": 3548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6676868200199265e-06,
+      "loss": 0.2853,
+      "step": 3549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662313775044665e-06,
+      "loss": 0.2706,
+      "step": 3550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6569422721301034e-06,
+      "loss": 0.2881,
+      "step": 3551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651572313185822e-06,
+      "loss": 0.3066,
+      "step": 3552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646203900120866e-06,
+      "loss": 0.2898,
+      "step": 3553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6408370348437165e-06,
+      "loss": 0.2839,
+      "step": 3554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635471719262308e-06,
+      "loss": 0.289,
+      "step": 3555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630107955284036e-06,
+      "loss": 0.301,
+      "step": 3556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62474574481573e-06,
+      "loss": 0.2899,
+      "step": 3557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.619385089763669e-06,
+      "loss": 0.2688,
+      "step": 3558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614025992033586e-06,
+      "loss": 0.2748,
+      "step": 3559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608668453530653e-06,
+      "loss": 0.2828,
+      "step": 3560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603312476159495e-06,
+      "loss": 0.2811,
+      "step": 3561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597958061824179e-06,
+      "loss": 0.2946,
+      "step": 3562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.592605212428209e-06,
+      "loss": 0.3074,
+      "step": 3563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587253929874549e-06,
+      "loss": 0.2811,
+      "step": 3564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5819042160655865e-06,
+      "loss": 0.2946,
+      "step": 3565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576556072903171e-06,
+      "loss": 0.277,
+      "step": 3566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571209502288578e-06,
+      "loss": 0.2852,
+      "step": 3567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56586450612253e-06,
+      "loss": 0.2886,
+      "step": 3568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56052108630519e-06,
+      "loss": 0.2865,
+      "step": 3569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555179244736161e-06,
+      "loss": 0.3032,
+      "step": 3570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549838983314477e-06,
+      "loss": 0.2871,
+      "step": 3571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544500303938629e-06,
+      "loss": 0.2867,
+      "step": 3572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539163208506524e-06,
+      "loss": 0.2957,
+      "step": 3573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.533827698915521e-06,
+      "loss": 0.2782,
+      "step": 3574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528493777062411e-06,
+      "loss": 0.2858,
+      "step": 3575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523161444843414e-06,
+      "loss": 0.2936,
+      "step": 3576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.517830704154196e-06,
+      "loss": 0.2911,
+      "step": 3577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512501556889849e-06,
+      "loss": 0.2912,
+      "step": 3578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.507174004944902e-06,
+      "loss": 0.2969,
+      "step": 3579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501848050213314e-06,
+      "loss": 0.2737,
+      "step": 3580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496523694588476e-06,
+      "loss": 0.2941,
+      "step": 3581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491200939963221e-06,
+      "loss": 0.3004,
+      "step": 3582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.485879788229798e-06,
+      "loss": 0.2888,
+      "step": 3583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480560241279891e-06,
+      "loss": 0.2871,
+      "step": 3584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475242301004623e-06,
+      "loss": 0.2932,
+      "step": 3585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46992596929453e-06,
+      "loss": 0.3049,
+      "step": 3586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46461124803959e-06,
+      "loss": 0.2924,
+      "step": 3587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459298139129201e-06,
+      "loss": 0.2887,
+      "step": 3588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453986644452187e-06,
+      "loss": 0.2812,
+      "step": 3589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448676765896806e-06,
+      "loss": 0.2856,
+      "step": 3590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443368505350729e-06,
+      "loss": 0.2761,
+      "step": 3591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43806186470106e-06,
+      "loss": 0.2939,
+      "step": 3592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432756845834331e-06,
+      "loss": 0.2888,
+      "step": 3593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4274534506364864e-06,
+      "loss": 0.2954,
+      "step": 3594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422151680992907e-06,
+      "loss": 0.2819,
+      "step": 3595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416851538788381e-06,
+      "loss": 0.2873,
+      "step": 3596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.411553025907126e-06,
+      "loss": 0.2941,
+      "step": 3597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4062561442327845e-06,
+      "loss": 0.2795,
+      "step": 3598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40096089564841e-06,
+      "loss": 0.2774,
+      "step": 3599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.395667282036483e-06,
+      "loss": 0.2894,
+      "step": 3600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390375305278895e-06,
+      "loss": 0.2758,
+      "step": 3601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.385084967256957e-06,
+      "loss": 0.3043,
+      "step": 3602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379796269851411e-06,
+      "loss": 0.2718,
+      "step": 3603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3745092149423976e-06,
+      "loss": 0.302,
+      "step": 3604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369223804409478e-06,
+      "loss": 0.2844,
+      "step": 3605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363940040131641e-06,
+      "loss": 0.2825,
+      "step": 3606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358657923987276e-06,
+      "loss": 0.2886,
+      "step": 3607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353377457854186e-06,
+      "loss": 0.2847,
+      "step": 3608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348098643609604e-06,
+      "loss": 0.2925,
+      "step": 3609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3428214831301585e-06,
+      "loss": 0.2892,
+      "step": 3610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337545978291896e-06,
+      "loss": 0.2808,
+      "step": 3611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.332272130970276e-06,
+      "loss": 0.2969,
+      "step": 3612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326999943040163e-06,
+      "loss": 0.2921,
+      "step": 3613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321729416375842e-06,
+      "loss": 0.2982,
+      "step": 3614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3164605528509995e-06,
+      "loss": 0.2803,
+      "step": 3615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311193354338727e-06,
+      "loss": 0.2942,
+      "step": 3616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305927822711539e-06,
+      "loss": 0.276,
+      "step": 3617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300663959841339e-06,
+      "loss": 0.2947,
+      "step": 3618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2954017675994565e-06,
+      "loss": 0.2809,
+      "step": 3619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29014124785661e-06,
+      "loss": 0.2993,
+      "step": 3620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284882402482931e-06,
+      "loss": 0.2732,
+      "step": 3621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279625233347957e-06,
+      "loss": 0.278,
+      "step": 3622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274369742320625e-06,
+      "loss": 0.2776,
+      "step": 3623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269115931269277e-06,
+      "loss": 0.3053,
+      "step": 3624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263863802061664e-06,
+      "loss": 0.2912,
+      "step": 3625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2586133565649275e-06,
+      "loss": 0.2765,
+      "step": 3626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.253364596645625e-06,
+      "loss": 0.2719,
+      "step": 3627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2481175241697024e-06,
+      "loss": 0.2754,
+      "step": 3628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242872141002506e-06,
+      "loss": 0.2983,
+      "step": 3629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237628449008794e-06,
+      "loss": 0.3044,
+      "step": 3630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23238645005271e-06,
+      "loss": 0.2997,
+      "step": 3631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227146145997801e-06,
+      "loss": 0.2736,
+      "step": 3632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221907538707012e-06,
+      "loss": 0.2872,
+      "step": 3633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216670630042677e-06,
+      "loss": 0.2773,
+      "step": 3634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211435421866544e-06,
+      "loss": 0.2858,
+      "step": 3635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206201916039741e-06,
+      "loss": 0.2951,
+      "step": 3636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200970114422791e-06,
+      "loss": 0.2902,
+      "step": 3637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195740018875622e-06,
+      "loss": 0.2792,
+      "step": 3638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190511631257548e-06,
+      "loss": 0.2906,
+      "step": 3639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18528495342727e-06,
+      "loss": 0.2777,
+      "step": 3640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180059987242897e-06,
+      "loss": 0.2697,
+      "step": 3641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.174836734561916e-06,
+      "loss": 0.2908,
+      "step": 3642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169615197241211e-06,
+      "loss": 0.2963,
+      "step": 3643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164395377137052e-06,
+      "loss": 0.2953,
+      "step": 3644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159177276105099e-06,
+      "loss": 0.2971,
+      "step": 3645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153960896000408e-06,
+      "loss": 0.2745,
+      "step": 3646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1487462386774135e-06,
+      "loss": 0.2939,
+      "step": 3647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1435333059899486e-06,
+      "loss": 0.2868,
+      "step": 3648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138322099791222e-06,
+      "loss": 0.3025,
+      "step": 3649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133112621933831e-06,
+      "loss": 0.2892,
+      "step": 3650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127904874269768e-06,
+      "loss": 0.2866,
+      "step": 3651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122698858650399e-06,
+      "loss": 0.2819,
+      "step": 3652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117494576926478e-06,
+      "loss": 0.2764,
+      "step": 3653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112292030948145e-06,
+      "loss": 0.2797,
+      "step": 3654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107091222564916e-06,
+      "loss": 0.2966,
+      "step": 3655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101892153625702e-06,
+      "loss": 0.2911,
+      "step": 3656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096694825978783e-06,
+      "loss": 0.2751,
+      "step": 3657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091499241471822e-06,
+      "loss": 0.2964,
+      "step": 3658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086305401951872e-06,
+      "loss": 0.2731,
+      "step": 3659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081113309265358e-06,
+      "loss": 0.2841,
+      "step": 3660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.075922965258077e-06,
+      "loss": 0.2906,
+      "step": 3661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.070734371775221e-06,
+      "loss": 0.2888,
+      "step": 3662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065547530661349e-06,
+      "loss": 0.2949,
+      "step": 3663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060362443760396e-06,
+      "loss": 0.2858,
+      "step": 3664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055179112915678e-06,
+      "loss": 0.3036,
+      "step": 3665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049997539969882e-06,
+      "loss": 0.289,
+      "step": 3666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044817726765078e-06,
+      "loss": 0.279,
+      "step": 3667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039639675142703e-06,
+      "loss": 0.2806,
+      "step": 3668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0344633869435665e-06,
+      "loss": 0.2881,
+      "step": 3669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029288864007862e-06,
+      "loss": 0.2761,
+      "step": 3670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.024116108175141e-06,
+      "loss": 0.2882,
+      "step": 3671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018945121284342e-06,
+      "loss": 0.2791,
+      "step": 3672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0137759051737615e-06,
+      "loss": 0.2707,
+      "step": 3673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008608461681071e-06,
+      "loss": 0.288,
+      "step": 3674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003442792643316e-06,
+      "loss": 0.2911,
+      "step": 3675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998278899896904e-06,
+      "loss": 0.2776,
+      "step": 3676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993116785277613e-06,
+      "loss": 0.3048,
+      "step": 3677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987956450620597e-06,
+      "loss": 0.3009,
+      "step": 3678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982797897760363e-06,
+      "loss": 0.2783,
+      "step": 3679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9776411285308005e-06,
+      "loss": 0.29,
+      "step": 3680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972486144765153e-06,
+      "loss": 0.297,
+      "step": 3681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967332948296028e-06,
+      "loss": 0.277,
+      "step": 3682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962181540955412e-06,
+      "loss": 0.2888,
+      "step": 3683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957031924574641e-06,
+      "loss": 0.2927,
+      "step": 3684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9518841009844195e-06,
+      "loss": 0.2977,
+      "step": 3685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946738072014815e-06,
+      "loss": 0.2921,
+      "step": 3686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.941593839495253e-06,
+      "loss": 0.298,
+      "step": 3687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93645140525453e-06,
+      "loss": 0.2765,
+      "step": 3688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9313107711207944e-06,
+      "loss": 0.2913,
+      "step": 3689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926171938921553e-06,
+      "loss": 0.2893,
+      "step": 3690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921034910483684e-06,
+      "loss": 0.278,
+      "step": 3691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.915899687633414e-06,
+      "loss": 0.2952,
+      "step": 3692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910766272196325e-06,
+      "loss": 0.278,
+      "step": 3693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905634665997371e-06,
+      "loss": 0.2822,
+      "step": 3694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900504870860848e-06,
+      "loss": 0.2827,
+      "step": 3695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895376888610415e-06,
+      "loss": 0.2853,
+      "step": 3696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8902507210690856e-06,
+      "loss": 0.2767,
+      "step": 3697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8851263700592235e-06,
+      "loss": 0.2972,
+      "step": 3698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880003837402559e-06,
+      "loss": 0.2791,
+      "step": 3699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874883124920161e-06,
+      "loss": 0.2887,
+      "step": 3700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869764234432466e-06,
+      "loss": 0.282,
+      "step": 3701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86464716775925e-06,
+      "loss": 0.2813,
+      "step": 3702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859531926719644e-06,
+      "loss": 0.2826,
+      "step": 3703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854418513132137e-06,
+      "loss": 0.2882,
+      "step": 3704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849306928814562e-06,
+      "loss": 0.2842,
+      "step": 3705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844197175584101e-06,
+      "loss": 0.2795,
+      "step": 3706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.839089255257288e-06,
+      "loss": 0.2817,
+      "step": 3707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83398316965e-06,
+      "loss": 0.29,
+      "step": 3708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828878920577473e-06,
+      "loss": 0.2845,
+      "step": 3709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.82377650985428e-06,
+      "loss": 0.2767,
+      "step": 3710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818675939294339e-06,
+      "loss": 0.2885,
+      "step": 3711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813577210710928e-06,
+      "loss": 0.2993,
+      "step": 3712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808480325916656e-06,
+      "loss": 0.2973,
+      "step": 3713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803385286723475e-06,
+      "loss": 0.2758,
+      "step": 3714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798292094942697e-06,
+      "loss": 0.2761,
+      "step": 3715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793200752384963e-06,
+      "loss": 0.2869,
+      "step": 3716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788111260860261e-06,
+      "loss": 0.2763,
+      "step": 3717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78302362217792e-06,
+      "loss": 0.2946,
+      "step": 3718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777937838146608e-06,
+      "loss": 0.2852,
+      "step": 3719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772853910574345e-06,
+      "loss": 0.2798,
+      "step": 3720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.767771841268476e-06,
+      "loss": 0.2911,
+      "step": 3721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762691632035691e-06,
+      "loss": 0.293,
+      "step": 3722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757613284682028e-06,
+      "loss": 0.2835,
+      "step": 3723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.752536801012845e-06,
+      "loss": 0.2852,
+      "step": 3724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747462182832856e-06,
+      "loss": 0.2776,
+      "step": 3725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7423894319461e-06,
+      "loss": 0.2948,
+      "step": 3726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7373185501559575e-06,
+      "loss": 0.2939,
+      "step": 3727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732249539265138e-06,
+      "loss": 0.2821,
+      "step": 3728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727182401075693e-06,
+      "loss": 0.2896,
+      "step": 3729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722117137389001e-06,
+      "loss": 0.292,
+      "step": 3730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717053750005786e-06,
+      "loss": 0.2845,
+      "step": 3731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711992240726093e-06,
+      "loss": 0.2825,
+      "step": 3732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7069326113493075e-06,
+      "loss": 0.2731,
+      "step": 3733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7018748636741415e-06,
+      "loss": 0.2794,
+      "step": 3734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696818999498636e-06,
+      "loss": 0.2897,
+      "step": 3735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691765020620174e-06,
+      "loss": 0.2778,
+      "step": 3736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6867129288354555e-06,
+      "loss": 0.2869,
+      "step": 3737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.681662725940514e-06,
+      "loss": 0.2905,
+      "step": 3738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.676614413730713e-06,
+      "loss": 0.2885,
+      "step": 3739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671567994000741e-06,
+      "loss": 0.2839,
+      "step": 3740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66652346854462e-06,
+      "loss": 0.2969,
+      "step": 3741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661480839155693e-06,
+      "loss": 0.2968,
+      "step": 3742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6564401076266256e-06,
+      "loss": 0.274,
+      "step": 3743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.651401275749419e-06,
+      "loss": 0.2882,
+      "step": 3744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646364345315392e-06,
+      "loss": 0.277,
+      "step": 3745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.641329318115183e-06,
+      "loss": 0.2782,
+      "step": 3746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.636296195938769e-06,
+      "loss": 0.2763,
+      "step": 3747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631264980575435e-06,
+      "loss": 0.283,
+      "step": 3748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626235673813794e-06,
+      "loss": 0.2777,
+      "step": 3749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6212082774417795e-06,
+      "loss": 0.3039,
+      "step": 3750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616182793246644e-06,
+      "loss": 0.2722,
+      "step": 3751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611159223014968e-06,
+      "loss": 0.2867,
+      "step": 3752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60613756853264e-06,
+      "loss": 0.2784,
+      "step": 3753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60111783158488e-06,
+      "loss": 0.272,
+      "step": 3754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596100013956217e-06,
+      "loss": 0.2854,
+      "step": 3755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591084117430495e-06,
+      "loss": 0.2705,
+      "step": 3756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.586070143790891e-06,
+      "loss": 0.2808,
+      "step": 3757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581058094819882e-06,
+      "loss": 0.2894,
+      "step": 3758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576047972299267e-06,
+      "loss": 0.2967,
+      "step": 3759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57103977801016e-06,
+      "loss": 0.2772,
+      "step": 3760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566033513732986e-06,
+      "loss": 0.2775,
+      "step": 3761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561029181247494e-06,
+      "loss": 0.2659,
+      "step": 3762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556026782332737e-06,
+      "loss": 0.2895,
+      "step": 3763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5510263187670765e-06,
+      "loss": 0.2886,
+      "step": 3764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5460277923282034e-06,
+      "loss": 0.2866,
+      "step": 3765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5410312047931035e-06,
+      "loss": 0.2845,
+      "step": 3766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536036557938076e-06,
+      "loss": 0.293,
+      "step": 3767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.53104385353874e-06,
+      "loss": 0.2767,
+      "step": 3768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.526053093370014e-06,
+      "loss": 0.2895,
+      "step": 3769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.521064279206129e-06,
+      "loss": 0.3013,
+      "step": 3770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516077412820622e-06,
+      "loss": 0.2817,
+      "step": 3771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5110924959863354e-06,
+      "loss": 0.2796,
+      "step": 3772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506109530475432e-06,
+      "loss": 0.2778,
+      "step": 3773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5011285180593655e-06,
+      "loss": 0.2798,
+      "step": 3774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496149460508899e-06,
+      "loss": 0.2907,
+      "step": 3775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491172359594109e-06,
+      "loss": 0.2966,
+      "step": 3776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486197217084363e-06,
+      "loss": 0.3004,
+      "step": 3777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481224034748345e-06,
+      "loss": 0.2869,
+      "step": 3778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476252814354036e-06,
+      "loss": 0.2795,
+      "step": 3779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4712835576687175e-06,
+      "loss": 0.2851,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466316266458977e-06,
+      "loss": 0.2867,
+      "step": 3781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461350942490701e-06,
+      "loss": 0.2811,
+      "step": 3782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456387587529073e-06,
+      "loss": 0.2961,
+      "step": 3783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.451426203338589e-06,
+      "loss": 0.2863,
+      "step": 3784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44646679168303e-06,
+      "loss": 0.292,
+      "step": 3785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441509354325487e-06,
+      "loss": 0.2784,
+      "step": 3786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436553893028341e-06,
+      "loss": 0.2936,
+      "step": 3787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431600409553272e-06,
+      "loss": 0.3,
+      "step": 3788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426648905661264e-06,
+      "loss": 0.2885,
+      "step": 3789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421699383112588e-06,
+      "loss": 0.2748,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4167518436668145e-06,
+      "loss": 0.2733,
+      "step": 3791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4118062890828094e-06,
+      "loss": 0.2839,
+      "step": 3792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40686272111873e-06,
+      "loss": 0.2944,
+      "step": 3793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401921141532034e-06,
+      "loss": 0.2704,
+      "step": 3794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3969815520794665e-06,
+      "loss": 0.2744,
+      "step": 3795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392043954517062e-06,
+      "loss": 0.2846,
+      "step": 3796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387108350600161e-06,
+      "loss": 0.2785,
+      "step": 3797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38217474208338e-06,
+      "loss": 0.2792,
+      "step": 3798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3772431307206274e-06,
+      "loss": 0.287,
+      "step": 3799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372313518265115e-06,
+      "loss": 0.2982,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367385906469333e-06,
+      "loss": 0.3053,
+      "step": 3801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3624602970850595e-06,
+      "loss": 0.281,
+      "step": 3802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357536691863364e-06,
+      "loss": 0.2724,
+      "step": 3803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3526150925546005e-06,
+      "loss": 0.2932,
+      "step": 3804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347695500908421e-06,
+      "loss": 0.2736,
+      "step": 3805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342777918673746e-06,
+      "loss": 0.2852,
+      "step": 3806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337862347598799e-06,
+      "loss": 0.3004,
+      "step": 3807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3329487894310766e-06,
+      "loss": 0.2908,
+      "step": 3808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328037245917361e-06,
+      "loss": 0.2869,
+      "step": 3809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.323127718803727e-06,
+      "loss": 0.2874,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318220209835524e-06,
+      "loss": 0.2647,
+      "step": 3811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313314720757385e-06,
+      "loss": 0.2697,
+      "step": 3812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308411253313228e-06,
+      "loss": 0.2709,
+      "step": 3813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303509809246247e-06,
+      "loss": 0.2963,
+      "step": 3814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298610390298926e-06,
+      "loss": 0.2788,
+      "step": 3815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.293712998213021e-06,
+      "loss": 0.2855,
+      "step": 3816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288817634729565e-06,
+      "loss": 0.2867,
+      "step": 3817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283924301588883e-06,
+      "loss": 0.2804,
+      "step": 3818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2790330005305665e-06,
+      "loss": 0.2892,
+      "step": 3819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.274143733293485e-06,
+      "loss": 0.2853,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269256501615792e-06,
+      "loss": 0.2769,
+      "step": 3821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2643713072349114e-06,
+      "loss": 0.2943,
+      "step": 3822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259488151887546e-06,
+      "loss": 0.2767,
+      "step": 3823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254607037309672e-06,
+      "loss": 0.294,
+      "step": 3824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2497279652365355e-06,
+      "loss": 0.275,
+      "step": 3825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244850937402669e-06,
+      "loss": 0.2869,
+      "step": 3826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.239975955541868e-06,
+      "loss": 0.2811,
+      "step": 3827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235103021387201e-06,
+      "loss": 0.2896,
+      "step": 3828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230232136671016e-06,
+      "loss": 0.2763,
+      "step": 3829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225363303124921e-06,
+      "loss": 0.295,
+      "step": 3830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220496522479808e-06,
+      "loss": 0.2972,
+      "step": 3831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21563179646583e-06,
+      "loss": 0.2803,
+      "step": 3832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210769126812409e-06,
+      "loss": 0.2715,
+      "step": 3833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205908515248244e-06,
+      "loss": 0.2984,
+      "step": 3834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201049963501292e-06,
+      "loss": 0.2819,
+      "step": 3835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196193473298783e-06,
+      "loss": 0.2818,
+      "step": 3836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191339046367219e-06,
+      "loss": 0.2959,
+      "step": 3837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186486684432357e-06,
+      "loss": 0.295,
+      "step": 3838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1816363892192355e-06,
+      "loss": 0.2862,
+      "step": 3839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176788162452143e-06,
+      "loss": 0.2813,
+      "step": 3840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171942005854636e-06,
+      "loss": 0.276,
+      "step": 3841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167097921149548e-06,
+      "loss": 0.2762,
+      "step": 3842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162255910058958e-06,
+      "loss": 0.2711,
+      "step": 3843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157415974304219e-06,
+      "loss": 0.2912,
+      "step": 3844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15257811560594e-06,
+      "loss": 0.2834,
+      "step": 3845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147742335683994e-06,
+      "loss": 0.287,
+      "step": 3846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.14290863625752e-06,
+      "loss": 0.2809,
+      "step": 3847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138077019044911e-06,
+      "loss": 0.2759,
+      "step": 3848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133247485763819e-06,
+      "loss": 0.2719,
+      "step": 3849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128420038131162e-06,
+      "loss": 0.2906,
+      "step": 3850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123594677863112e-06,
+      "loss": 0.2731,
+      "step": 3851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118771406675093e-06,
+      "loss": 0.2763,
+      "step": 3852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113950226281802e-06,
+      "loss": 0.2798,
+      "step": 3853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109131138397177e-06,
+      "loss": 0.2836,
+      "step": 3854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104314144734423e-06,
+      "loss": 0.2836,
+      "step": 3855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099499247005992e-06,
+      "loss": 0.2961,
+      "step": 3856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094686446923591e-06,
+      "loss": 0.2873,
+      "step": 3857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089875746198195e-06,
+      "loss": 0.2825,
+      "step": 3858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085067146540015e-06,
+      "loss": 0.2812,
+      "step": 3859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080260649658529e-06,
+      "loss": 0.264,
+      "step": 3860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075456257262457e-06,
+      "loss": 0.2815,
+      "step": 3861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070653971059773e-06,
+      "loss": 0.3049,
+      "step": 3862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065853792757712e-06,
+      "loss": 0.2979,
+      "step": 3863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061055724062748e-06,
+      "loss": 0.283,
+      "step": 3864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056259766680609e-06,
+      "loss": 0.2881,
+      "step": 3865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051465922316271e-06,
+      "loss": 0.2825,
+      "step": 3866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.04667419267396e-06,
+      "loss": 0.2762,
+      "step": 3867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041884579457154e-06,
+      "loss": 0.2921,
+      "step": 3868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.037097084368574e-06,
+      "loss": 0.2842,
+      "step": 3869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032311709110185e-06,
+      "loss": 0.2895,
+      "step": 3870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027528455383209e-06,
+      "loss": 0.2781,
+      "step": 3871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0227473248881045e-06,
+      "loss": 0.288,
+      "step": 3872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017968319324575e-06,
+      "loss": 0.2859,
+      "step": 3873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013191440391576e-06,
+      "loss": 0.2772,
+      "step": 3874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.008416689787302e-06,
+      "loss": 0.2817,
+      "step": 3875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003644069209191e-06,
+      "loss": 0.2767,
+      "step": 3876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998873580353921e-06,
+      "loss": 0.2856,
+      "step": 3877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994105224917416e-06,
+      "loss": 0.2744,
+      "step": 3878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989339004594846e-06,
+      "loss": 0.2831,
+      "step": 3879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984574921080614e-06,
+      "loss": 0.2723,
+      "step": 3880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979812976068362e-06,
+      "loss": 0.3039,
+      "step": 3881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975053171250983e-06,
+      "loss": 0.2809,
+      "step": 3882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970295508320596e-06,
+      "loss": 0.2911,
+      "step": 3883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.965539988968572e-06,
+      "loss": 0.2859,
+      "step": 3884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960786614885507e-06,
+      "loss": 0.2902,
+      "step": 3885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956035387761241e-06,
+      "loss": 0.2957,
+      "step": 3886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951286309284851e-06,
+      "loss": 0.2877,
+      "step": 3887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9465393811446485e-06,
+      "loss": 0.2743,
+      "step": 3888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.941794605028176e-06,
+      "loss": 0.2892,
+      "step": 3889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937051982622223e-06,
+      "loss": 0.2878,
+      "step": 3890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932311515612801e-06,
+      "loss": 0.277,
+      "step": 3891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927573205685167e-06,
+      "loss": 0.2797,
+      "step": 3892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9228370545238e-06,
+      "loss": 0.2809,
+      "step": 3893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918103063812414e-06,
+      "loss": 0.3073,
+      "step": 3894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913371235233964e-06,
+      "loss": 0.2954,
+      "step": 3895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.908641570470624e-06,
+      "loss": 0.2977,
+      "step": 3896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903914071203807e-06,
+      "loss": 0.2839,
+      "step": 3897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.899188739114151e-06,
+      "loss": 0.2859,
+      "step": 3898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894465575881526e-06,
+      "loss": 0.2927,
+      "step": 3899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889744583185034e-06,
+      "loss": 0.2811,
+      "step": 3900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885025762703001e-06,
+      "loss": 0.277,
+      "step": 3901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880309116112978e-06,
+      "loss": 0.2831,
+      "step": 3902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875594645091755e-06,
+      "loss": 0.2844,
+      "step": 3903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870882351315336e-06,
+      "loss": 0.2789,
+      "step": 3904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866172236458953e-06,
+      "loss": 0.2792,
+      "step": 3905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861464302197073e-06,
+      "loss": 0.2997,
+      "step": 3906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856758550203379e-06,
+      "loss": 0.2784,
+      "step": 3907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.852054982150778e-06,
+      "loss": 0.3038,
+      "step": 3908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847353599711404e-06,
+      "loss": 0.3033,
+      "step": 3909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84265440455661e-06,
+      "loss": 0.298,
+      "step": 3910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.837957398356979e-06,
+      "loss": 0.2765,
+      "step": 3911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8332625827823054e-06,
+      "loss": 0.2716,
+      "step": 3912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828569959501618e-06,
+      "loss": 0.2739,
+      "step": 3913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823879530183155e-06,
+      "loss": 0.2867,
+      "step": 3914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819191296494374e-06,
+      "loss": 0.2775,
+      "step": 3915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814505260101963e-06,
+      "loss": 0.2904,
+      "step": 3916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80982142267182e-06,
+      "loss": 0.2927,
+      "step": 3917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8051397858690643e-06,
+      "loss": 0.2888,
+      "step": 3918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8004603513580295e-06,
+      "loss": 0.2832,
+      "step": 3919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7957831208022676e-06,
+      "loss": 0.2726,
+      "step": 3920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7911080958645553e-06,
+      "loss": 0.2878,
+      "step": 3921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7864352782068727e-06,
+      "loss": 0.285,
+      "step": 3922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7817646694904185e-06,
+      "loss": 0.2838,
+      "step": 3923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7770962713756154e-06,
+      "loss": 0.2868,
+      "step": 3924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7724300855220906e-06,
+      "loss": 0.2839,
+      "step": 3925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7677661135886823e-06,
+      "loss": 0.2798,
+      "step": 3926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7631043572334546e-06,
+      "loss": 0.2697,
+      "step": 3927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7584448181136723e-06,
+      "loss": 0.2867,
+      "step": 3928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.753787497885816e-06,
+      "loss": 0.2811,
+      "step": 3929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7491323982055773e-06,
+      "loss": 0.2788,
+      "step": 3930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.744479520727854e-06,
+      "loss": 0.2883,
+      "step": 3931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.739828867106767e-06,
+      "loss": 0.2813,
+      "step": 3932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7351804389956324e-06,
+      "loss": 0.297,
+      "step": 3933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730534238046979e-06,
+      "loss": 0.2999,
+      "step": 3934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7258902659125517e-06,
+      "loss": 0.2941,
+      "step": 3935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7212485242432885e-06,
+      "loss": 0.2618,
+      "step": 3936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7166090146893506e-06,
+      "loss": 0.3018,
+      "step": 3937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7119717389000953e-06,
+      "loss": 0.2925,
+      "step": 3938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7073366985240876e-06,
+      "loss": 0.2838,
+      "step": 3939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.702703895209099e-06,
+      "loss": 0.274,
+      "step": 3940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6980733306021035e-06,
+      "loss": 0.2788,
+      "step": 3941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6934450063492788e-06,
+      "loss": 0.2848,
+      "step": 3942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6888189240960148e-06,
+      "loss": 0.2806,
+      "step": 3943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.684195085486891e-06,
+      "loss": 0.2793,
+      "step": 3944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6795734921657045e-06,
+      "loss": 0.2733,
+      "step": 3945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.67495414577544e-06,
+      "loss": 0.2808,
+      "step": 3946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.670337047958288e-06,
+      "loss": 0.2772,
+      "step": 3947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.665722200355646e-06,
+      "loss": 0.2613,
+      "step": 3948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661109604608104e-06,
+      "loss": 0.2764,
+      "step": 3949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.656499262355453e-06,
+      "loss": 0.2795,
+      "step": 3950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6518911752366847e-06,
+      "loss": 0.2911,
+      "step": 3951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6472853448899847e-06,
+      "loss": 0.2867,
+      "step": 3952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6426817729527454e-06,
+      "loss": 0.2798,
+      "step": 3953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6380804610615485e-06,
+      "loss": 0.274,
+      "step": 3954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.633481410852171e-06,
+      "loss": 0.2728,
+      "step": 3955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6288846239595955e-06,
+      "loss": 0.2666,
+      "step": 3956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6242901020179898e-06,
+      "loss": 0.2881,
+      "step": 3957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.619697846660719e-06,
+      "loss": 0.2884,
+      "step": 3958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.615107859520348e-06,
+      "loss": 0.2878,
+      "step": 3959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6105201422286307e-06,
+      "loss": 0.2866,
+      "step": 3960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6059346964165133e-06,
+      "loss": 0.274,
+      "step": 3961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.601351523714135e-06,
+      "loss": 0.2823,
+      "step": 3962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.596770625750826e-06,
+      "loss": 0.2729,
+      "step": 3963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.592192004155115e-06,
+      "loss": 0.2835,
+      "step": 3964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.587615660554713e-06,
+      "loss": 0.2647,
+      "step": 3965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5830415965765206e-06,
+      "loss": 0.2891,
+      "step": 3966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578469813846638e-06,
+      "loss": 0.2704,
+      "step": 3967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5739003139903416e-06,
+      "loss": 0.2921,
+      "step": 3968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5693330986321096e-06,
+      "loss": 0.31,
+      "step": 3969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.564768169395598e-06,
+      "loss": 0.2686,
+      "step": 3970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5602055279036517e-06,
+      "loss": 0.2897,
+      "step": 3971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.555645175778304e-06,
+      "loss": 0.2812,
+      "step": 3972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551087114640771e-06,
+      "loss": 0.2873,
+      "step": 3973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546531346111464e-06,
+      "loss": 0.2862,
+      "step": 3974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5419778718099693e-06,
+      "loss": 0.2703,
+      "step": 3975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5374266933550584e-06,
+      "loss": 0.2871,
+      "step": 3976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5328778123646943e-06,
+      "loss": 0.2733,
+      "step": 3977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5283312304560156e-06,
+      "loss": 0.2925,
+      "step": 3978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5237869492453423e-06,
+      "loss": 0.265,
+      "step": 3979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5192449703481877e-06,
+      "loss": 0.2886,
+      "step": 3980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5147052953792347e-06,
+      "loss": 0.2679,
+      "step": 3981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5101679259523537e-06,
+      "loss": 0.2835,
+      "step": 3982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5056328636805904e-06,
+      "loss": 0.2804,
+      "step": 3983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5011001101761723e-06,
+      "loss": 0.2795,
+      "step": 3984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4965696670505123e-06,
+      "loss": 0.2773,
+      "step": 3985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4920415359141944e-06,
+      "loss": 0.277,
+      "step": 3986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4875157183769793e-06,
+      "loss": 0.2828,
+      "step": 3987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482992216047816e-06,
+      "loss": 0.2906,
+      "step": 3988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4784710305348157e-06,
+      "loss": 0.2689,
+      "step": 3989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473952163445281e-06,
+      "loss": 0.2874,
+      "step": 3990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4694356163856814e-06,
+      "loss": 0.2829,
+      "step": 3991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4649213909616597e-06,
+      "loss": 0.2659,
+      "step": 3992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.460409488778039e-06,
+      "loss": 0.2887,
+      "step": 3993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4558999114388148e-06,
+      "loss": 0.2842,
+      "step": 3994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.2719,
+      "step": 3995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4468877377053945e-06,
+      "loss": 0.2583,
+      "step": 3996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.442385144515056e-06,
+      "loss": 0.2891,
+      "step": 3997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437884882576824e-06,
+      "loss": 0.296,
+      "step": 3998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433386953490555e-06,
+      "loss": 0.2903,
+      "step": 3999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4288913588552717e-06,
+      "loss": 0.2731,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4243981002691795e-06,
+      "loss": 0.2829,
+      "step": 4001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.419907179329641e-06,
+      "loss": 0.2842,
+      "step": 4002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4154185976331922e-06,
+      "loss": 0.2713,
+      "step": 4003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4109323567755394e-06,
+      "loss": 0.2838,
+      "step": 4004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.406448458351549e-06,
+      "loss": 0.273,
+      "step": 4005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4019669039552695e-06,
+      "loss": 0.2732,
+      "step": 4006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397487695179903e-06,
+      "loss": 0.2792,
+      "step": 4007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393010833617818e-06,
+      "loss": 0.2849,
+      "step": 4008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.388536320860559e-06,
+      "loss": 0.2809,
+      "step": 4009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3840641584988276e-06,
+      "loss": 0.3003,
+      "step": 4010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3795943481224857e-06,
+      "loss": 0.2845,
+      "step": 4011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3751268913205727e-06,
+      "loss": 0.2794,
+      "step": 4012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3706617896812775e-06,
+      "loss": 0.2751,
+      "step": 4013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.36619904479196e-06,
+      "loss": 0.2855,
+      "step": 4014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3617386582391388e-06,
+      "loss": 0.2714,
+      "step": 4015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3572806316084917e-06,
+      "loss": 0.2771,
+      "step": 4016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3528249664848667e-06,
+      "loss": 0.2626,
+      "step": 4017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3483716644522636e-06,
+      "loss": 0.2836,
+      "step": 4018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.34392072709384e-06,
+      "loss": 0.2805,
+      "step": 4019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3394721559919275e-06,
+      "loss": 0.2734,
+      "step": 4020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.335025952727998e-06,
+      "loss": 0.2917,
+      "step": 4021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330582118882697e-06,
+      "loss": 0.281,
+      "step": 4022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.326140656035818e-06,
+      "loss": 0.2887,
+      "step": 4023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3217015657663145e-06,
+      "loss": 0.2864,
+      "step": 4024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3172648496522974e-06,
+      "loss": 0.2788,
+      "step": 4025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.312830509271029e-06,
+      "loss": 0.2863,
+      "step": 4026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3083985461989367e-06,
+      "loss": 0.2742,
+      "step": 4027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3039689620115932e-06,
+      "loss": 0.2774,
+      "step": 4028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2995417582837285e-06,
+      "loss": 0.2822,
+      "step": 4029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295116936589232e-06,
+      "loss": 0.2845,
+      "step": 4030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290694498501138e-06,
+      "loss": 0.2733,
+      "step": 4031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2862744455916327e-06,
+      "loss": 0.2889,
+      "step": 4032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2818567794320667e-06,
+      "loss": 0.274,
+      "step": 4033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2774415015929297e-06,
+      "loss": 0.2893,
+      "step": 4034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.273028613643866e-06,
+      "loss": 0.2817,
+      "step": 4035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268618117153671e-06,
+      "loss": 0.2723,
+      "step": 4036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2642100136902863e-06,
+      "loss": 0.2715,
+      "step": 4037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259804304820814e-06,
+      "loss": 0.281,
+      "step": 4038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2554009921114916e-06,
+      "loss": 0.2807,
+      "step": 4039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2510000771277093e-06,
+      "loss": 0.303,
+      "step": 4040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246601561434011e-06,
+      "loss": 0.2825,
+      "step": 4041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2422054465940743e-06,
+      "loss": 0.2706,
+      "step": 4042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2378117341707417e-06,
+      "loss": 0.2932,
+      "step": 4043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233420425725985e-06,
+      "loss": 0.2842,
+      "step": 4044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.229031522820929e-06,
+      "loss": 0.2798,
+      "step": 4045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2246450270158425e-06,
+      "loss": 0.2814,
+      "step": 4046
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.220260939870138e-06,
+      "loss": 0.2834,
+      "step": 4047
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2158792629423673e-06,
+      "loss": 0.2874,
+      "step": 4048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2114999977902385e-06,
+      "loss": 0.289,
+      "step": 4049
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2071231459705856e-06,
+      "loss": 0.2597,
+      "step": 4050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2027487090393996e-06,
+      "loss": 0.2728,
+      "step": 4051
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.198376688551802e-06,
+      "loss": 0.2767,
+      "step": 4052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1940070860620586e-06,
+      "loss": 0.2719,
+      "step": 4053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1896399031235816e-06,
+      "loss": 0.2729,
+      "step": 4054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185275141288915e-06,
+      "loss": 0.2746,
+      "step": 4055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1809128021097445e-06,
+      "loss": 0.2643,
+      "step": 4056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1765528871368946e-06,
+      "loss": 0.2887,
+      "step": 4057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.172195397920326e-06,
+      "loss": 0.2715,
+      "step": 4058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.167840336009146e-06,
+      "loss": 0.2766,
+      "step": 4059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1634877029515877e-06,
+      "loss": 0.2795,
+      "step": 4060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159137500295024e-06,
+      "loss": 0.2821,
+      "step": 4061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154789729585971e-06,
+      "loss": 0.2767,
+      "step": 4062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1504443923700713e-06,
+      "loss": 0.2695,
+      "step": 4063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1461014901921015e-06,
+      "loss": 0.2871,
+      "step": 4064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141761024595984e-06,
+      "loss": 0.2896,
+      "step": 4065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137422997124765e-06,
+      "loss": 0.2748,
+      "step": 4066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.133087409320624e-06,
+      "loss": 0.2847,
+      "step": 4067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.128754262724879e-06,
+      "loss": 0.2853,
+      "step": 4068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1244235588779715e-06,
+      "loss": 0.2698,
+      "step": 4069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1200952993194856e-06,
+      "loss": 0.2626,
+      "step": 4070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1157694855881306e-06,
+      "loss": 0.28,
+      "step": 4071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1114461192217393e-06,
+      "loss": 0.2812,
+      "step": 4072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1071252017572916e-06,
+      "loss": 0.2737,
+      "step": 4073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1028067347308787e-06,
+      "loss": 0.2776,
+      "step": 4074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.098490719677736e-06,
+      "loss": 0.2898,
+      "step": 4075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0941771581322167e-06,
+      "loss": 0.2835,
+      "step": 4076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0898660516278043e-06,
+      "loss": 0.2769,
+      "step": 4077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085557401697111e-06,
+      "loss": 0.2813,
+      "step": 4078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0812512098718717e-06,
+      "loss": 0.2672,
+      "step": 4079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0769474776829566e-06,
+      "loss": 0.2911,
+      "step": 4080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0726462066603534e-06,
+      "loss": 0.2699,
+      "step": 4081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0683473983331733e-06,
+      "loss": 0.2676,
+      "step": 4082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.064051054229662e-06,
+      "loss": 0.284,
+      "step": 4083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0597571758771806e-06,
+      "loss": 0.2667,
+      "step": 4084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055465764802211e-06,
+      "loss": 0.2829,
+      "step": 4085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0511768225303727e-06,
+      "loss": 0.284,
+      "step": 4086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.046890350586391e-06,
+      "loss": 0.2695,
+      "step": 4087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0426063504941216e-06,
+      "loss": 0.2745,
+      "step": 4088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0383248237765383e-06,
+      "loss": 0.2836,
+      "step": 4089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.034045771955736e-06,
+      "loss": 0.2814,
+      "step": 4090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0297691965529353e-06,
+      "loss": 0.2734,
+      "step": 4091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.025495099088468e-06,
+      "loss": 0.2791,
+      "step": 4092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0212234810817877e-06,
+      "loss": 0.2787,
+      "step": 4093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0169543440514726e-06,
+      "loss": 0.2903,
+      "step": 4094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.012687689515208e-06,
+      "loss": 0.2823,
+      "step": 4095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.008423518989809e-06,
+      "loss": 0.2675,
+      "step": 4096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0041618339912017e-06,
+      "loss": 0.2822,
+      "step": 4097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9999026360344176e-06,
+      "loss": 0.2708,
+      "step": 4098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995645926633625e-06,
+      "loss": 0.2817,
+      "step": 4099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9913917073020926e-06,
+      "loss": 0.2821,
+      "step": 4100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9871399795522062e-06,
+      "loss": 0.272,
+      "step": 4101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9828907448954745e-06,
+      "loss": 0.2668,
+      "step": 4102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.978644004842507e-06,
+      "loss": 0.2784,
+      "step": 4103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9743997609030386e-06,
+      "loss": 0.2797,
+      "step": 4104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.970158014585908e-06,
+      "loss": 0.2861,
+      "step": 4105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9659187673990654e-06,
+      "loss": 0.279,
+      "step": 4106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9616820208495833e-06,
+      "loss": 0.261,
+      "step": 4107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.957447776443638e-06,
+      "loss": 0.2625,
+      "step": 4108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953216035686507e-06,
+      "loss": 0.3022,
+      "step": 4109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.948986800082595e-06,
+      "loss": 0.2708,
+      "step": 4110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.944760071135403e-06,
+      "loss": 0.2711,
+      "step": 4111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9405358503475536e-06,
+      "loss": 0.2653,
+      "step": 4112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9363141392207638e-06,
+      "loss": 0.29,
+      "step": 4113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.932094939255865e-06,
+      "loss": 0.2668,
+      "step": 4114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9278782519527993e-06,
+      "loss": 0.29,
+      "step": 4115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.923664078810611e-06,
+      "loss": 0.2884,
+      "step": 4116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9194524213274467e-06,
+      "loss": 0.2609,
+      "step": 4117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9152432810005737e-06,
+      "loss": 0.2797,
+      "step": 4118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9110366593263417e-06,
+      "loss": 0.2728,
+      "step": 4119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9068325578002264e-06,
+      "loss": 0.2744,
+      "step": 4120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.902630977916796e-06,
+      "loss": 0.2851,
+      "step": 4121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.898431921169723e-06,
+      "loss": 0.2664,
+      "step": 4122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8942353890517894e-06,
+      "loss": 0.2868,
+      "step": 4123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8900413830548722e-06,
+      "loss": 0.2828,
+      "step": 4124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8858499046699507e-06,
+      "loss": 0.2724,
+      "step": 4125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8816609553871146e-06,
+      "loss": 0.2729,
+      "step": 4126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8774745366955415e-06,
+      "loss": 0.2695,
+      "step": 4127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8732906500835266e-06,
+      "loss": 0.2863,
+      "step": 4128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.869109297038444e-06,
+      "loss": 0.2659,
+      "step": 4129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864930479046777e-06,
+      "loss": 0.2956,
+      "step": 4130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8607541975941144e-06,
+      "loss": 0.2678,
+      "step": 4131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.856580454165132e-06,
+      "loss": 0.2781,
+      "step": 4132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.852409250243614e-06,
+      "loss": 0.2895,
+      "step": 4133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848240587312433e-06,
+      "loss": 0.2944,
+      "step": 4134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.844074466853558e-06,
+      "loss": 0.2818,
+      "step": 4135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8399108903480645e-06,
+      "loss": 0.2929,
+      "step": 4136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8357498592761123e-06,
+      "loss": 0.2783,
+      "step": 4137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.831591375116961e-06,
+      "loss": 0.2942,
+      "step": 4138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8274354393489654e-06,
+      "loss": 0.2609,
+      "step": 4139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8232820534495697e-06,
+      "loss": 0.2874,
+      "step": 4140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8191312188953225e-06,
+      "loss": 0.2607,
+      "step": 4141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.814982937161852e-06,
+      "loss": 0.2777,
+      "step": 4142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.810837209723886e-06,
+      "loss": 0.2816,
+      "step": 4143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8066940380552455e-06,
+      "loss": 0.2725,
+      "step": 4144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.80255342362884e-06,
+      "loss": 0.2804,
+      "step": 4145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7984153679166672e-06,
+      "loss": 0.2812,
+      "step": 4146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7942798723898235e-06,
+      "loss": 0.2651,
+      "step": 4147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7901469385184866e-06,
+      "loss": 0.2803,
+      "step": 4148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786016567771933e-06,
+      "loss": 0.2725,
+      "step": 4149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7818887616185164e-06,
+      "loss": 0.2674,
+      "step": 4150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7777635215256814e-06,
+      "loss": 0.2838,
+      "step": 4151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7736408489599708e-06,
+      "loss": 0.2684,
+      "step": 4152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7695207453870056e-06,
+      "loss": 0.2813,
+      "step": 4153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765403212271489e-06,
+      "loss": 0.2747,
+      "step": 4154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7612882510772265e-06,
+      "loss": 0.274,
+      "step": 4155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7571758632670897e-06,
+      "loss": 0.2676,
+      "step": 4156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.753066050303054e-06,
+      "loss": 0.2641,
+      "step": 4157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7489588136461653e-06,
+      "loss": 0.2747,
+      "step": 4158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7448541547565598e-06,
+      "loss": 0.2694,
+      "step": 4159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7407520750934557e-06,
+      "loss": 0.2726,
+      "step": 4160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.736652576115156e-06,
+      "loss": 0.2676,
+      "step": 4161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7325556592790403e-06,
+      "loss": 0.2694,
+      "step": 4162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7284613260415837e-06,
+      "loss": 0.2675,
+      "step": 4163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7243695778583258e-06,
+      "loss": 0.2762,
+      "step": 4164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7202804161839016e-06,
+      "loss": 0.2854,
+      "step": 4165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716193842472019e-06,
+      "loss": 0.2722,
+      "step": 4166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7121098581754635e-06,
+      "loss": 0.2692,
+      "step": 4167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.70802846474611e-06,
+      "loss": 0.2823,
+      "step": 4168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.703949663634904e-06,
+      "loss": 0.2737,
+      "step": 4169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6998734562918715e-06,
+      "loss": 0.2645,
+      "step": 4170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.695799844166116e-06,
+      "loss": 0.2825,
+      "step": 4171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6917288287058164e-06,
+      "loss": 0.2752,
+      "step": 4172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.687660411358237e-06,
+      "loss": 0.2857,
+      "step": 4173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6835945935697103e-06,
+      "loss": 0.2832,
+      "step": 4174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6795313767856436e-06,
+      "loss": 0.2702,
+      "step": 4175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6754707624505284e-06,
+      "loss": 0.2722,
+      "step": 4176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6714127520079223e-06,
+      "loss": 0.2745,
+      "step": 4177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.667357346900459e-06,
+      "loss": 0.2736,
+      "step": 4178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6633045485698515e-06,
+      "loss": 0.277,
+      "step": 4179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6592543584568798e-06,
+      "loss": 0.2864,
+      "step": 4180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.655206778001399e-06,
+      "loss": 0.2708,
+      "step": 4181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6511618086423365e-06,
+      "loss": 0.2842,
+      "step": 4182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.647119451817688e-06,
+      "loss": 0.276,
+      "step": 4183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6430797089645298e-06,
+      "loss": 0.2688,
+      "step": 4184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6390425815189956e-06,
+      "loss": 0.2789,
+      "step": 4185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6350080709163037e-06,
+      "loss": 0.2781,
+      "step": 4186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6309761785907327e-06,
+      "loss": 0.2744,
+      "step": 4187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.626946905975629e-06,
+      "loss": 0.2757,
+      "step": 4188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622920254503416e-06,
+      "loss": 0.2788,
+      "step": 4189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6188962256055796e-06,
+      "loss": 0.2714,
+      "step": 4190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6148748207126726e-06,
+      "loss": 0.2889,
+      "step": 4191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.610856041254317e-06,
+      "loss": 0.267,
+      "step": 4192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6068398886592005e-06,
+      "loss": 0.2763,
+      "step": 4193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60282636435508e-06,
+      "loss": 0.271,
+      "step": 4194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598815469768776e-06,
+      "loss": 0.2651,
+      "step": 4195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5948072063261685e-06,
+      "loss": 0.2723,
+      "step": 4196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.590801575452214e-06,
+      "loss": 0.2743,
+      "step": 4197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5867985785709236e-06,
+      "loss": 0.2787,
+      "step": 4198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5827982171053724e-06,
+      "loss": 0.2635,
+      "step": 4199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5788004924777065e-06,
+      "loss": 0.2615,
+      "step": 4200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5748054061091265e-06,
+      "loss": 0.2681,
+      "step": 4201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5708129594198972e-06,
+      "loss": 0.2556,
+      "step": 4202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5668231538293466e-06,
+      "loss": 0.2739,
+      "step": 4203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5628359907558585e-06,
+      "loss": 0.2809,
+      "step": 4204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558851471616889e-06,
+      "loss": 0.2766,
+      "step": 4205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.554869597828943e-06,
+      "loss": 0.2678,
+      "step": 4206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5508903708075862e-06,
+      "loss": 0.2899,
+      "step": 4207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5469137919674504e-06,
+      "loss": 0.2787,
+      "step": 4208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5429398627222178e-06,
+      "loss": 0.293,
+      "step": 4209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.538968584484637e-06,
+      "loss": 0.2847,
+      "step": 4210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5349999586665077e-06,
+      "loss": 0.2736,
+      "step": 4211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.531033986678687e-06,
+      "loss": 0.275,
+      "step": 4212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527070669931091e-06,
+      "loss": 0.274,
+      "step": 4213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.52311000983269e-06,
+      "loss": 0.277,
+      "step": 4214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5191520077915087e-06,
+      "loss": 0.2723,
+      "step": 4215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5151966652146353e-06,
+      "loss": 0.2562,
+      "step": 4216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511243983508198e-06,
+      "loss": 0.2725,
+      "step": 4217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507293964077396e-06,
+      "loss": 0.2676,
+      "step": 4218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503346608326469e-06,
+      "loss": 0.2907,
+      "step": 4219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4994019176587102e-06,
+      "loss": 0.2672,
+      "step": 4220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4954598934764774e-06,
+      "loss": 0.2719,
+      "step": 4221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.491520537181167e-06,
+      "loss": 0.2724,
+      "step": 4222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.487583850173232e-06,
+      "loss": 0.2719,
+      "step": 4223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4836498338521787e-06,
+      "loss": 0.3025,
+      "step": 4224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.479718489616557e-06,
+      "loss": 0.2701,
+      "step": 4225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4757898188639793e-06,
+      "loss": 0.2682,
+      "step": 4226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4718638229910974e-06,
+      "loss": 0.2702,
+      "step": 4227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.46794050339361e-06,
+      "loss": 0.2825,
+      "step": 4228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464019861466276e-06,
+      "loss": 0.2753,
+      "step": 4229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4601018986028912e-06,
+      "loss": 0.2646,
+      "step": 4230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4561866161963033e-06,
+      "loss": 0.291,
+      "step": 4231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4522740156384118e-06,
+      "loss": 0.2748,
+      "step": 4232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.448364098320154e-06,
+      "loss": 0.2654,
+      "step": 4233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444456865631519e-06,
+      "loss": 0.2699,
+      "step": 4234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4405523189615387e-06,
+      "loss": 0.2584,
+      "step": 4235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.43665045969829e-06,
+      "loss": 0.2756,
+      "step": 4236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4327512892288997e-06,
+      "loss": 0.2855,
+      "step": 4237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4288548089395316e-06,
+      "loss": 0.2704,
+      "step": 4238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4249610202153985e-06,
+      "loss": 0.2839,
+      "step": 4239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.421069924440754e-06,
+      "loss": 0.2748,
+      "step": 4240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4171815229988916e-06,
+      "loss": 0.293,
+      "step": 4241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4132958172721544e-06,
+      "loss": 0.2794,
+      "step": 4242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409412808641921e-06,
+      "loss": 0.2852,
+      "step": 4243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.405532498488612e-06,
+      "loss": 0.2682,
+      "step": 4244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.40165488819169e-06,
+      "loss": 0.2778,
+      "step": 4245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.397779979129653e-06,
+      "loss": 0.2795,
+      "step": 4246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393907772680051e-06,
+      "loss": 0.2798,
+      "step": 4247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3900382702194613e-06,
+      "loss": 0.2905,
+      "step": 4248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3861714731235e-06,
+      "loss": 0.2682,
+      "step": 4249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.382307382766832e-06,
+      "loss": 0.2782,
+      "step": 4250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3784460005231512e-06,
+      "loss": 0.2695,
+      "step": 4251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3745873277651867e-06,
+      "loss": 0.2836,
+      "step": 4252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3707313658647137e-06,
+      "loss": 0.2675,
+      "step": 4253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.366878116192537e-06,
+      "loss": 0.2845,
+      "step": 4254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3630275801184975e-06,
+      "loss": 0.2818,
+      "step": 4255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.359179759011474e-06,
+      "loss": 0.287,
+      "step": 4256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355334654239374e-06,
+      "loss": 0.2688,
+      "step": 4257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351492267169151e-06,
+      "loss": 0.2755,
+      "step": 4258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.347652599166781e-06,
+      "loss": 0.267,
+      "step": 4259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3438156515972764e-06,
+      "loss": 0.2864,
+      "step": 4260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339981425824689e-06,
+      "loss": 0.2587,
+      "step": 4261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3361499232120923e-06,
+      "loss": 0.2767,
+      "step": 4262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332321145121602e-06,
+      "loss": 0.2639,
+      "step": 4263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.328495092914358e-06,
+      "loss": 0.2622,
+      "step": 4264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3246717679505337e-06,
+      "loss": 0.2847,
+      "step": 4265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3208511715893334e-06,
+      "loss": 0.2776,
+      "step": 4266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.317033305188989e-06,
+      "loss": 0.27,
+      "step": 4267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3132181701067624e-06,
+      "loss": 0.2585,
+      "step": 4268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3094057676989514e-06,
+      "loss": 0.2789,
+      "step": 4269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3055960993208704e-06,
+      "loss": 0.2711,
+      "step": 4270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3017891663268743e-06,
+      "loss": 0.2704,
+      "step": 4271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.297984970070336e-06,
+      "loss": 0.2727,
+      "step": 4272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.294183511903656e-06,
+      "loss": 0.2623,
+      "step": 4273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2903847931782707e-06,
+      "loss": 0.2798,
+      "step": 4274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2865888152446325e-06,
+      "loss": 0.2853,
+      "step": 4275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282795579452224e-06,
+      "loss": 0.2677,
+      "step": 4276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2790050871495518e-06,
+      "loss": 0.263,
+      "step": 4277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2752173396841436e-06,
+      "loss": 0.2769,
+      "step": 4278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.271432338402564e-06,
+      "loss": 0.2825,
+      "step": 4279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267650084650386e-06,
+      "loss": 0.2774,
+      "step": 4280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2638705797722104e-06,
+      "loss": 0.2979,
+      "step": 4281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.26009382511167e-06,
+      "loss": 0.275,
+      "step": 4282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256319822011409e-06,
+      "loss": 0.2843,
+      "step": 4283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2525485718130933e-06,
+      "loss": 0.27,
+      "step": 4284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2487800758574208e-06,
+      "loss": 0.2867,
+      "step": 4285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2450143354840994e-06,
+      "loss": 0.2672,
+      "step": 4286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2412513520318626e-06,
+      "loss": 0.2812,
+      "step": 4287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2374911268384614e-06,
+      "loss": 0.2797,
+      "step": 4288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2337336612406644e-06,
+      "loss": 0.2852,
+      "step": 4289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229978956574268e-06,
+      "loss": 0.2816,
+      "step": 4290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2262270141740796e-06,
+      "loss": 0.2701,
+      "step": 4291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2224778353739206e-06,
+      "loss": 0.2719,
+      "step": 4292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2187314215066447e-06,
+      "loss": 0.2687,
+      "step": 4293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2149877739041047e-06,
+      "loss": 0.2691,
+      "step": 4294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2112468938971865e-06,
+      "loss": 0.2843,
+      "step": 4295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2075087828157803e-06,
+      "loss": 0.2783,
+      "step": 4296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2037734419887967e-06,
+      "loss": 0.2653,
+      "step": 4297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.20004087274416e-06,
+      "loss": 0.2872,
+      "step": 4298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196311076408808e-06,
+      "loss": 0.2869,
+      "step": 4299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1925840543086994e-06,
+      "loss": 0.2793,
+      "step": 4300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1888598077688005e-06,
+      "loss": 0.2841,
+      "step": 4301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1851383381130886e-06,
+      "loss": 0.2928,
+      "step": 4302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181419646664562e-06,
+      "loss": 0.2811,
+      "step": 4303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.177703734745227e-06,
+      "loss": 0.282,
+      "step": 4304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173990603676096e-06,
+      "loss": 0.2721,
+      "step": 4305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170280254777205e-06,
+      "loss": 0.2874,
+      "step": 4306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.166572689367592e-06,
+      "loss": 0.2642,
+      "step": 4307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1628679087653084e-06,
+      "loss": 0.2922,
+      "step": 4308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1591659142874133e-06,
+      "loss": 0.2872,
+      "step": 4309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1554667072499747e-06,
+      "loss": 0.2773,
+      "step": 4310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.151770288968079e-06,
+      "loss": 0.2677,
+      "step": 4311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1480766607558103e-06,
+      "loss": 0.2775,
+      "step": 4312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1443858239262607e-06,
+      "loss": 0.2822,
+      "step": 4313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140697779791542e-06,
+      "loss": 0.2888,
+      "step": 4314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1370125296627585e-06,
+      "loss": 0.2648,
+      "step": 4315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133330074850033e-06,
+      "loss": 0.2719,
+      "step": 4316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1296504166624863e-06,
+      "loss": 0.2833,
+      "step": 4317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1259735564082494e-06,
+      "loss": 0.2651,
+      "step": 4318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1222994953944577e-06,
+      "loss": 0.2686,
+      "step": 4319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.118628234927249e-06,
+      "loss": 0.273,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114959776311767e-06,
+      "loss": 0.2772,
+      "step": 4321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1112941208521633e-06,
+      "loss": 0.2663,
+      "step": 4322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1076312698515855e-06,
+      "loss": 0.2797,
+      "step": 4323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1039712246121947e-06,
+      "loss": 0.2953,
+      "step": 4324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.100313986435143e-06,
+      "loss": 0.2812,
+      "step": 4325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0966595566205893e-06,
+      "loss": 0.2787,
+      "step": 4326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0930079364676982e-06,
+      "loss": 0.2977,
+      "step": 4327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0893591272746304e-06,
+      "loss": 0.2736,
+      "step": 4328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0857131303385482e-06,
+      "loss": 0.2651,
+      "step": 4329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082069946955615e-06,
+      "loss": 0.2569,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0784295784209896e-06,
+      "loss": 0.2674,
+      "step": 4331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.074792026028841e-06,
+      "loss": 0.2755,
+      "step": 4332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0711572910723277e-06,
+      "loss": 0.2817,
+      "step": 4333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0675253748436055e-06,
+      "loss": 0.2746,
+      "step": 4334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063896278633837e-06,
+      "loss": 0.2694,
+      "step": 4335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0602700037331736e-06,
+      "loss": 0.2708,
+      "step": 4336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0566465514307666e-06,
+      "loss": 0.2766,
+      "step": 4337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0530259230147686e-06,
+      "loss": 0.2725,
+      "step": 4338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0494081197723206e-06,
+      "loss": 0.2665,
+      "step": 4339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045793142989564e-06,
+      "loss": 0.2718,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0421809939516323e-06,
+      "loss": 0.2678,
+      "step": 4341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0385716739426552e-06,
+      "loss": 0.2939,
+      "step": 4342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0349651842457606e-06,
+      "loss": 0.2586,
+      "step": 4343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0313615261430653e-06,
+      "loss": 0.2632,
+      "step": 4344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0277607009156765e-06,
+      "loss": 0.2852,
+      "step": 4345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0241627098437043e-06,
+      "loss": 0.2628,
+      "step": 4346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.020567554206242e-06,
+      "loss": 0.2763,
+      "step": 4347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016975235281381e-06,
+      "loss": 0.2723,
+      "step": 4348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013385754346201e-06,
+      "loss": 0.2621,
+      "step": 4349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0097991126767725e-06,
+      "loss": 0.2587,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0062153115481575e-06,
+      "loss": 0.272,
+      "step": 4351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002634352234406e-06,
+      "loss": 0.2724,
+      "step": 4352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999056236008564e-06,
+      "loss": 0.2763,
+      "step": 4353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995480964142663e-06,
+      "loss": 0.274,
+      "step": 4354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9919085379077172e-06,
+      "loss": 0.2722,
+      "step": 4355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9883389585737434e-06,
+      "loss": 0.2732,
+      "step": 4356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9847722274097325e-06,
+      "loss": 0.2676,
+      "step": 4357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9812083456836673e-06,
+      "loss": 0.2905,
+      "step": 4358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977647314662523e-06,
+      "loss": 0.2725,
+      "step": 4359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.974089135612255e-06,
+      "loss": 0.2623,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9705338097978065e-06,
+      "loss": 0.2784,
+      "step": 4361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966981338483107e-06,
+      "loss": 0.2925,
+      "step": 4362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9634317229310686e-06,
+      "loss": 0.2681,
+      "step": 4363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9598849644035944e-06,
+      "loss": 0.2849,
+      "step": 4364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9563410641615666e-06,
+      "loss": 0.2865,
+      "step": 4365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.952800023464849e-06,
+      "loss": 0.2676,
+      "step": 4366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9492618435722985e-06,
+      "loss": 0.2928,
+      "step": 4367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9457265257417436e-06,
+      "loss": 0.2664,
+      "step": 4368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9421940712300046e-06,
+      "loss": 0.2623,
+      "step": 4369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938664481292879e-06,
+      "loss": 0.268,
+      "step": 4370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.935137757185145e-06,
+      "loss": 0.2716,
+      "step": 4371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.931613900160564e-06,
+      "loss": 0.2884,
+      "step": 4372
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9280929114718795e-06,
+      "loss": 0.2832,
+      "step": 4373
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.92457479237081e-06,
+      "loss": 0.2722,
+      "step": 4374
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9210595441080627e-06,
+      "loss": 0.2778,
+      "step": 4375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9175471679333146e-06,
+      "loss": 0.2761,
+      "step": 4376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914037665095231e-06,
+      "loss": 0.272,
+      "step": 4377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105310368414477e-06,
+      "loss": 0.2639,
+      "step": 4378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.907027284418579e-06,
+      "loss": 0.2842,
+      "step": 4379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9035264090722261e-06,
+      "loss": 0.2736,
+      "step": 4380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9000284120469582e-06,
+      "loss": 0.2781,
+      "step": 4381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8965332945863224e-06,
+      "loss": 0.2708,
+      "step": 4382
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.893041057932845e-06,
+      "loss": 0.2643,
+      "step": 4383
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8895517033280242e-06,
+      "loss": 0.2776,
+      "step": 4384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8860652320123408e-06,
+      "loss": 0.2774,
+      "step": 4385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8825816452252432e-06,
+      "loss": 0.28,
+      "step": 4386
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8791009442051545e-06,
+      "loss": 0.282,
+      "step": 4387
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8756231301894801e-06,
+      "loss": 0.287,
+      "step": 4388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8721482044145921e-06,
+      "loss": 0.2769,
+      "step": 4389
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8686761681158328e-06,
+      "loss": 0.272,
+      "step": 4390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8652070225275277e-06,
+      "loss": 0.2557,
+      "step": 4391
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8617407688829658e-06,
+      "loss": 0.2765,
+      "step": 4392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8582774084144107e-06,
+      "loss": 0.2927,
+      "step": 4393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8548169423530982e-06,
+      "loss": 0.2753,
+      "step": 4394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.851359371929231e-06,
+      "loss": 0.2762,
+      "step": 4395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8479046983719927e-06,
+      "loss": 0.2877,
+      "step": 4396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8444529229095264e-06,
+      "loss": 0.2814,
+      "step": 4397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8410040467689472e-06,
+      "loss": 0.272,
+      "step": 4398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8375580711763453e-06,
+      "loss": 0.2603,
+      "step": 4399
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8341149973567707e-06,
+      "loss": 0.2643,
+      "step": 4400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8306748265342521e-06,
+      "loss": 0.2784,
+      "step": 4401
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8272375599317771e-06,
+      "loss": 0.2607,
+      "step": 4402
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8238031987713067e-06,
+      "loss": 0.2804,
+      "step": 4403
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8203717442737646e-06,
+      "loss": 0.2847,
+      "step": 4404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8169431976590412e-06,
+      "loss": 0.2733,
+      "step": 4405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.813517560146002e-06,
+      "loss": 0.2659,
+      "step": 4406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810094832952467e-06,
+      "loss": 0.277,
+      "step": 4407
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8066750172952241e-06,
+      "loss": 0.2769,
+      "step": 4408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8032581143900342e-06,
+      "loss": 0.2772,
+      "step": 4409
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7998441254516142e-06,
+      "loss": 0.267,
+      "step": 4410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7964330516936435e-06,
+      "loss": 0.2597,
+      "step": 4411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7930248943287753e-06,
+      "loss": 0.2934,
+      "step": 4412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7896196545686184e-06,
+      "loss": 0.2681,
+      "step": 4413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7862173336237443e-06,
+      "loss": 0.2789,
+      "step": 4414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7828179327036886e-06,
+      "loss": 0.2878,
+      "step": 4415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7794214530169473e-06,
+      "loss": 0.2834,
+      "step": 4416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776027895770983e-06,
+      "loss": 0.2844,
+      "step": 4417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7726372621722154e-06,
+      "loss": 0.2831,
+      "step": 4418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.76924955342602e-06,
+      "loss": 0.2728,
+      "step": 4419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7658647707367438e-06,
+      "loss": 0.2801,
+      "step": 4420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7624829153076828e-06,
+      "loss": 0.2731,
+      "step": 4421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7591039883411009e-06,
+      "loss": 0.2768,
+      "step": 4422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7557279910382151e-06,
+      "loss": 0.264,
+      "step": 4423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7523549245992022e-06,
+      "loss": 0.2788,
+      "step": 4424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7489847902231993e-06,
+      "loss": 0.2853,
+      "step": 4425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7456175891082972e-06,
+      "loss": 0.2775,
+      "step": 4426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7422533224515447e-06,
+      "loss": 0.2731,
+      "step": 4427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7388919914489544e-06,
+      "loss": 0.259,
+      "step": 4428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7355335972954834e-06,
+      "loss": 0.2776,
+      "step": 4429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.732178141185057e-06,
+      "loss": 0.2928,
+      "step": 4430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7288256243105484e-06,
+      "loss": 0.2753,
+      "step": 4431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7254760478637833e-06,
+      "loss": 0.2616,
+      "step": 4432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7221294130355515e-06,
+      "loss": 0.2662,
+      "step": 4433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.718785721015591e-06,
+      "loss": 0.276,
+      "step": 4434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7154449729925937e-06,
+      "loss": 0.2776,
+      "step": 4435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7121071701542058e-06,
+      "loss": 0.2833,
+      "step": 4436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7087723136870239e-06,
+      "loss": 0.2725,
+      "step": 4437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7054404047766037e-06,
+      "loss": 0.2801,
+      "step": 4438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7021114446074471e-06,
+      "loss": 0.2708,
+      "step": 4439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6987854343630083e-06,
+      "loss": 0.2728,
+      "step": 4440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6954623752256971e-06,
+      "loss": 0.2972,
+      "step": 4441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6921422683768707e-06,
+      "loss": 0.273,
+      "step": 4442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6888251149968327e-06,
+      "loss": 0.2932,
+      "step": 4443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6855109162648475e-06,
+      "loss": 0.277,
+      "step": 4444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6821996733591205e-06,
+      "loss": 0.2819,
+      "step": 4445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.678891387456808e-06,
+      "loss": 0.2722,
+      "step": 4446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6755860597340157e-06,
+      "loss": 0.2713,
+      "step": 4447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.672283691365796e-06,
+      "loss": 0.2726,
+      "step": 4448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668984283526155e-06,
+      "loss": 0.2747,
+      "step": 4449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6656878373880415e-06,
+      "loss": 0.2915,
+      "step": 4450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6623943541233466e-06,
+      "loss": 0.2802,
+      "step": 4451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.659103834902922e-06,
+      "loss": 0.2747,
+      "step": 4452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6558162808965495e-06,
+      "loss": 0.2791,
+      "step": 4453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6525316932729718e-06,
+      "loss": 0.275,
+      "step": 4454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6492500731998651e-06,
+      "loss": 0.2582,
+      "step": 4455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6459714218438582e-06,
+      "loss": 0.2686,
+      "step": 4456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.642695740370519e-06,
+      "loss": 0.2852,
+      "step": 4457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6394230299443614e-06,
+      "loss": 0.2693,
+      "step": 4458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636153291728848e-06,
+      "loss": 0.2636,
+      "step": 4459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6328865268863792e-06,
+      "loss": 0.2628,
+      "step": 4460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6296227365782968e-06,
+      "loss": 0.2703,
+      "step": 4461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.626361921964893e-06,
+      "loss": 0.259,
+      "step": 4462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6231040842053947e-06,
+      "loss": 0.2645,
+      "step": 4463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6198492244579723e-06,
+      "loss": 0.2873,
+      "step": 4464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.616597343879741e-06,
+      "loss": 0.2769,
+      "step": 4465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6133484436267533e-06,
+      "loss": 0.2633,
+      "step": 4466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6101025248540024e-06,
+      "loss": 0.2699,
+      "step": 4467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6068595887154226e-06,
+      "loss": 0.2698,
+      "step": 4468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6036196363638867e-06,
+      "loss": 0.2723,
+      "step": 4469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6003826689512093e-06,
+      "loss": 0.2645,
+      "step": 4470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5971486876281427e-06,
+      "loss": 0.2619,
+      "step": 4471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5939176935443735e-06,
+      "loss": 0.292,
+      "step": 4472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5906896878485346e-06,
+      "loss": 0.2701,
+      "step": 4473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.2775,
+      "step": 4474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5842426462098392e-06,
+      "loss": 0.2809,
+      "step": 4475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5810236125589285e-06,
+      "loss": 0.2562,
+      "step": 4476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5778075718798313e-06,
+      "loss": 0.2703,
+      "step": 4477
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5745945253158601e-06,
+      "loss": 0.2729,
+      "step": 4478
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.571384474009262e-06,
+      "loss": 0.2833,
+      "step": 4479
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5681774191012177e-06,
+      "loss": 0.2843,
+      "step": 4480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5649733617318507e-06,
+      "loss": 0.2704,
+      "step": 4481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5617723030402076e-06,
+      "loss": 0.2765,
+      "step": 4482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5585742441642792e-06,
+      "loss": 0.2757,
+      "step": 4483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5553791862409828e-06,
+      "loss": 0.2731,
+      "step": 4484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5521871304061697e-06,
+      "loss": 0.2713,
+      "step": 4485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.54899807779463e-06,
+      "loss": 0.2773,
+      "step": 4486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5458120295400792e-06,
+      "loss": 0.2762,
+      "step": 4487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5426289867751664e-06,
+      "loss": 0.2574,
+      "step": 4488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5394489506314746e-06,
+      "loss": 0.2793,
+      "step": 4489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5362719222395117e-06,
+      "loss": 0.2789,
+      "step": 4490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5330979027287285e-06,
+      "loss": 0.2711,
+      "step": 4491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5299268932274936e-06,
+      "loss": 0.2859,
+      "step": 4492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5267588948631095e-06,
+      "loss": 0.2718,
+      "step": 4493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5235939087618145e-06,
+      "loss": 0.2665,
+      "step": 4494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5204319360487674e-06,
+      "loss": 0.2796,
+      "step": 4495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5172729778480576e-06,
+      "loss": 0.2572,
+      "step": 4496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5141170352827094e-06,
+      "loss": 0.2648,
+      "step": 4497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5109641094746664e-06,
+      "loss": 0.2751,
+      "step": 4498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507814201544804e-06,
+      "loss": 0.2726,
+      "step": 4499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.504667312612924e-06,
+      "loss": 0.2774,
+      "step": 4500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.501523443797752e-06,
+      "loss": 0.2563,
+      "step": 4501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4983825962169496e-06,
+      "loss": 0.2696,
+      "step": 4502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4952447709870944e-06,
+      "loss": 0.2687,
+      "step": 4503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4921099692236895e-06,
+      "loss": 0.268,
+      "step": 4504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4889781920411727e-06,
+      "loss": 0.2675,
+      "step": 4505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4858494405528944e-06,
+      "loss": 0.2709,
+      "step": 4506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4827237158711405e-06,
+      "loss": 0.2638,
+      "step": 4507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.479601019107113e-06,
+      "loss": 0.2734,
+      "step": 4508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.476481351370942e-06,
+      "loss": 0.2722,
+      "step": 4509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4733647137716756e-06,
+      "loss": 0.2687,
+      "step": 4510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4702511074172888e-06,
+      "loss": 0.285,
+      "step": 4511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4671405334146805e-06,
+      "loss": 0.2777,
+      "step": 4512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4640329928696683e-06,
+      "loss": 0.2733,
+      "step": 4513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4609284868869899e-06,
+      "loss": 0.2665,
+      "step": 4514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4578270165703113e-06,
+      "loss": 0.2734,
+      "step": 4515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4547285830222114e-06,
+      "loss": 0.2717,
+      "step": 4516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4516331873441914e-06,
+      "loss": 0.273,
+      "step": 4517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4485408306366787e-06,
+      "loss": 0.2672,
+      "step": 4518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4454515139990122e-06,
+      "loss": 0.2723,
+      "step": 4519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4423652385294539e-06,
+      "loss": 0.2613,
+      "step": 4520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4392820053251855e-06,
+      "loss": 0.2754,
+      "step": 4521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.436201815482301e-06,
+      "loss": 0.2683,
+      "step": 4522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4331246700958234e-06,
+      "loss": 0.2714,
+      "step": 4523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4300505702596856e-06,
+      "loss": 0.2848,
+      "step": 4524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4269795170667356e-06,
+      "loss": 0.2816,
+      "step": 4525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4239115116087488e-06,
+      "loss": 0.27,
+      "step": 4526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4208465549764027e-06,
+      "loss": 0.2827,
+      "step": 4527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.417784648259307e-06,
+      "loss": 0.2731,
+      "step": 4528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4147257925459745e-06,
+      "loss": 0.2829,
+      "step": 4529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4116699889238372e-06,
+      "loss": 0.2805,
+      "step": 4530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.408617238479244e-06,
+      "loss": 0.2692,
+      "step": 4531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.405567542297457e-06,
+      "loss": 0.263,
+      "step": 4532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4025209014626494e-06,
+      "loss": 0.28,
+      "step": 4533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3994773170579168e-06,
+      "loss": 0.269,
+      "step": 4534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3964367901652575e-06,
+      "loss": 0.2695,
+      "step": 4535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3933993218655928e-06,
+      "loss": 0.2689,
+      "step": 4536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3903649132387487e-06,
+      "loss": 0.2766,
+      "step": 4537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3873335653634669e-06,
+      "loss": 0.2709,
+      "step": 4538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3843052793174027e-06,
+      "loss": 0.2605,
+      "step": 4539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.381280056177121e-06,
+      "loss": 0.2878,
+      "step": 4540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.378257897018096e-06,
+      "loss": 0.2636,
+      "step": 4541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3752388029147157e-06,
+      "loss": 0.266,
+      "step": 4542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.372222774940275e-06,
+      "loss": 0.2618,
+      "step": 4543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3692098141669851e-06,
+      "loss": 0.2625,
+      "step": 4544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3661999216659606e-06,
+      "loss": 0.2714,
+      "step": 4545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.363193098507225e-06,
+      "loss": 0.2564,
+      "step": 4546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3601893457597182e-06,
+      "loss": 0.2713,
+      "step": 4547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3571886644912802e-06,
+      "loss": 0.2816,
+      "step": 4548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3541910557686622e-06,
+      "loss": 0.278,
+      "step": 4549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.351196520657525e-06,
+      "loss": 0.272,
+      "step": 4550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.348205060222435e-06,
+      "loss": 0.2657,
+      "step": 4551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3452166755268637e-06,
+      "loss": 0.272,
+      "step": 4552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.342231367633192e-06,
+      "loss": 0.2808,
+      "step": 4553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339249137602705e-06,
+      "loss": 0.2799,
+      "step": 4554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3362699864955964e-06,
+      "loss": 0.2725,
+      "step": 4555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3332939153709634e-06,
+      "loss": 0.2679,
+      "step": 4556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3303209252868043e-06,
+      "loss": 0.2716,
+      "step": 4557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3273510173000325e-06,
+      "loss": 0.2778,
+      "step": 4558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3243841924664536e-06,
+      "loss": 0.278,
+      "step": 4559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3214204518407869e-06,
+      "loss": 0.2607,
+      "step": 4560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.318459796476651e-06,
+      "loss": 0.2759,
+      "step": 4561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3155022274265671e-06,
+      "loss": 0.2649,
+      "step": 4562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.312547745741959e-06,
+      "loss": 0.2684,
+      "step": 4563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3095963524731526e-06,
+      "loss": 0.2804,
+      "step": 4564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306648048669381e-06,
+      "loss": 0.2581,
+      "step": 4565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303702835378774e-06,
+      "loss": 0.2799,
+      "step": 4566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3007607136483614e-06,
+      "loss": 0.2716,
+      "step": 4567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297821684524081e-06,
+      "loss": 0.2777,
+      "step": 4568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.294885749050764e-06,
+      "loss": 0.2692,
+      "step": 4569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2919529082721417e-06,
+      "loss": 0.2781,
+      "step": 4570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2890231632308526e-06,
+      "loss": 0.2874,
+      "step": 4571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.286096514968429e-06,
+      "loss": 0.2824,
+      "step": 4572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2831729645253021e-06,
+      "loss": 0.2651,
+      "step": 4573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2802525129408039e-06,
+      "loss": 0.2597,
+      "step": 4574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2773351612531614e-06,
+      "loss": 0.2782,
+      "step": 4575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2744209104995054e-06,
+      "loss": 0.2795,
+      "step": 4576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2715097617158601e-06,
+      "loss": 0.2855,
+      "step": 4577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2686017159371443e-06,
+      "loss": 0.2621,
+      "step": 4578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2656967741971837e-06,
+      "loss": 0.2852,
+      "step": 4579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2627949375286864e-06,
+      "loss": 0.2733,
+      "step": 4580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2598962069632715e-06,
+      "loss": 0.2701,
+      "step": 4581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2570005835314425e-06,
+      "loss": 0.2674,
+      "step": 4582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.254108068262604e-06,
+      "loss": 0.2689,
+      "step": 4583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.251218662185052e-06,
+      "loss": 0.2568,
+      "step": 4584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2483323663259805e-06,
+      "loss": 0.2695,
+      "step": 4585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2454491817114744e-06,
+      "loss": 0.2657,
+      "step": 4586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2425691093665182e-06,
+      "loss": 0.2516,
+      "step": 4587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2396921503149839e-06,
+      "loss": 0.277,
+      "step": 4588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2368183055796411e-06,
+      "loss": 0.2707,
+      "step": 4589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2339475761821485e-06,
+      "loss": 0.2889,
+      "step": 4590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2310799631430592e-06,
+      "loss": 0.2615,
+      "step": 4591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2282154674818203e-06,
+      "loss": 0.2645,
+      "step": 4592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2253540902167683e-06,
+      "loss": 0.2803,
+      "step": 4593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2224958323651304e-06,
+      "loss": 0.2903,
+      "step": 4594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2196406949430274e-06,
+      "loss": 0.2851,
+      "step": 4595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2167886789654659e-06,
+      "loss": 0.2712,
+      "step": 4596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2139397854463497e-06,
+      "loss": 0.2782,
+      "step": 4597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2110940153984696e-06,
+      "loss": 0.2739,
+      "step": 4598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2082513698335018e-06,
+      "loss": 0.2644,
+      "step": 4599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2054118497620205e-06,
+      "loss": 0.2743,
+      "step": 4600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.20257545619348e-06,
+      "loss": 0.2619,
+      "step": 4601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1997421901362271e-06,
+      "loss": 0.2727,
+      "step": 4602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1969120525974988e-06,
+      "loss": 0.2704,
+      "step": 4603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1940850445834163e-06,
+      "loss": 0.2596,
+      "step": 4604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1912611670989904e-06,
+      "loss": 0.2659,
+      "step": 4605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188440421148116e-06,
+      "loss": 0.2762,
+      "step": 4606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1856228077335775e-06,
+      "loss": 0.2793,
+      "step": 4607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1828083278570467e-06,
+      "loss": 0.2711,
+      "step": 4608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1799969825190794e-06,
+      "loss": 0.2631,
+      "step": 4609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1771887727191156e-06,
+      "loss": 0.2665,
+      "step": 4610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1743836994554848e-06,
+      "loss": 0.2806,
+      "step": 4611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1715817637253968e-06,
+      "loss": 0.2723,
+      "step": 4612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1687829665249505e-06,
+      "loss": 0.2724,
+      "step": 4613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1659873088491269e-06,
+      "loss": 0.2699,
+      "step": 4614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1631947916917906e-06,
+      "loss": 0.2662,
+      "step": 4615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.160405416045689e-06,
+      "loss": 0.2633,
+      "step": 4616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157619182902453e-06,
+      "loss": 0.28,
+      "step": 4617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154836093252597e-06,
+      "loss": 0.2849,
+      "step": 4618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1520561480855197e-06,
+      "loss": 0.2703,
+      "step": 4619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1492793483894981e-06,
+      "loss": 0.2703,
+      "step": 4620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1465056951516962e-06,
+      "loss": 0.2626,
+      "step": 4621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1437351893581526e-06,
+      "loss": 0.2756,
+      "step": 4622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1409678319937923e-06,
+      "loss": 0.2745,
+      "step": 4623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1382036240424198e-06,
+      "loss": 0.2813,
+      "step": 4624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1354425664867187e-06,
+      "loss": 0.2836,
+      "step": 4625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.132684660308253e-06,
+      "loss": 0.2565,
+      "step": 4626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1299299064874681e-06,
+      "loss": 0.2809,
+      "step": 4627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1271783060036844e-06,
+      "loss": 0.2663,
+      "step": 4628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1244298598351078e-06,
+      "loss": 0.269,
+      "step": 4629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1216845689588173e-06,
+      "loss": 0.269,
+      "step": 4630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1189424343507716e-06,
+      "loss": 0.2723,
+      "step": 4631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1162034569858115e-06,
+      "loss": 0.2703,
+      "step": 4632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1134676378376464e-06,
+      "loss": 0.2661,
+      "step": 4633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1107349778788746e-06,
+      "loss": 0.2684,
+      "step": 4634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1080054780809602e-06,
+      "loss": 0.2734,
+      "step": 4635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1052791394142515e-06,
+      "loss": 0.2683,
+      "step": 4636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1025559628479687e-06,
+      "loss": 0.2725,
+      "step": 4637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0998359493502097e-06,
+      "loss": 0.2847,
+      "step": 4638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0971190998879455e-06,
+      "loss": 0.278,
+      "step": 4639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094405415427029e-06,
+      "loss": 0.2622,
+      "step": 4640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0916948969321784e-06,
+      "loss": 0.2759,
+      "step": 4641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0889875453669951e-06,
+      "loss": 0.2682,
+      "step": 4642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0862833616939506e-06,
+      "loss": 0.2771,
+      "step": 4643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0835823468743866e-06,
+      "loss": 0.278,
+      "step": 4644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0808845018685265e-06,
+      "loss": 0.2703,
+      "step": 4645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0781898276354619e-06,
+      "loss": 0.297,
+      "step": 4646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0754983251331551e-06,
+      "loss": 0.2565,
+      "step": 4647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0728099953184468e-06,
+      "loss": 0.2766,
+      "step": 4648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0701248391470408e-06,
+      "loss": 0.2831,
+      "step": 4649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0674428575735252e-06,
+      "loss": 0.2578,
+      "step": 4650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.064764051551349e-06,
+      "loss": 0.256,
+      "step": 4651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0620884220328353e-06,
+      "loss": 0.2442,
+      "step": 4652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0594159699691808e-06,
+      "loss": 0.2759,
+      "step": 4653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.05674669631045e-06,
+      "loss": 0.2664,
+      "step": 4654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0540806020055749e-06,
+      "loss": 0.2716,
+      "step": 4655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0514176880023651e-06,
+      "loss": 0.2684,
+      "step": 4656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0487579552474915e-06,
+      "loss": 0.2818,
+      "step": 4657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.046101404686497e-06,
+      "loss": 0.2844,
+      "step": 4658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0434480372637946e-06,
+      "loss": 0.2681,
+      "step": 4659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0407978539226626e-06,
+      "loss": 0.2755,
+      "step": 4660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0381508556052523e-06,
+      "loss": 0.2787,
+      "step": 4661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.035507043252577e-06,
+      "loss": 0.2791,
+      "step": 4662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0328664178045211e-06,
+      "loss": 0.2767,
+      "step": 4663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0302289801998356e-06,
+      "loss": 0.2784,
+      "step": 4664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.027594731376136e-06,
+      "loss": 0.2698,
+      "step": 4665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0249636722699086e-06,
+      "loss": 0.2658,
+      "step": 4666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.022335803816501e-06,
+      "loss": 0.2638,
+      "step": 4667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0197111269501292e-06,
+      "loss": 0.2715,
+      "step": 4668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0170896426038734e-06,
+      "loss": 0.2706,
+      "step": 4669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0144713517096782e-06,
+      "loss": 0.2789,
+      "step": 4670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0118562551983546e-06,
+      "loss": 0.262,
+      "step": 4671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0092443539995788e-06,
+      "loss": 0.2656,
+      "step": 4672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0066356490418871e-06,
+      "loss": 0.2667,
+      "step": 4673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0040301412526854e-06,
+      "loss": 0.2713,
+      "step": 4674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0014278315582382e-06,
+      "loss": 0.2839,
+      "step": 4675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.988287208836723e-07,
+      "loss": 0.2906,
+      "step": 4676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.962328101529827e-07,
+      "loss": 0.2798,
+      "step": 4677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.936401002890228e-07,
+      "loss": 0.2789,
+      "step": 4678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.910505922135093e-07,
+      "loss": 0.2609,
+      "step": 4679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.884642868470185e-07,
+      "loss": 0.2833,
+      "step": 4680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85881185108989e-07,
+      "loss": 0.2618,
+      "step": 4681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.833012879177251e-07,
+      "loss": 0.2672,
+      "step": 4682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.807245961903855e-07,
+      "loss": 0.2812,
+      "step": 4683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.781511108429908e-07,
+      "loss": 0.2879,
+      "step": 4684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.75580832790427e-07,
+      "loss": 0.2766,
+      "step": 4685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.73013762946432e-07,
+      "loss": 0.258,
+      "step": 4686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.704499022236102e-07,
+      "loss": 0.2606,
+      "step": 4687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.678892515334194e-07,
+      "loss": 0.2629,
+      "step": 4688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.653318117861798e-07,
+      "loss": 0.2731,
+      "step": 4689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.627775838910691e-07,
+      "loss": 0.2667,
+      "step": 4690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.60226568756122e-07,
+      "loss": 0.2851,
+      "step": 4691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.576787672882315e-07,
+      "loss": 0.2626,
+      "step": 4692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.551341803931523e-07,
+      "loss": 0.2684,
+      "step": 4693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.525928089754888e-07,
+      "loss": 0.2729,
+      "step": 4694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.500546539387101e-07,
+      "loss": 0.2739,
+      "step": 4695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.475197161851358e-07,
+      "loss": 0.2637,
+      "step": 4696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.449879966159426e-07,
+      "loss": 0.283,
+      "step": 4697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.424594961311695e-07,
+      "loss": 0.2676,
+      "step": 4698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.39934215629702e-07,
+      "loss": 0.2591,
+      "step": 4699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.374121560092852e-07,
+      "loss": 0.2664,
+      "step": 4700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.34893318166521e-07,
+      "loss": 0.2748,
+      "step": 4701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.323777029968605e-07,
+      "loss": 0.2729,
+      "step": 4702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.298653113946166e-07,
+      "loss": 0.2773,
+      "step": 4703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.273561442529511e-07,
+      "loss": 0.2653,
+      "step": 4704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.248502024638773e-07,
+      "loss": 0.2624,
+      "step": 4705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.223474869182691e-07,
+      "loss": 0.2885,
+      "step": 4706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.198479985058495e-07,
+      "loss": 0.284,
+      "step": 4707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.173517381151908e-07,
+      "loss": 0.2626,
+      "step": 4708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.148587066337245e-07,
+      "loss": 0.2622,
+      "step": 4709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.123689049477302e-07,
+      "loss": 0.283,
+      "step": 4710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.09882333942339e-07,
+      "loss": 0.2675,
+      "step": 4711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.07398994501536e-07,
+      "loss": 0.27,
+      "step": 4712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.049188875081527e-07,
+      "loss": 0.273,
+      "step": 4713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.024420138438784e-07,
+      "loss": 0.2671,
+      "step": 4714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.999683743892485e-07,
+      "loss": 0.2522,
+      "step": 4715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.974979700236463e-07,
+      "loss": 0.2702,
+      "step": 4716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.950308016253129e-07,
+      "loss": 0.272,
+      "step": 4717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.925668700713297e-07,
+      "loss": 0.28,
+      "step": 4718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.901061762376362e-07,
+      "loss": 0.2653,
+      "step": 4719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.876487209990136e-07,
+      "loss": 0.2812,
+      "step": 4720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.851945052290967e-07,
+      "loss": 0.2686,
+      "step": 4721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.827435298003639e-07,
+      "loss": 0.2678,
+      "step": 4722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.802957955841473e-07,
+      "loss": 0.2672,
+      "step": 4723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.778513034506208e-07,
+      "loss": 0.2684,
+      "step": 4724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.754100542688126e-07,
+      "loss": 0.2714,
+      "step": 4725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.729720489065896e-07,
+      "loss": 0.2935,
+      "step": 4726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.70537288230675e-07,
+      "loss": 0.2798,
+      "step": 4727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.681057731066322e-07,
+      "loss": 0.2637,
+      "step": 4728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.656775043988697e-07,
+      "loss": 0.267,
+      "step": 4729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.632524829706468e-07,
+      "loss": 0.2814,
+      "step": 4730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.608307096840673e-07,
+      "loss": 0.2823,
+      "step": 4731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.58412185400076e-07,
+      "loss": 0.2682,
+      "step": 4732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.559969109784671e-07,
+      "loss": 0.2646,
+      "step": 4733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.535848872778762e-07,
+      "loss": 0.2753,
+      "step": 4734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.511761151557885e-07,
+      "loss": 0.2725,
+      "step": 4735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.487705954685288e-07,
+      "loss": 0.2721,
+      "step": 4736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.463683290712632e-07,
+      "loss": 0.2768,
+      "step": 4737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.439693168180096e-07,
+      "loss": 0.2767,
+      "step": 4738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.415735595616204e-07,
+      "loss": 0.277,
+      "step": 4739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.391810581537973e-07,
+      "loss": 0.2889,
+      "step": 4740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.367918134450814e-07,
+      "loss": 0.2659,
+      "step": 4741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.344058262848564e-07,
+      "loss": 0.2728,
+      "step": 4742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.320230975213473e-07,
+      "loss": 0.2639,
+      "step": 4743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.296436280016218e-07,
+      "loss": 0.2733,
+      "step": 4744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.272674185715868e-07,
+      "loss": 0.2613,
+      "step": 4745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.248944700759954e-07,
+      "loss": 0.2651,
+      "step": 4746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.225247833584338e-07,
+      "loss": 0.2654,
+      "step": 4747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.201583592613383e-07,
+      "loss": 0.2615,
+      "step": 4748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.177951986259769e-07,
+      "loss": 0.2746,
+      "step": 4749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.154353022924577e-07,
+      "loss": 0.2764,
+      "step": 4750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.130786710997362e-07,
+      "loss": 0.2817,
+      "step": 4751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.107253058856024e-07,
+      "loss": 0.2663,
+      "step": 4752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.083752074866769e-07,
+      "loss": 0.2683,
+      "step": 4753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.060283767384347e-07,
+      "loss": 0.2737,
+      "step": 4754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.036848144751774e-07,
+      "loss": 0.2673,
+      "step": 4755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.013445215300509e-07,
+      "loss": 0.2731,
+      "step": 4756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.99007498735035e-07,
+      "loss": 0.3008,
+      "step": 4757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.966737469209474e-07,
+      "loss": 0.2675,
+      "step": 4758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.943432669174477e-07,
+      "loss": 0.2563,
+      "step": 4759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.920160595530257e-07,
+      "loss": 0.2816,
+      "step": 4760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.896921256550094e-07,
+      "loss": 0.2858,
+      "step": 4761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.873714660495702e-07,
+      "loss": 0.2641,
+      "step": 4762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.850540815617025e-07,
+      "loss": 0.2733,
+      "step": 4763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.827399730152485e-07,
+      "loss": 0.269,
+      "step": 4764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.804291412328802e-07,
+      "loss": 0.2629,
+      "step": 4765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.781215870361014e-07,
+      "loss": 0.28,
+      "step": 4766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.758173112452605e-07,
+      "loss": 0.2778,
+      "step": 4767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.73516314679531e-07,
+      "loss": 0.2676,
+      "step": 4768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.712185981569232e-07,
+      "loss": 0.2618,
+      "step": 4769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.689241624942856e-07,
+      "loss": 0.2573,
+      "step": 4770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.666330085072948e-07,
+      "loss": 0.2778,
+      "step": 4771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.643451370104649e-07,
+      "loss": 0.2994,
+      "step": 4772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.620605488171418e-07,
+      "loss": 0.2606,
+      "step": 4773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.597792447394992e-07,
+      "loss": 0.284,
+      "step": 4774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.575012255885517e-07,
+      "loss": 0.2693,
+      "step": 4775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.552264921741404e-07,
+      "loss": 0.2696,
+      "step": 4776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.529550453049395e-07,
+      "loss": 0.2853,
+      "step": 4777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.506868857884564e-07,
+      "loss": 0.2763,
+      "step": 4778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.484220144310283e-07,
+      "loss": 0.2722,
+      "step": 4779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.461604320378247e-07,
+      "loss": 0.2613,
+      "step": 4780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.439021394128443e-07,
+      "loss": 0.2618,
+      "step": 4781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.416471373589163e-07,
+      "loss": 0.2662,
+      "step": 4782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.393954266777048e-07,
+      "loss": 0.2711,
+      "step": 4783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.371470081696929e-07,
+      "loss": 0.2584,
+      "step": 4784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.349018826342059e-07,
+      "loss": 0.2746,
+      "step": 4785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.326600508693904e-07,
+      "loss": 0.2699,
+      "step": 4786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.304215136722237e-07,
+      "loss": 0.2786,
+      "step": 4787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281862718385157e-07,
+      "loss": 0.2756,
+      "step": 4788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.259543261628988e-07,
+      "loss": 0.2585,
+      "step": 4789
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.237256774388368e-07,
+      "loss": 0.2549,
+      "step": 4790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.21500326458624e-07,
+      "loss": 0.2638,
+      "step": 4791
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.192782740133741e-07,
+      "loss": 0.2681,
+      "step": 4792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.170595208930409e-07,
+      "loss": 0.2574,
+      "step": 4793
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.148440678863921e-07,
+      "loss": 0.2758,
+      "step": 4794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.12631915781028e-07,
+      "loss": 0.283,
+      "step": 4795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.104230653633793e-07,
+      "loss": 0.2706,
+      "step": 4796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.082175174186967e-07,
+      "loss": 0.2547,
+      "step": 4797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.060152727310565e-07,
+      "loss": 0.285,
+      "step": 4798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.038163320833691e-07,
+      "loss": 0.274,
+      "step": 4799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.016206962573602e-07,
+      "loss": 0.2786,
+      "step": 4800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.994283660335876e-07,
+      "loss": 0.2756,
+      "step": 4801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.972393421914314e-07,
+      "loss": 0.2598,
+      "step": 4802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.950536255090934e-07,
+      "loss": 0.2726,
+      "step": 4803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.928712167636076e-07,
+      "loss": 0.2656,
+      "step": 4804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.906921167308212e-07,
+      "loss": 0.2629,
+      "step": 4805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.885163261854133e-07,
+      "loss": 0.2711,
+      "step": 4806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.863438459008853e-07,
+      "loss": 0.2598,
+      "step": 4807
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.841746766495583e-07,
+      "loss": 0.266,
+      "step": 4808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.820088192025809e-07,
+      "loss": 0.2619,
+      "step": 4809
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.798462743299216e-07,
+      "loss": 0.2575,
+      "step": 4810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.776870428003701e-07,
+      "loss": 0.2803,
+      "step": 4811
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.755311253815411e-07,
+      "loss": 0.262,
+      "step": 4812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.733785228398703e-07,
+      "loss": 0.2829,
+      "step": 4813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.712292359406137e-07,
+      "loss": 0.2675,
+      "step": 4814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.690832654478497e-07,
+      "loss": 0.2881,
+      "step": 4815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.669406121244748e-07,
+      "loss": 0.2737,
+      "step": 4816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.648012767322131e-07,
+      "loss": 0.2802,
+      "step": 4817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.62665260031603e-07,
+      "loss": 0.2696,
+      "step": 4818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.605325627820036e-07,
+      "loss": 0.2569,
+      "step": 4819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.584031857415974e-07,
+      "loss": 0.2854,
+      "step": 4820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.562771296673831e-07,
+      "loss": 0.2742,
+      "step": 4821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.541543953151797e-07,
+      "loss": 0.2622,
+      "step": 4822
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.520349834396278e-07,
+      "loss": 0.2714,
+      "step": 4823
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.499188947941836e-07,
+      "loss": 0.265,
+      "step": 4824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.478061301311234e-07,
+      "loss": 0.2941,
+      "step": 4825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.456966902015405e-07,
+      "loss": 0.2649,
+      "step": 4826
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.435905757553463e-07,
+      "loss": 0.2609,
+      "step": 4827
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.414877875412751e-07,
+      "loss": 0.2659,
+      "step": 4828
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.393883263068712e-07,
+      "loss": 0.2733,
+      "step": 4829
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.372921927984988e-07,
+      "loss": 0.2585,
+      "step": 4830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.351993877613427e-07,
+      "loss": 0.2798,
+      "step": 4831
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.33109911939398e-07,
+      "loss": 0.2773,
+      "step": 4832
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.310237660754837e-07,
+      "loss": 0.2646,
+      "step": 4833
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.289409509112288e-07,
+      "loss": 0.2601,
+      "step": 4834
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.26861467187081e-07,
+      "loss": 0.2735,
+      "step": 4835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.247853156423022e-07,
+      "loss": 0.2559,
+      "step": 4836
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2271249701497e-07,
+      "loss": 0.2817,
+      "step": 4837
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.206430120419804e-07,
+      "loss": 0.2806,
+      "step": 4838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.185768614590404e-07,
+      "loss": 0.2738,
+      "step": 4839
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.165140460006702e-07,
+      "loss": 0.2694,
+      "step": 4840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.144545664002122e-07,
+      "loss": 0.2686,
+      "step": 4841
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.123984233898139e-07,
+      "loss": 0.2631,
+      "step": 4842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.103456177004408e-07,
+      "loss": 0.2802,
+      "step": 4843
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.082961500618722e-07,
+      "loss": 0.267,
+      "step": 4844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.062500212027011e-07,
+      "loss": 0.275,
+      "step": 4845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.042072318503323e-07,
+      "loss": 0.2614,
+      "step": 4846
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.021677827309824e-07,
+      "loss": 0.2676,
+      "step": 4847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.001316745696806e-07,
+      "loss": 0.269,
+      "step": 4848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.980989080902721e-07,
+      "loss": 0.2625,
+      "step": 4849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.960694840154113e-07,
+      "loss": 0.2899,
+      "step": 4850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.940434030665632e-07,
+      "loss": 0.2734,
+      "step": 4851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.92020665964007e-07,
+      "loss": 0.2836,
+      "step": 4852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.900012734268313e-07,
+      "loss": 0.2783,
+      "step": 4853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.87985226172938e-07,
+      "loss": 0.2798,
+      "step": 4854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.859725249190362e-07,
+      "loss": 0.2713,
+      "step": 4855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.83963170380647e-07,
+      "loss": 0.2712,
+      "step": 4856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.819571632721044e-07,
+      "loss": 0.2623,
+      "step": 4857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.79954504306548e-07,
+      "loss": 0.2731,
+      "step": 4858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.779551941959272e-07,
+      "loss": 0.277,
+      "step": 4859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.759592336510078e-07,
+      "loss": 0.2563,
+      "step": 4860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.739666233813568e-07,
+      "loss": 0.2788,
+      "step": 4861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.719773640953541e-07,
+      "loss": 0.2533,
+      "step": 4862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.699914565001874e-07,
+      "loss": 0.2791,
+      "step": 4863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.680089013018508e-07,
+      "loss": 0.2626,
+      "step": 4864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.660296992051529e-07,
+      "loss": 0.277,
+      "step": 4865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.640538509137039e-07,
+      "loss": 0.2654,
+      "step": 4866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.620813571299233e-07,
+      "loss": 0.2681,
+      "step": 4867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6011221855504e-07,
+      "loss": 0.2692,
+      "step": 4868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.581464358890864e-07,
+      "loss": 0.2824,
+      "step": 4869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.561840098309068e-07,
+      "loss": 0.2655,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.542249410781486e-07,
+      "loss": 0.2759,
+      "step": 4871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.522692303272659e-07,
+      "loss": 0.2869,
+      "step": 4872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.503168782735213e-07,
+      "loss": 0.2803,
+      "step": 4873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.48367885610982e-07,
+      "loss": 0.2729,
+      "step": 4874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.464222530325192e-07,
+      "loss": 0.2661,
+      "step": 4875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.444799812298129e-07,
+      "loss": 0.2819,
+      "step": 4876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.425410708933454e-07,
+      "loss": 0.2763,
+      "step": 4877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.406055227124074e-07,
+      "loss": 0.2704,
+      "step": 4878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.386733373750897e-07,
+      "loss": 0.2619,
+      "step": 4879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.367445155682893e-07,
+      "loss": 0.2881,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.348190579777124e-07,
+      "loss": 0.2684,
+      "step": 4881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.328969652878624e-07,
+      "loss": 0.27,
+      "step": 4882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.309782381820472e-07,
+      "loss": 0.2632,
+      "step": 4883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.290628773423834e-07,
+      "loss": 0.2775,
+      "step": 4884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.271508834497863e-07,
+      "loss": 0.2557,
+      "step": 4885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.252422571839766e-07,
+      "loss": 0.2644,
+      "step": 4886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.23336999223476e-07,
+      "loss": 0.2729,
+      "step": 4887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.214351102456095e-07,
+      "loss": 0.2539,
+      "step": 4888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.195365909265037e-07,
+      "loss": 0.2768,
+      "step": 4889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.17641441941088e-07,
+      "loss": 0.257,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.157496639630955e-07,
+      "loss": 0.286,
+      "step": 4891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.138612576650581e-07,
+      "loss": 0.2736,
+      "step": 4892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.119762237183079e-07,
+      "loss": 0.2783,
+      "step": 4893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.100945627929843e-07,
+      "loss": 0.2693,
+      "step": 4894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.082162755580211e-07,
+      "loss": 0.2659,
+      "step": 4895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.063413626811542e-07,
+      "loss": 0.2621,
+      "step": 4896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.044698248289225e-07,
+      "loss": 0.278,
+      "step": 4897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.026016626666641e-07,
+      "loss": 0.265,
+      "step": 4898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.007368768585141e-07,
+      "loss": 0.2822,
+      "step": 4899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.988754680674113e-07,
+      "loss": 0.2766,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.970174369550906e-07,
+      "loss": 0.2493,
+      "step": 4901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.951627841820894e-07,
+      "loss": 0.257,
+      "step": 4902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.933115104077413e-07,
+      "loss": 0.2592,
+      "step": 4903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.914636162901799e-07,
+      "loss": 0.2638,
+      "step": 4904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896191024863384e-07,
+      "loss": 0.2731,
+      "step": 4905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.877779696519447e-07,
+      "loss": 0.2816,
+      "step": 4906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.859402184415307e-07,
+      "loss": 0.2542,
+      "step": 4907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.841058495084194e-07,
+      "loss": 0.2557,
+      "step": 4908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.82274863504737e-07,
+      "loss": 0.2746,
+      "step": 4909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.804472610814015e-07,
+      "loss": 0.2659,
+      "step": 4910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.786230428881345e-07,
+      "loss": 0.261,
+      "step": 4911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7680220957344814e-07,
+      "loss": 0.27,
+      "step": 4912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.749847617846559e-07,
+      "loss": 0.2712,
+      "step": 4913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7317070016786535e-07,
+      "loss": 0.2676,
+      "step": 4914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7136002536798176e-07,
+      "loss": 0.2689,
+      "step": 4915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6955273802870573e-07,
+      "loss": 0.2713,
+      "step": 4916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.677488387925322e-07,
+      "loss": 0.2814,
+      "step": 4917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6594832830075375e-07,
+      "loss": 0.2747,
+      "step": 4918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.641512071934573e-07,
+      "loss": 0.2798,
+      "step": 4919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.623574761095251e-07,
+      "loss": 0.2703,
+      "step": 4920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.605671356866337e-07,
+      "loss": 0.2833,
+      "step": 4921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.587801865612529e-07,
+      "loss": 0.2782,
+      "step": 4922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.569966293686512e-07,
+      "loss": 0.2721,
+      "step": 4923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5521646474288803e-07,
+      "loss": 0.2732,
+      "step": 4924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.534396933168161e-07,
+      "loss": 0.272,
+      "step": 4925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.516663157220846e-07,
+      "loss": 0.2641,
+      "step": 4926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4989633258913366e-07,
+      "loss": 0.2752,
+      "step": 4927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4812974454719657e-07,
+      "loss": 0.2663,
+      "step": 4928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4636655222430436e-07,
+      "loss": 0.2708,
+      "step": 4929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4460675624727443e-07,
+      "loss": 0.2666,
+      "step": 4930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4285035724172086e-07,
+      "loss": 0.2709,
+      "step": 4931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.410973558320486e-07,
+      "loss": 0.2695,
+      "step": 4932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3934775264145246e-07,
+      "loss": 0.2648,
+      "step": 4933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3760154829192605e-07,
+      "loss": 0.2784,
+      "step": 4934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3585874340424825e-07,
+      "loss": 0.274,
+      "step": 4935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.341193385979903e-07,
+      "loss": 0.2665,
+      "step": 4936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.323833344915185e-07,
+      "loss": 0.2691,
+      "step": 4937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.306507317019837e-07,
+      "loss": 0.2742,
+      "step": 4938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.289215308453365e-07,
+      "loss": 0.2672,
+      "step": 4939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2719573253631076e-07,
+      "loss": 0.2802,
+      "step": 4940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.254733373884334e-07,
+      "loss": 0.2849,
+      "step": 4941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.237543460140192e-07,
+      "loss": 0.2835,
+      "step": 4942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.220387590241781e-07,
+      "loss": 0.2759,
+      "step": 4943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.203265770288023e-07,
+      "loss": 0.2681,
+      "step": 4944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.186178006365815e-07,
+      "loss": 0.2751,
+      "step": 4945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1691243045498874e-07,
+      "loss": 0.2678,
+      "step": 4946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1521046709029014e-07,
+      "loss": 0.2598,
+      "step": 4947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.135119111475383e-07,
+      "loss": 0.2618,
+      "step": 4948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1181676323057365e-07,
+      "loss": 0.2645,
+      "step": 4949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1012502394202735e-07,
+      "loss": 0.2833,
+      "step": 4950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.084366938833195e-07,
+      "loss": 0.2722,
+      "step": 4951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0675177365465444e-07,
+      "loss": 0.2817,
+      "step": 4952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.269,
+      "step": 4953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0339216508221813e-07,
+      "loss": 0.2745,
+      "step": 4954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0171747793279925e-07,
+      "loss": 0.2581,
+      "step": 4955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.000462030021257e-07,
+      "loss": 0.2593,
+      "step": 4956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9837834088434e-07,
+      "loss": 0.2742,
+      "step": 4957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.967138921723734e-07,
+      "loss": 0.2775,
+      "step": 4958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9505285745794155e-07,
+      "loss": 0.2682,
+      "step": 4959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.933952373315486e-07,
+      "loss": 0.2786,
+      "step": 4960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.917410323824833e-07,
+      "loss": 0.2724,
+      "step": 4961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.900902431988196e-07,
+      "loss": 0.2633,
+      "step": 4962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.884428703674181e-07,
+      "loss": 0.2579,
+      "step": 4963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.867989144739259e-07,
+      "loss": 0.254,
+      "step": 4964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.851583761027733e-07,
+      "loss": 0.2898,
+      "step": 4965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.83521255837177e-07,
+      "loss": 0.2701,
+      "step": 4966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8188755425913825e-07,
+      "loss": 0.2523,
+      "step": 4967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.802572719494446e-07,
+      "loss": 0.2902,
+      "step": 4968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.786304094876636e-07,
+      "loss": 0.26,
+      "step": 4969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.770069674521515e-07,
+      "loss": 0.2715,
+      "step": 4970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7538694642004656e-07,
+      "loss": 0.2761,
+      "step": 4971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7377034696727264e-07,
+      "loss": 0.2666,
+      "step": 4972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.721571696685333e-07,
+      "loss": 0.2582,
+      "step": 4973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.705474150973187e-07,
+      "loss": 0.2632,
+      "step": 4974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.689410838259011e-07,
+      "loss": 0.2609,
+      "step": 4975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.673381764253381e-07,
+      "loss": 0.2696,
+      "step": 4976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.657386934654672e-07,
+      "loss": 0.2671,
+      "step": 4977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6414263551490805e-07,
+      "loss": 0.258,
+      "step": 4978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.625500031410667e-07,
+      "loss": 0.2783,
+      "step": 4979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.60960796910127e-07,
+      "loss": 0.2612,
+      "step": 4980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5937501738705584e-07,
+      "loss": 0.2675,
+      "step": 4981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5779266513560564e-07,
+      "loss": 0.2577,
+      "step": 4982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5621374071830527e-07,
+      "loss": 0.2589,
+      "step": 4983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.54638244696468e-07,
+      "loss": 0.2595,
+      "step": 4984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5306617763018804e-07,
+      "loss": 0.2808,
+      "step": 4985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5149754007833606e-07,
+      "loss": 0.2746,
+      "step": 4986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.499323325985726e-07,
+      "loss": 0.2732,
+      "step": 4987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.483705557473327e-07,
+      "loss": 0.2778,
+      "step": 4988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4681221007982877e-07,
+      "loss": 0.2505,
+      "step": 4989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.452572961500633e-07,
+      "loss": 0.2751,
+      "step": 4990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.437058145108074e-07,
+      "loss": 0.2553,
+      "step": 4991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.421577657136221e-07,
+      "loss": 0.2889,
+      "step": 4992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.406131503088417e-07,
+      "loss": 0.2803,
+      "step": 4993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.390719688455824e-07,
+      "loss": 0.2824,
+      "step": 4994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.37534221871737e-07,
+      "loss": 0.28,
+      "step": 4995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.359999099339806e-07,
+      "loss": 0.2697,
+      "step": 4996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.344690335777645e-07,
+      "loss": 0.279,
+      "step": 4997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.329415933473212e-07,
+      "loss": 0.2775,
+      "step": 4998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.314175897856597e-07,
+      "loss": 0.2803,
+      "step": 4999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2989702343457e-07,
+      "loss": 0.2725,
+      "step": 5000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.283798948346162e-07,
+      "loss": 0.2827,
+      "step": 5001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2686620452514027e-07,
+      "loss": 0.269,
+      "step": 5002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2535595304426735e-07,
+      "loss": 0.2689,
+      "step": 5003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2384914092889575e-07,
+      "loss": 0.2752,
+      "step": 5004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.223457687147002e-07,
+      "loss": 0.2736,
+      "step": 5005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2084583693613557e-07,
+      "loss": 0.2617,
+      "step": 5006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1934934612642964e-07,
+      "loss": 0.2723,
+      "step": 5007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1785629681759244e-07,
+      "loss": 0.2828,
+      "step": 5008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1636668954040715e-07,
+      "loss": 0.2722,
+      "step": 5009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.148805248244313e-07,
+      "loss": 0.2618,
+      "step": 5010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.133978031980045e-07,
+      "loss": 0.2826,
+      "step": 5011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1191852518823507e-07,
+      "loss": 0.2596,
+      "step": 5012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1044269132101457e-07,
+      "loss": 0.2587,
+      "step": 5013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.089703021210033e-07,
+      "loss": 0.2713,
+      "step": 5014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.075013581116426e-07,
+      "loss": 0.269,
+      "step": 5015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0603585981514583e-07,
+      "loss": 0.2742,
+      "step": 5016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0457380775250066e-07,
+      "loss": 0.2646,
+      "step": 5017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0311520244347134e-07,
+      "loss": 0.2783,
+      "step": 5018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.016600444065987e-07,
+      "loss": 0.267,
+      "step": 5019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.002083341591933e-07,
+      "loss": 0.2759,
+      "step": 5020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.987600722173445e-07,
+      "loss": 0.2582,
+      "step": 5021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9731525909591274e-07,
+      "loss": 0.2755,
+      "step": 5022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.958738953085338e-07,
+      "loss": 0.2695,
+      "step": 5023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9443598136761787e-07,
+      "loss": 0.2708,
+      "step": 5024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9300151778434593e-07,
+      "loss": 0.2532,
+      "step": 5025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9157050506867567e-07,
+      "loss": 0.2507,
+      "step": 5026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.901429437293357e-07,
+      "loss": 0.2704,
+      "step": 5027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8871883427382676e-07,
+      "loss": 0.2655,
+      "step": 5028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.872981772084282e-07,
+      "loss": 0.2594,
+      "step": 5029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.858809730381851e-07,
+      "loss": 0.278,
+      "step": 5030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.844672222669165e-07,
+      "loss": 0.2693,
+      "step": 5031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.830569253972193e-07,
+      "loss": 0.3029,
+      "step": 5032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.816500829304558e-07,
+      "loss": 0.2609,
+      "step": 5033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.802466953667615e-07,
+      "loss": 0.2773,
+      "step": 5034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7884676320504846e-07,
+      "loss": 0.2611,
+      "step": 5035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7745028694299513e-07,
+      "loss": 0.2639,
+      "step": 5036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.760572670770523e-07,
+      "loss": 0.2596,
+      "step": 5037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.746677041024426e-07,
+      "loss": 0.2557,
+      "step": 5038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7328159851316093e-07,
+      "loss": 0.275,
+      "step": 5039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7189895080197206e-07,
+      "loss": 0.2656,
+      "step": 5040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.705197614604116e-07,
+      "loss": 0.265,
+      "step": 5041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6914403097878405e-07,
+      "loss": 0.2668,
+      "step": 5042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6777175984616823e-07,
+      "loss": 0.2764,
+      "step": 5043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6640294855040826e-07,
+      "loss": 0.2847,
+      "step": 5044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6503759757812276e-07,
+      "loss": 0.2708,
+      "step": 5045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6367570741469784e-07,
+      "loss": 0.2744,
+      "step": 5046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6231727854428845e-07,
+      "loss": 0.2657,
+      "step": 5047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6096231144982166e-07,
+      "loss": 0.2816,
+      "step": 5048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5961080661299096e-07,
+      "loss": 0.2649,
+      "step": 5049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.582627645142599e-07,
+      "loss": 0.261,
+      "step": 5050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.56918185632864e-07,
+      "loss": 0.2749,
+      "step": 5051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.55577070446803e-07,
+      "loss": 0.2504,
+      "step": 5052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.54239419432849e-07,
+      "loss": 0.2817,
+      "step": 5053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5290523306654045e-07,
+      "loss": 0.2813,
+      "step": 5054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.515745118221835e-07,
+      "loss": 0.2828,
+      "step": 5055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.502472561728564e-07,
+      "loss": 0.2716,
+      "step": 5056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4892346659040055e-07,
+      "loss": 0.2598,
+      "step": 5057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4760314354542737e-07,
+      "loss": 0.269,
+      "step": 5058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4628628750731577e-07,
+      "loss": 0.2761,
+      "step": 5059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4497289894421127e-07,
+      "loss": 0.2891,
+      "step": 5060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.436629783230293e-07,
+      "loss": 0.2706,
+      "step": 5061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.423565261094507e-07,
+      "loss": 0.2787,
+      "step": 5062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.410535427679206e-07,
+      "loss": 0.2844,
+      "step": 5063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.397540287616562e-07,
+      "loss": 0.2586,
+      "step": 5064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3845798455263693e-07,
+      "loss": 0.2647,
+      "step": 5065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.37165410601613e-07,
+      "loss": 0.2654,
+      "step": 5066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3587630736809586e-07,
+      "loss": 0.2593,
+      "step": 5067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.345906753103666e-07,
+      "loss": 0.2687,
+      "step": 5068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.333085148854708e-07,
+      "loss": 0.2553,
+      "step": 5069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3202982654922156e-07,
+      "loss": 0.2704,
+      "step": 5070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3075461075619533e-07,
+      "loss": 0.2772,
+      "step": 5071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2948286795973495e-07,
+      "loss": 0.2604,
+      "step": 5072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2821459861194994e-07,
+      "loss": 0.2692,
+      "step": 5073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2694980316371406e-07,
+      "loss": 0.2706,
+      "step": 5074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2568848206466544e-07,
+      "loss": 0.2627,
+      "step": 5075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2443063576320756e-07,
+      "loss": 0.2658,
+      "step": 5076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2317626470650832e-07,
+      "loss": 0.2703,
+      "step": 5077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.21925369340501e-07,
+      "loss": 0.2672,
+      "step": 5078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2067795010988324e-07,
+      "loss": 0.2847,
+      "step": 5079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1943400745811473e-07,
+      "loss": 0.2604,
+      "step": 5080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.181935418274217e-07,
+      "loss": 0.2596,
+      "step": 5081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1695655365879363e-07,
+      "loss": 0.2806,
+      "step": 5082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1572304339198435e-07,
+      "loss": 0.2832,
+      "step": 5083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.144930114655086e-07,
+      "loss": 0.2559,
+      "step": 5084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1326645831664883e-07,
+      "loss": 0.294,
+      "step": 5085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1204338438144734e-07,
+      "loss": 0.2731,
+      "step": 5086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1082379009470854e-07,
+      "loss": 0.2792,
+      "step": 5087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0960767589000565e-07,
+      "loss": 0.2626,
+      "step": 5088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0839504219966944e-07,
+      "loss": 0.2746,
+      "step": 5089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0718588945479402e-07,
+      "loss": 0.2802,
+      "step": 5090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.059802180852377e-07,
+      "loss": 0.2585,
+      "step": 5091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.047780285196188e-07,
+      "loss": 0.2617,
+      "step": 5092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0357932118532318e-07,
+      "loss": 0.2665,
+      "step": 5093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.023840965084911e-07,
+      "loss": 0.2686,
+      "step": 5094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0119235491402934e-07,
+      "loss": 0.2566,
+      "step": 5095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0000409682560783e-07,
+      "loss": 0.2667,
+      "step": 5096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9881932266565318e-07,
+      "loss": 0.2691,
+      "step": 5097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9763803285535843e-07,
+      "loss": 0.2741,
+      "step": 5098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9646022781467544e-07,
+      "loss": 0.2784,
+      "step": 5099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9528590796231596e-07,
+      "loss": 0.2545,
+      "step": 5100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9411507371575598e-07,
+      "loss": 0.2721,
+      "step": 5101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9294772549122932e-07,
+      "loss": 0.2496,
+      "step": 5102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.917838637037317e-07,
+      "loss": 0.2733,
+      "step": 5103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9062348876702107e-07,
+      "loss": 0.2816,
+      "step": 5104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8946660109361193e-07,
+      "loss": 0.2669,
+      "step": 5105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.883132010947841e-07,
+      "loss": 0.2828,
+      "step": 5106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716328918057415e-07,
+      "loss": 0.2581,
+      "step": 5107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8601686575977718e-07,
+      "loss": 0.2913,
+      "step": 5108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.848739312399528e-07,
+      "loss": 0.2603,
+      "step": 5109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8373448602741707e-07,
+      "loss": 0.2801,
+      "step": 5110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8259853052724486e-07,
+      "loss": 0.2768,
+      "step": 5111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8146606514327315e-07,
+      "loss": 0.2723,
+      "step": 5112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8033709027809655e-07,
+      "loss": 0.2738,
+      "step": 5113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7921160633306845e-07,
+      "loss": 0.2718,
+      "step": 5114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7808961370830325e-07,
+      "loss": 0.2794,
+      "step": 5115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7697111280267077e-07,
+      "loss": 0.2652,
+      "step": 5116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7585610401380403e-07,
+      "loss": 0.278,
+      "step": 5117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7474458773808932e-07,
+      "loss": 0.2639,
+      "step": 5118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.736365643706761e-07,
+      "loss": 0.2707,
+      "step": 5119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7253203430546927e-07,
+      "loss": 0.2584,
+      "step": 5120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.714309979351325e-07,
+      "loss": 0.2675,
+      "step": 5121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7033345565108826e-07,
+      "loss": 0.2843,
+      "step": 5122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6923940784351556e-07,
+      "loss": 0.2557,
+      "step": 5123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6814885490135103e-07,
+      "loss": 0.2709,
+      "step": 5124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.670617972122912e-07,
+      "loss": 0.2666,
+      "step": 5125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6597823516278699e-07,
+      "loss": 0.2581,
+      "step": 5126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6489816913804911e-07,
+      "loss": 0.2761,
+      "step": 5127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.638215995220438e-07,
+      "loss": 0.2596,
+      "step": 5128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6274852669749486e-07,
+      "loss": 0.2719,
+      "step": 5129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6167895104588384e-07,
+      "loss": 0.2544,
+      "step": 5130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6061287294744653e-07,
+      "loss": 0.2748,
+      "step": 5131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5955029278117761e-07,
+      "loss": 0.2741,
+      "step": 5132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5849121092482822e-07,
+      "loss": 0.2695,
+      "step": 5133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5743562775490385e-07,
+      "loss": 0.2631,
+      "step": 5134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5638354364666986e-07,
+      "loss": 0.275,
+      "step": 5135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5533495897414264e-07,
+      "loss": 0.2504,
+      "step": 5136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5428987411009843e-07,
+      "loss": 0.2629,
+      "step": 5137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5324828942606894e-07,
+      "loss": 0.2697,
+      "step": 5138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.522102052923402e-07,
+      "loss": 0.272,
+      "step": 5139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5117562207795256e-07,
+      "loss": 0.2615,
+      "step": 5140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5014454015070624e-07,
+      "loss": 0.2687,
+      "step": 5141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4911695987715247e-07,
+      "loss": 0.2689,
+      "step": 5142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4809288162259904e-07,
+      "loss": 0.2763,
+      "step": 5143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.470723057511092e-07,
+      "loss": 0.2682,
+      "step": 5144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4605523262550158e-07,
+      "loss": 0.2685,
+      "step": 5145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4504166260734698e-07,
+      "loss": 0.2603,
+      "step": 5146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4403159605697492e-07,
+      "loss": 0.2635,
+      "step": 5147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4302503333346484e-07,
+      "loss": 0.2854,
+      "step": 5148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4202197479465496e-07,
+      "loss": 0.2739,
+      "step": 5149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4102242079713335e-07,
+      "loss": 0.2605,
+      "step": 5150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4002637169624688e-07,
+      "loss": 0.2652,
+      "step": 5151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3903382784609344e-07,
+      "loss": 0.2662,
+      "step": 5152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3804478959952406e-07,
+      "loss": 0.258,
+      "step": 5153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.370592573081464e-07,
+      "loss": 0.2678,
+      "step": 5154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.360772313223202e-07,
+      "loss": 0.2592,
+      "step": 5155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.350987119911562e-07,
+      "loss": 0.2703,
+      "step": 5156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3412369966252393e-07,
+      "loss": 0.2662,
+      "step": 5157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3315219468304164e-07,
+      "loss": 0.2605,
+      "step": 5158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3218419739808308e-07,
+      "loss": 0.2702,
+      "step": 5159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.312197081517741e-07,
+      "loss": 0.2707,
+      "step": 5160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3025872728699262e-07,
+      "loss": 0.2624,
+      "step": 5161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2930125514537317e-07,
+      "loss": 0.2838,
+      "step": 5162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.283472920672968e-07,
+      "loss": 0.2767,
+      "step": 5163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2739683839190108e-07,
+      "loss": 0.273,
+      "step": 5164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2644989445707688e-07,
+      "loss": 0.2812,
+      "step": 5165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.255064605994627e-07,
+      "loss": 0.2913,
+      "step": 5166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2456653715445465e-07,
+      "loss": 0.2643,
+      "step": 5167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.236301244561977e-07,
+      "loss": 0.2689,
+      "step": 5168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.226972228375889e-07,
+      "loss": 0.2829,
+      "step": 5169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2176783263027847e-07,
+      "loss": 0.2758,
+      "step": 5170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.208419541646666e-07,
+      "loss": 0.2642,
+      "step": 5171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1991958776990664e-07,
+      "loss": 0.2571,
+      "step": 5172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1900073377390297e-07,
+      "loss": 0.2687,
+      "step": 5173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1808539250330986e-07,
+      "loss": 0.271,
+      "step": 5174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1717356428353477e-07,
+      "loss": 0.2747,
+      "step": 5175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1626524943873397e-07,
+      "loss": 0.2598,
+      "step": 5176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1536044829181802e-07,
+      "loss": 0.2759,
+      "step": 5177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1445916116444522e-07,
+      "loss": 0.2629,
+      "step": 5178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1356138837702702e-07,
+      "loss": 0.2687,
+      "step": 5179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1266713024872366e-07,
+      "loss": 0.2811,
+      "step": 5180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1177638709744642e-07,
+      "loss": 0.2694,
+      "step": 5181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1088915923985755e-07,
+      "loss": 0.2601,
+      "step": 5182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1000544699137027e-07,
+      "loss": 0.2667,
+      "step": 5183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0912525066614554e-07,
+      "loss": 0.2697,
+      "step": 5184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0824857057709525e-07,
+      "loss": 0.2623,
+      "step": 5185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0737540703588457e-07,
+      "loss": 0.2486,
+      "step": 5186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0650576035292293e-07,
+      "loss": 0.258,
+      "step": 5187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0563963083737306e-07,
+      "loss": 0.2727,
+      "step": 5188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0477701879714863e-07,
+      "loss": 0.2803,
+      "step": 5189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.039179245389077e-07,
+      "loss": 0.2583,
+      "step": 5190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0306234836806372e-07,
+      "loss": 0.2643,
+      "step": 5191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0221029058877563e-07,
+      "loss": 0.2762,
+      "step": 5192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0136175150395222e-07,
+      "loss": 0.2581,
+      "step": 5193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0051673141525109e-07,
+      "loss": 0.2854,
+      "step": 5194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.967523062308193e-08,
+      "loss": 0.2584,
+      "step": 5195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.883724942659878e-08,
+      "loss": 0.2571,
+      "step": 5196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.800278812370667e-08,
+      "loss": 0.2612,
+      "step": 5197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.717184701105831e-08,
+      "loss": 0.2672,
+      "step": 5198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.634442638405849e-08,
+      "loss": 0.2811,
+      "step": 5199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.552052653685528e-08,
+      "loss": 0.2667,
+      "step": 5200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.470014776234881e-08,
+      "loss": 0.2721,
+      "step": 5201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.388329035218579e-08,
+      "loss": 0.2761,
+      "step": 5202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.306995459676171e-08,
+      "loss": 0.2709,
+      "step": 5203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.226014078521972e-08,
+      "loss": 0.2639,
+      "step": 5204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.145384920545175e-08,
+      "loss": 0.2693,
+      "step": 5205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.065108014409629e-08,
+      "loss": 0.259,
+      "step": 5206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.985183388654062e-08,
+      "loss": 0.2621,
+      "step": 5207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.905611071691855e-08,
+      "loss": 0.2761,
+      "step": 5208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.82639109181127e-08,
+      "loss": 0.2709,
+      "step": 5209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.747523477175224e-08,
+      "loss": 0.269,
+      "step": 5210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.669008255821287e-08,
+      "loss": 0.2583,
+      "step": 5211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.590845455662133e-08,
+      "loss": 0.2732,
+      "step": 5212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.513035104484646e-08,
+      "loss": 0.2792,
+      "step": 5213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.435577229950698e-08,
+      "loss": 0.2689,
+      "step": 5214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.358471859596817e-08,
+      "loss": 0.2702,
+      "step": 5215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.281719020834301e-08,
+      "loss": 0.2845,
+      "step": 5216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.205318740948875e-08,
+      "loss": 0.2591,
+      "step": 5217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.12927104710104e-08,
+      "loss": 0.2726,
+      "step": 5218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.053575966326055e-08,
+      "loss": 0.2631,
+      "step": 5219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.97823352553384e-08,
+      "loss": 0.2805,
+      "step": 5220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.903243751508748e-08,
+      "loss": 0.26,
+      "step": 5221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.828606670909789e-08,
+      "loss": 0.2606,
+      "step": 5222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.754322310270846e-08,
+      "loss": 0.2692,
+      "step": 5223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.68039069600024e-08,
+      "loss": 0.2625,
+      "step": 5224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.606811854380724e-08,
+      "loss": 0.2678,
+      "step": 5225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.533585811569932e-08,
+      "loss": 0.2644,
+      "step": 5226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.46071259360004e-08,
+      "loss": 0.2574,
+      "step": 5227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.388192226377433e-08,
+      "loss": 0.2741,
+      "step": 5228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.316024735683491e-08,
+      "loss": 0.2866,
+      "step": 5229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.244210147173913e-08,
+      "loss": 0.2744,
+      "step": 5230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.172748486379055e-08,
+      "loss": 0.2655,
+      "step": 5231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.101639778703706e-08,
+      "loss": 0.2618,
+      "step": 5232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.03088404942731e-08,
+      "loss": 0.2673,
+      "step": 5233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.960481323703639e-08,
+      "loss": 0.2587,
+      "step": 5234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.890431626561222e-08,
+      "loss": 0.2719,
+      "step": 5235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.82073498290281e-08,
+      "loss": 0.2682,
+      "step": 5236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.751391417505804e-08,
+      "loss": 0.2786,
+      "step": 5237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.68240095502204e-08,
+      "loss": 0.2685,
+      "step": 5238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.613763619977898e-08,
+      "loss": 0.2693,
+      "step": 5239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.545479436774083e-08,
+      "loss": 0.2628,
+      "step": 5240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.477548429685954e-08,
+      "loss": 0.2661,
+      "step": 5241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.409970622863082e-08,
+      "loss": 0.259,
+      "step": 5242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.342746040329584e-08,
+      "loss": 0.2689,
+      "step": 5243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.275874705984008e-08,
+      "loss": 0.2825,
+      "step": 5244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.20935664359934e-08,
+      "loss": 0.2703,
+      "step": 5245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.143191876822996e-08,
+      "loss": 0.2655,
+      "step": 5246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.077380429176716e-08,
+      "loss": 0.2585,
+      "step": 5247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.011922324056674e-08,
+      "loss": 0.2572,
+      "step": 5248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9468175847333664e-08,
+      "loss": 0.2717,
+      "step": 5249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8820662343518354e-08,
+      "loss": 0.2662,
+      "step": 5250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8176682959312223e-08,
+      "loss": 0.261,
+      "step": 5251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.753623792365437e-08,
+      "loss": 0.2527,
+      "step": 5252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.689932746422267e-08,
+      "loss": 0.2816,
+      "step": 5253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6265951807441545e-08,
+      "loss": 0.2643,
+      "step": 5254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.563611117847867e-08,
+      "loss": 0.2665,
+      "step": 5255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.500980580124271e-08,
+      "loss": 0.2733,
+      "step": 5256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4387035898388896e-08,
+      "loss": 0.2605,
+      "step": 5257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.376780169131346e-08,
+      "loss": 0.265,
+      "step": 5258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.315210340015475e-08,
+      "loss": 0.2745,
+      "step": 5259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.253994124379658e-08,
+      "loss": 0.272,
+      "step": 5260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1931315439864846e-08,
+      "loss": 0.278,
+      "step": 5261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.132622620472538e-08,
+      "loss": 0.2686,
+      "step": 5262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.072467375349277e-08,
+      "loss": 0.2727,
+      "step": 5263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.012665830001817e-08,
+      "loss": 0.2576,
+      "step": 5264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.953218005689931e-08,
+      "loss": 0.2645,
+      "step": 5265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.894123923547378e-08,
+      "loss": 0.2833,
+      "step": 5266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8353836045824666e-08,
+      "loss": 0.2667,
+      "step": 5267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.77699706967738e-08,
+      "loss": 0.2673,
+      "step": 5268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.718964339588739e-08,
+      "loss": 0.2882,
+      "step": 5269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6612854349474824e-08,
+      "loss": 0.2883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6039603762585427e-08,
+      "loss": 0.2509,
+      "step": 5271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.546989183901063e-08,
+      "loss": 0.2656,
+      "step": 5272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.490371878128619e-08,
+      "loss": 0.2725,
+      "step": 5273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4341084790688885e-08,
+      "loss": 0.264,
+      "step": 5274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.378199006723427e-08,
+      "loss": 0.2742,
+      "step": 5275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.322643480968558e-08,
+      "loss": 0.2594,
+      "step": 5276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2674419215540385e-08,
+      "loss": 0.2607,
+      "step": 5277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.212594348104615e-08,
+      "loss": 0.2721,
+      "step": 5278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.158100780118468e-08,
+      "loss": 0.271,
+      "step": 5279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.103961236968323e-08,
+      "loss": 0.2578,
+      "step": 5280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0501757379010073e-08,
+      "loss": 0.2672,
+      "step": 5281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.996744302037447e-08,
+      "loss": 0.2624,
+      "step": 5282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.943666948372449e-08,
+      "loss": 0.2642,
+      "step": 5283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.890943695775473e-08,
+      "loss": 0.2684,
+      "step": 5284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8385745629895276e-08,
+      "loss": 0.2772,
+      "step": 5285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7865595686321644e-08,
+      "loss": 0.2684,
+      "step": 5286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.734898731194925e-08,
+      "loss": 0.2736,
+      "step": 5287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.683592069043229e-08,
+      "loss": 0.2843,
+      "step": 5288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.632639600416932e-08,
+      "loss": 0.2833,
+      "step": 5289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5820413434296545e-08,
+      "loss": 0.276,
+      "step": 5290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5317973160692295e-08,
+      "loss": 0.264,
+      "step": 5291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.481907536197704e-08,
+      "loss": 0.2861,
+      "step": 5292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.432372021550778e-08,
+      "loss": 0.2603,
+      "step": 5293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3831907897388105e-08,
+      "loss": 0.2854,
+      "step": 5294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.334363858245593e-08,
+      "loss": 0.2509,
+      "step": 5295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2858912444294624e-08,
+      "loss": 0.2784,
+      "step": 5296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.237772965522412e-08,
+      "loss": 0.2889,
+      "step": 5297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.19000903863087e-08,
+      "loss": 0.2562,
+      "step": 5298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.142599480734698e-08,
+      "loss": 0.2576,
+      "step": 5299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.095544308688414e-08,
+      "loss": 0.2681,
+      "step": 5300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.048843539220303e-08,
+      "loss": 0.2705,
+      "step": 5301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0024971889324184e-08,
+      "loss": 0.2755,
+      "step": 5302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9565052743012468e-08,
+      "loss": 0.2666,
+      "step": 5303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9108678116768206e-08,
+      "loss": 0.2702,
+      "step": 5304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8655848172837175e-08,
+      "loss": 0.2812,
+      "step": 5305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8206563072198377e-08,
+      "loss": 0.2842,
+      "step": 5306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7760822974576272e-08,
+      "loss": 0.2852,
+      "step": 5307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7318628038431882e-08,
+      "loss": 0.2657,
+      "step": 5308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6879978420966123e-08,
+      "loss": 0.2762,
+      "step": 5309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6444874278122033e-08,
+      "loss": 0.2686,
+      "step": 5310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6013315764580326e-08,
+      "loss": 0.2575,
+      "step": 5311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.55853030337605e-08,
+      "loss": 0.2668,
+      "step": 5312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.516083623782195e-08,
+      "loss": 0.2748,
+      "step": 5313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4739915527665082e-08,
+      "loss": 0.2477,
+      "step": 5314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4322541052927973e-08,
+      "loss": 0.2589,
+      "step": 5315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.39087129619886e-08,
+      "loss": 0.2613,
+      "step": 5316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.349843140196373e-08,
+      "loss": 0.2797,
+      "step": 5317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3091696518711127e-08,
+      "loss": 0.2686,
+      "step": 5318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.268850845682402e-08,
+      "loss": 0.2669,
+      "step": 5319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2288867359637754e-08,
+      "loss": 0.255,
+      "step": 5320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.189277336922757e-08,
+      "loss": 0.2706,
+      "step": 5321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1500226626404164e-08,
+      "loss": 0.2574,
+      "step": 5322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1111227270720348e-08,
+      "loss": 0.2585,
+      "step": 5323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.072577544046661e-08,
+      "loss": 0.2729,
+      "step": 5324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0343871272670013e-08,
+      "loss": 0.2639,
+      "step": 5325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9965514903101945e-08,
+      "loss": 0.2805,
+      "step": 5326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.959070646626704e-08,
+      "loss": 0.2723,
+      "step": 5327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.921944609541093e-08,
+      "loss": 0.2642,
+      "step": 5328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8851733922519157e-08,
+      "loss": 0.258,
+      "step": 5329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.848757007831381e-08,
+      "loss": 0.2665,
+      "step": 5330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8126954692255784e-08,
+      "loss": 0.2716,
+      "step": 5331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7769887892544746e-08,
+      "loss": 0.2726,
+      "step": 5332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.741636980612027e-08,
+      "loss": 0.2833,
+      "step": 5333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.706640055865849e-08,
+      "loss": 0.2701,
+      "step": 5334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6719980274574333e-08,
+      "loss": 0.2762,
+      "step": 5335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6377109077021503e-08,
+      "loss": 0.2722,
+      "step": 5336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6037787087891387e-08,
+      "loss": 0.2861,
+      "step": 5337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5702014427814162e-08,
+      "loss": 0.2649,
+      "step": 5338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.536979121615878e-08,
+      "loss": 0.2697,
+      "step": 5339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5041117571030772e-08,
+      "loss": 0.2675,
+      "step": 5340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4715993609274449e-08,
+      "loss": 0.2823,
+      "step": 5341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4394419446474017e-08,
+      "loss": 0.2656,
+      "step": 5342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4076395196948034e-08,
+      "loss": 0.288,
+      "step": 5343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.376192097375495e-08,
+      "loss": 0.269,
+      "step": 5344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3450996888693113e-08,
+      "loss": 0.2556,
+      "step": 5345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3143623052296328e-08,
+      "loss": 0.2668,
+      "step": 5346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2839799573836076e-08,
+      "loss": 0.2709,
+      "step": 5347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2539526561323734e-08,
+      "loss": 0.2666,
+      "step": 5348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2242804121506135e-08,
+      "loss": 0.2492,
+      "step": 5349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1949632359870012e-08,
+      "loss": 0.2682,
+      "step": 5350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1660011380638658e-08,
+      "loss": 0.2755,
+      "step": 5351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1373941286771939e-08,
+      "loss": 0.2673,
+      "step": 5352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1091422179970724e-08,
+      "loss": 0.2665,
+      "step": 5353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0812454160670227e-08,
+      "loss": 0.2604,
+      "step": 5354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0537037328044453e-08,
+      "loss": 0.2559,
+      "step": 5355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0265171780005079e-08,
+      "loss": 0.2626,
+      "step": 5356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.996857613200351e-09,
+      "loss": 0.2764,
+      "step": 5357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.73209492301841e-09,
+      "loss": 0.2656,
+      "step": 5358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.470883803581744e-09,
+      "loss": 0.2757,
+      "step": 5359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.213224347752736e-09,
+      "loss": 0.2696,
+      "step": 5360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.95911664712923e-09,
+      "loss": 0.2548,
+      "step": 5361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.708560792047849e-09,
+      "loss": 0.2604,
+      "step": 5362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.461556871581789e-09,
+      "loss": 0.269,
+      "step": 5363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.218104973541918e-09,
+      "loss": 0.2672,
+      "step": 5364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.978205184475674e-09,
+      "loss": 0.2716,
+      "step": 5365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.741857589668167e-09,
+      "loss": 0.2752,
+      "step": 5366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.509062273142186e-09,
+      "loss": 0.2892,
+      "step": 5367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.279819317657088e-09,
+      "loss": 0.2661,
+      "step": 5368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.054128804708793e-09,
+      "loss": 0.2556,
+      "step": 5369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.831990814532008e-09,
+      "loss": 0.2616,
+      "step": 5370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.613405426096897e-09,
+      "loss": 0.2792,
+      "step": 5371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.398372717111301e-09,
+      "loss": 0.2689,
+      "step": 5372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.186892764019625e-09,
+      "loss": 0.2646,
+      "step": 5373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9789656420050636e-09,
+      "loss": 0.2549,
+      "step": 5374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.774591424984044e-09,
+      "loss": 0.2679,
+      "step": 5375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.573770185614003e-09,
+      "loss": 0.2669,
+      "step": 5376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.376501995286721e-09,
+      "loss": 0.2604,
+      "step": 5377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.182786924131655e-09,
+      "loss": 0.2533,
+      "step": 5378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.992625041015942e-09,
+      "loss": 0.2637,
+      "step": 5379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8060164135410595e-09,
+      "loss": 0.2735,
+      "step": 5380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.622961108048385e-09,
+      "loss": 0.2772,
+      "step": 5381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.443459189614752e-09,
+      "loss": 0.2695,
+      "step": 5382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.267510722052448e-09,
+      "loss": 0.2605,
+      "step": 5383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.09511576791255e-09,
+      "loss": 0.2749,
+      "step": 5384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.926274388482698e-09,
+      "loss": 0.2781,
+      "step": 5385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7609866437848805e-09,
+      "loss": 0.2664,
+      "step": 5386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5992525925798714e-09,
+      "loss": 0.2797,
+      "step": 5387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4410722923650108e-09,
+      "loss": 0.2595,
+      "step": 5388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.286445799374205e-09,
+      "loss": 0.2681,
+      "step": 5389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1353731685768163e-09,
+      "loss": 0.2751,
+      "step": 5390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9878544536798837e-09,
+      "loss": 0.2599,
+      "step": 5391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8438897071259018e-09,
+      "loss": 0.2756,
+      "step": 5392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7034789800972626e-09,
+      "loss": 0.2834,
+      "step": 5393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.566622322508483e-09,
+      "loss": 0.27,
+      "step": 5394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4333197830128663e-09,
+      "loss": 0.2601,
+      "step": 5395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3035714090002823e-09,
+      "loss": 0.2678,
+      "step": 5396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1773772465960576e-09,
+      "loss": 0.2609,
+      "step": 5397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.054737340662083e-09,
+      "loss": 0.2806,
+      "step": 5398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9356517347990377e-09,
+      "loss": 0.2581,
+      "step": 5399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.820120471340836e-09,
+      "loss": 0.2535,
+      "step": 5400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7081435913590682e-09,
+      "loss": 0.2689,
+      "step": 5401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5997211346630016e-09,
+      "loss": 0.2796,
+      "step": 5402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4948531397962484e-09,
+      "loss": 0.2631,
+      "step": 5403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3935396440400984e-09,
+      "loss": 0.2643,
+      "step": 5404
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2957806834112963e-09,
+      "loss": 0.278,
+      "step": 5405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2015762926642639e-09,
+      "loss": 0.2617,
+      "step": 5406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1109265052888784e-09,
+      "loss": 0.2708,
+      "step": 5407
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0238313535104739e-09,
+      "loss": 0.2639,
+      "step": 5408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.402908682920598e-10,
+      "loss": 0.2603,
+      "step": 5409
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.603050793332124e-10,
+      "loss": 0.2919,
+      "step": 5410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.838740150678536e-10,
+      "loss": 0.2623,
+      "step": 5411
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.109977026686921e-10,
+      "loss": 0.2678,
+      "step": 5412
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.416761680427819e-10,
+      "loss": 0.2778,
+      "step": 5413
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.759094358337436e-10,
+      "loss": 0.2592,
+      "step": 5414
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.136975294228741e-10,
+      "loss": 0.2794,
+      "step": 5415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.550404709269263e-10,
+      "loss": 0.2489,
+      "step": 5416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9993828119699875e-10,
+      "loss": 0.2618,
+      "step": 5417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4839097982408696e-10,
+      "loss": 0.2681,
+      "step": 5418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.003985851313118e-10,
+      "loss": 0.265,
+      "step": 5419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.559611141816909e-10,
+      "loss": 0.269,
+      "step": 5420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1507858277258764e-10,
+      "loss": 0.2521,
+      "step": 5421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7775100543793166e-10,
+      "loss": 0.2584,
+      "step": 5422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4397839544710856e-10,
+      "loss": 0.2854,
+      "step": 5423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1376076480718035e-10,
+      "loss": 0.2652,
+      "step": 5424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.7098124260665e-11,
+      "loss": 0.28,
+      "step": 5425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.399048328442625e-11,
+      "loss": 0.2762,
+      "step": 5426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.443785009522472e-11,
+      "loss": 0.2581,
+      "step": 5427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.844023164416676e-11,
+      "loss": 0.2676,
+      "step": 5428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5997633617814702e-11,
+      "loss": 0.268,
+      "step": 5429
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.11006043818685e-12,
+      "loss": 0.2706,
+      "step": 5430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.777515267198382e-12,
+      "loss": 0.2766,
+      "step": 5431
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.2678,
+      "step": 5432
+    },
+    {
+      "epoch": 1.0,
+      "step": 5432,
+      "total_flos": 788519610777600.0,
+      "train_loss": 0.3368299818530525,
+      "train_runtime": 29073.4245,
+      "train_samples_per_second": 47.824,
+      "train_steps_per_second": 0.187
+    }
+  ],
+  "max_steps": 5432,
+  "num_train_epochs": 1,
+  "total_flos": 788519610777600.0,
+  "trial_name": null,
+  "trial_params": null
+}