diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32586 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999861795734095,
+  "eval_steps": 500,
+  "global_step": 5426,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.226993865030675e-06,
+      "loss": 2.9133,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.45398773006135e-06,
+      "loss": 3.055,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.680981595092025e-06,
+      "loss": 2.9751,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9079754601227e-06,
+      "loss": 3.0182,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.134969325153374e-06,
+      "loss": 2.8012,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.36196319018405e-06,
+      "loss": 2.9888,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.588957055214725e-06,
+      "loss": 2.714,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.8159509202454e-06,
+      "loss": 2.3919,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1042944785276074e-05,
+      "loss": 2.7006,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2269938650306748e-05,
+      "loss": 2.2474,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3496932515337424e-05,
+      "loss": 2.2551,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.47239263803681e-05,
+      "loss": 2.1976,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5950920245398772e-05,
+      "loss": 2.1565,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.717791411042945e-05,
+      "loss": 2.1496,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8404907975460123e-05,
+      "loss": 2.1257,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.96319018404908e-05,
+      "loss": 1.8707,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.085889570552147e-05,
+      "loss": 2.1615,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.208588957055215e-05,
+      "loss": 1.8606,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3312883435582822e-05,
+      "loss": 2.0974,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4539877300613496e-05,
+      "loss": 2.0825,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5766871165644174e-05,
+      "loss": 1.9132,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6993865030674848e-05,
+      "loss": 1.9175,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.822085889570552e-05,
+      "loss": 1.7998,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.94478527607362e-05,
+      "loss": 1.7691,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.067484662576687e-05,
+      "loss": 1.6433,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1901840490797544e-05,
+      "loss": 1.6845,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.312883435582822e-05,
+      "loss": 1.8083,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.43558282208589e-05,
+      "loss": 1.784,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.558282208588957e-05,
+      "loss": 1.7445,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6809815950920246e-05,
+      "loss": 1.7455,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8036809815950924e-05,
+      "loss": 1.7485,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.92638036809816e-05,
+      "loss": 1.5979,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.049079754601227e-05,
+      "loss": 1.6039,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.171779141104294e-05,
+      "loss": 1.5644,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2944785276073626e-05,
+      "loss": 1.6367,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.41717791411043e-05,
+      "loss": 1.651,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.539877300613497e-05,
+      "loss": 1.5372,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6625766871165645e-05,
+      "loss": 1.7554,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.785276073619632e-05,
+      "loss": 1.5379,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.907975460122699e-05,
+      "loss": 1.583,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.030674846625767e-05,
+      "loss": 1.5597,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.153374233128835e-05,
+      "loss": 1.7612,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.276073619631902e-05,
+      "loss": 1.595,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3987730061349695e-05,
+      "loss": 1.6658,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.521472392638037e-05,
+      "loss": 1.7242,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.644171779141104e-05,
+      "loss": 1.5264,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.766871165644172e-05,
+      "loss": 1.5155,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.88957055214724e-05,
+      "loss": 1.6741,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.012269938650307e-05,
+      "loss": 1.5787,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.134969325153375e-05,
+      "loss": 1.5553,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.257668711656443e-05,
+      "loss": 1.6424,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.380368098159509e-05,
+      "loss": 1.5311,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.503067484662577e-05,
+      "loss": 1.3966,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.625766871165644e-05,
+      "loss": 1.491,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.748466257668711e-05,
+      "loss": 1.4504,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.87116564417178e-05,
+      "loss": 1.6267,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.993865030674847e-05,
+      "loss": 1.4846,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.116564417177914e-05,
+      "loss": 1.4432,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.239263803680982e-05,
+      "loss": 1.5841,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.361963190184049e-05,
+      "loss": 1.5825,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.484662576687118e-05,
+      "loss": 1.5897,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.607361963190185e-05,
+      "loss": 1.4428,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.730061349693252e-05,
+      "loss": 1.4481,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.85276073619632e-05,
+      "loss": 1.5164,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.975460122699386e-05,
+      "loss": 1.5669,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.098159509202454e-05,
+      "loss": 1.5615,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.220858895705523e-05,
+      "loss": 1.4287,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.343558282208588e-05,
+      "loss": 1.5482,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.466257668711657e-05,
+      "loss": 1.4216,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.588957055214725e-05,
+      "loss": 1.5713,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.711656441717791e-05,
+      "loss": 1.3897,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.83435582822086e-05,
+      "loss": 1.4794,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.957055214723928e-05,
+      "loss": 1.4433,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.079754601226993e-05,
+      "loss": 1.3892,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.202453987730062e-05,
+      "loss": 1.494,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.325153374233129e-05,
+      "loss": 1.5278,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.447852760736196e-05,
+      "loss": 1.3703,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.570552147239264e-05,
+      "loss": 1.4735,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.693251533742331e-05,
+      "loss": 1.5073,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.815950920245399e-05,
+      "loss": 1.3267,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.938650306748467e-05,
+      "loss": 1.4909,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010061349693251534,
+      "loss": 1.2488,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010184049079754601,
+      "loss": 1.3247,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001030674846625767,
+      "loss": 1.4525,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010429447852760737,
+      "loss": 1.4053,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010552147239263804,
+      "loss": 1.6705,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010674846625766872,
+      "loss": 1.5301,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010797546012269939,
+      "loss": 1.533,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010920245398773006,
+      "loss": 1.3868,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011042944785276075,
+      "loss": 1.3824,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011165644171779142,
+      "loss": 1.442,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011288343558282209,
+      "loss": 1.5594,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011411042944785277,
+      "loss": 1.4202,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011533742331288344,
+      "loss": 1.5025,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011656441717791411,
+      "loss": 1.5514,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001177914110429448,
+      "loss": 1.5556,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011901840490797547,
+      "loss": 1.5445,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012024539877300614,
+      "loss": 1.5812,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012147239263803682,
+      "loss": 1.3252,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001226993865030675,
+      "loss": 1.58,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012392638036809816,
+      "loss": 1.3717,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012515337423312886,
+      "loss": 1.337,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012638036809815953,
+      "loss": 1.3888,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012760736196319017,
+      "loss": 1.4114,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012883435582822084,
+      "loss": 1.4551,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013006134969325154,
+      "loss": 1.6447,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001312883435582822,
+      "loss": 1.2774,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013251533742331288,
+      "loss": 1.3539,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013374233128834358,
+      "loss": 1.3999,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013496932515337422,
+      "loss": 1.5652,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001361963190184049,
+      "loss": 1.3521,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001374233128834356,
+      "loss": 1.3553,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013865030674846626,
+      "loss": 1.4753,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013987730061349693,
+      "loss": 1.5041,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014110429447852763,
+      "loss": 1.2509,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014233128834355828,
+      "loss": 1.5168,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014355828220858895,
+      "loss": 1.4064,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014478527607361964,
+      "loss": 1.4439,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014601226993865031,
+      "loss": 1.4274,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014723926380368098,
+      "loss": 1.5336,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014846625766871168,
+      "loss": 1.4122,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014969325153374235,
+      "loss": 1.3413,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000150920245398773,
+      "loss": 1.3997,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001521472392638037,
+      "loss": 1.417,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015337423312883436,
+      "loss": 1.5019,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015460122699386504,
+      "loss": 1.3009,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015582822085889573,
+      "loss": 1.4166,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001570552147239264,
+      "loss": 1.3501,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015828220858895705,
+      "loss": 1.3895,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015950920245398772,
+      "loss": 1.3085,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016073619631901842,
+      "loss": 1.3562,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016196319018404909,
+      "loss": 1.3536,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016319018404907976,
+      "loss": 1.4299,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016441717791411045,
+      "loss": 1.3528,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001656441717791411,
+      "loss": 1.3283,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016687116564417177,
+      "loss": 1.3906,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016809815950920247,
+      "loss": 1.2988,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016932515337423314,
+      "loss": 1.4344,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001705521472392638,
+      "loss": 1.328,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001717791411042945,
+      "loss": 1.4753,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017300613496932518,
+      "loss": 1.4446,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017423312883435582,
+      "loss": 1.3035,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017546012269938652,
+      "loss": 1.336,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001766871165644172,
+      "loss": 1.3953,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017791411042944786,
+      "loss": 1.3547,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017914110429447856,
+      "loss": 1.3653,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018036809815950923,
+      "loss": 1.4712,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018159509202453987,
+      "loss": 1.277,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018282208588957057,
+      "loss": 1.3809,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018404907975460124,
+      "loss": 1.353,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001852760736196319,
+      "loss": 1.2671,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018650306748466258,
+      "loss": 1.329,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018773006134969328,
+      "loss": 1.3542,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018895705521472392,
+      "loss": 1.4236,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001901840490797546,
+      "loss": 1.3466,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001914110429447853,
+      "loss": 1.2041,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019263803680981596,
+      "loss": 1.2636,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019386503067484663,
+      "loss": 1.2953,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019509202453987733,
+      "loss": 1.3976,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019631901840490797,
+      "loss": 1.4372,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019754601226993864,
+      "loss": 1.4172,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019877300613496934,
+      "loss": 1.2351,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 1.4594,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999998218429567,
+      "loss": 1.3737,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999992873718898,
+      "loss": 1.395,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199999839658699,
+      "loss": 1.4466,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999997149488575,
+      "loss": 1.3668,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999955460770887,
+      "loss": 1.2873,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999935863531025,
+      "loss": 1.255,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999912703173148,
+      "loss": 1.2899,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999885979705512,
+      "loss": 1.3567,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999985569313763,
+      "loss": 1.3086,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999821843480297,
+      "loss": 1.2711,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999784430745575,
+      "loss": 1.4246,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999743454946799,
+      "loss": 1.4704,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999698916098562,
+      "loss": 1.571,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999650814216742,
+      "loss": 1.2829,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999959914931847,
+      "loss": 1.2935,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999954392142216,
+      "loss": 1.4999,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999948513054749,
+      "loss": 1.4001,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999422776715404,
+      "loss": 1.4315,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999356859948123,
+      "loss": 1.3782,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999287380269133,
+      "loss": 1.3725,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999214337703195,
+      "loss": 1.552,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999137732276326,
+      "loss": 1.2235,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999057564015833,
+      "loss": 1.4118,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999897383295027,
+      "loss": 1.4871,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998886539109479,
+      "loss": 1.402,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999879568252456,
+      "loss": 1.4212,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998701263227887,
+      "loss": 1.2837,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998603281253107,
+      "loss": 1.4063,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999850173663513,
+      "loss": 1.3891,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998396629410133,
+      "loss": 1.26,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998287959615573,
+      "loss": 1.2396,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999817572729017,
+      "loss": 1.2441,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998059932473914,
+      "loss": 1.3385,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999794057520806,
+      "loss": 1.2471,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999781765553514,
+      "loss": 1.2956,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997691173498955,
+      "loss": 1.4159,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997561129144565,
+      "loss": 1.4816,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997427522518315,
+      "loss": 1.2821,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997290353667804,
+      "loss": 1.3105,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999714962264191,
+      "loss": 1.4725,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997005329490776,
+      "loss": 1.4484,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999685747426582,
+      "loss": 1.1717,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996706057019722,
+      "loss": 1.3503,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996551077806435,
+      "loss": 1.3575,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996392536681174,
+      "loss": 1.3482,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999623043370044,
+      "loss": 1.5001,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996064768921983,
+      "loss": 1.156,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999589554240484,
+      "loss": 1.3218,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995722754209303,
+      "loss": 1.2664,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999554640439694,
+      "loss": 1.3725,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995366493030589,
+      "loss": 1.1872,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995183020174353,
+      "loss": 1.4838,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994995985893606,
+      "loss": 1.4477,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999480539025499,
+      "loss": 1.4168,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994611233326422,
+      "loss": 1.6,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999441351517708,
+      "loss": 1.2846,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994212235877406,
+      "loss": 1.5858,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994007395499134,
+      "loss": 1.3058,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993798994115237,
+      "loss": 1.3324,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993587031799984,
+      "loss": 1.3545,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993371508628886,
+      "loss": 1.281,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993152424678748,
+      "loss": 1.2807,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999292978002763,
+      "loss": 1.3583,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992703574754864,
+      "loss": 1.2874,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992473808941045,
+      "loss": 1.3659,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992240482668048,
+      "loss": 1.3034,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999200359601901,
+      "loss": 1.4119,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999176314907833,
+      "loss": 1.3809,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991519141931692,
+      "loss": 1.3961,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991271574666035,
+      "loss": 1.3355,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991020447369568,
+      "loss": 1.4137,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990765760131776,
+      "loss": 1.3358,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990507513043405,
+      "loss": 1.4359,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999024570619647,
+      "loss": 1.3345,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989980339684261,
+      "loss": 1.3421,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989711413601332,
+      "loss": 1.2213,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199894389280435,
+      "loss": 1.426,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998916288310786,
+      "loss": 1.2883,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998888327889277,
+      "loss": 1.3506,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988600115497854,
+      "loss": 1.3957,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998831339302401,
+      "loss": 1.3449,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199880231115734,
+      "loss": 1.4988,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987729271249457,
+      "loss": 1.549,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987431872156878,
+      "loss": 1.4106,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987130914401633,
+      "loss": 1.3517,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986826398090955,
+      "loss": 1.4112,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998651832333335,
+      "loss": 1.2307,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986206690238587,
+      "loss": 1.4371,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985891498917706,
+      "loss": 1.3221,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985572749483014,
+      "loss": 1.3698,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998525044204809,
+      "loss": 1.3083,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998492457672777,
+      "loss": 1.3395,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984595153638165,
+      "loss": 1.3562,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998426217289666,
+      "loss": 1.3313,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983925634621893,
+      "loss": 1.4963,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983585538933785,
+      "loss": 1.5295,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998324188595351,
+      "loss": 1.3574,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982894675803518,
+      "loss": 1.3337,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982543908607524,
+      "loss": 1.3864,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982189584490519,
+      "loss": 1.2631,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981831703578742,
+      "loss": 1.3177,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998147026599972,
+      "loss": 1.463,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981105271882235,
+      "loss": 1.2233,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980736721356336,
+      "loss": 1.3262,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980364614553348,
+      "loss": 1.3905,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979988951605856,
+      "loss": 1.4483,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979609732647717,
+      "loss": 1.4095,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979226957814045,
+      "loss": 1.1959,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978840627241236,
+      "loss": 1.4526,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997845074106694,
+      "loss": 1.3379,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978057299430079,
+      "loss": 1.3522,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977660302470843,
+      "loss": 1.3332,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997725975033069,
+      "loss": 1.4385,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976855643152338,
+      "loss": 1.3786,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976447981079782,
+      "loss": 1.2158,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997603676425827,
+      "loss": 1.3663,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975621992834327,
+      "loss": 1.3057,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975203666955745,
+      "loss": 1.2422,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997478178677158,
+      "loss": 1.3122,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974356352432147,
+      "loss": 1.4501,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973927364089043,
+      "loss": 1.2732,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997349482189512,
+      "loss": 1.2324,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997305872600449,
+      "loss": 1.4399,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997261907657255,
+      "loss": 1.5414,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997217587375595,
+      "loss": 1.3618,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997172911771261,
+      "loss": 1.354,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971278808601716,
+      "loss": 1.2714,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997082494658372,
+      "loss": 1.3236,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970367531820337,
+      "loss": 1.4577,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996990656447455,
+      "loss": 1.4747,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969442044710613,
+      "loss": 1.4719,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968973972694037,
+      "loss": 1.3618,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199685023485916,
+      "loss": 1.1732,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968027172571356,
+      "loss": 1.4505,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019967548444802612,
+      "loss": 1.3874,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967066165455942,
+      "loss": 1.2825,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966580334703196,
+      "loss": 1.3764,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996609095271748,
+      "loss": 1.197,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965598019673167,
+      "loss": 1.2907,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965101535745895,
+      "loss": 1.433,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964601501112566,
+      "loss": 1.4297,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964097915951357,
+      "loss": 1.2658,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963590780441696,
+      "loss": 1.3734,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963080094764286,
+      "loss": 1.3946,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962565859101087,
+      "loss": 1.2515,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962048073635336,
+      "loss": 1.3725,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961526738551522,
+      "loss": 1.5055,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199610018540354,
+      "loss": 1.2875,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960473420274004,
+      "loss": 1.3318,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959941437455614,
+      "loss": 1.3646,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959405905769785,
+      "loss": 1.322,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995886682540734,
+      "loss": 1.4699,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995832419656035,
+      "loss": 1.3175,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957778019422172,
+      "loss": 1.2536,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995722829418741,
+      "loss": 1.2587,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956675021051945,
+      "loss": 1.43,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995611820021291,
+      "loss": 1.4215,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955557831868707,
+      "loss": 1.2245,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954993916219008,
+      "loss": 1.3355,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954426453464746,
+      "loss": 1.3529,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953855443808108,
+      "loss": 1.2456,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995328088745256,
+      "loss": 1.2831,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952702784602818,
+      "loss": 1.2316,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952121135464874,
+      "loss": 1.4306,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951535940245977,
+      "loss": 1.2444,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950947199154638,
+      "loss": 1.3893,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950354912400635,
+      "loss": 1.3107,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949759080195006,
+      "loss": 1.35,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994915970275006,
+      "loss": 1.4179,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948556780279356,
+      "loss": 1.2205,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994795031299773,
+      "loss": 1.3611,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947340301121268,
+      "loss": 1.3528,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946726744867336,
+      "loss": 1.298,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946109644454544,
+      "loss": 1.281,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945489000102777,
+      "loss": 1.4903,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944864812033178,
+      "loss": 1.5107,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994423708046816,
+      "loss": 1.2177,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994360580563138,
+      "loss": 1.304,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994297098774778,
+      "loss": 1.4084,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942332627043556,
+      "loss": 1.2874,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994169072374616,
+      "loss": 1.1817,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941045278084308,
+      "loss": 1.367,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940396290287987,
+      "loss": 1.3391,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939743760588442,
+      "loss": 1.3201,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939087689218175,
+      "loss": 1.3155,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938428076410954,
+      "loss": 1.3715,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937764922401806,
+      "loss": 1.4434,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993709822742703,
+      "loss": 1.3175,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019936427991724167,
+      "loss": 1.3147,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935754215532042,
+      "loss": 1.2638,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935076899090724,
+      "loss": 1.3556,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934396042641555,
+      "loss": 1.2999,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993371164642713,
+      "loss": 1.3627,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933023710691315,
+      "loss": 1.2443,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932332235679225,
+      "loss": 1.2654,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931637221637242,
+      "loss": 1.353,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930938668813016,
+      "loss": 1.2648,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930236577455444,
+      "loss": 1.2313,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929530947814693,
+      "loss": 1.2778,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928821780142194,
+      "loss": 1.4433,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928109074690624,
+      "loss": 1.3975,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992739283171394,
+      "loss": 1.2093,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992667305146734,
+      "loss": 1.4877,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925949734207299,
+      "loss": 1.2254,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992522288019154,
+      "loss": 1.2159,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992449248967906,
+      "loss": 1.2056,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923758562930094,
+      "loss": 1.308,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923021100206164,
+      "loss": 1.2907,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992228010177003,
+      "loss": 1.3554,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992153556788572,
+      "loss": 1.3918,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920787498818528,
+      "loss": 1.2498,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920035894834993,
+      "loss": 1.2315,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919280756202932,
+      "loss": 1.3674,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918522083191403,
+      "loss": 1.2367,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917759876070736,
+      "loss": 1.262,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916994135112516,
+      "loss": 1.2442,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916224860589587,
+      "loss": 1.3386,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019915452052776051,
+      "loss": 1.3748,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914675711947273,
+      "loss": 1.2694,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913895838379872,
+      "loss": 1.3321,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913112432351726,
+      "loss": 1.32,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912325494141978,
+      "loss": 1.2755,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911535024031026,
+      "loss": 1.4258,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910741022300519,
+      "loss": 1.2682,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909943489233376,
+      "loss": 1.2658,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909142425113767,
+      "loss": 1.2079,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908337830227123,
+      "loss": 1.2834,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907529704860137,
+      "loss": 1.2887,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906718049300749,
+      "loss": 1.2534,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905902863838167,
+      "loss": 1.2917,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990508414876285,
+      "loss": 1.3415,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990426190436652,
+      "loss": 1.3396,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019903436130942157,
+      "loss": 1.2498,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990260682878399,
+      "loss": 1.3885,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901773998187513,
+      "loss": 1.1937,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900937639449477,
+      "loss": 1.3459,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900097752867886,
+      "loss": 1.3642,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019899254338742007,
+      "loss": 1.3218,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898407397372355,
+      "loss": 1.3259,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989755692906071,
+      "loss": 1.369,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989670293411011,
+      "loss": 1.5343,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989584541282484,
+      "loss": 1.3461,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019894984365510449,
+      "loss": 1.2003,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019894119792473734,
+      "loss": 1.2929,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019893251694022768,
+      "loss": 1.3947,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892380070466855,
+      "loss": 1.2045,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989150492211657,
+      "loss": 1.3784,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019890626249283744,
+      "loss": 1.1913,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889744052281456,
+      "loss": 1.2795,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019888858331424048,
+      "loss": 1.2899,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887969087027113,
+      "loss": 1.1906,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887076319407502,
+      "loss": 1.3078,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019886180028883321,
+      "loss": 1.3758,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885280215773933,
+      "loss": 1.438,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988437688039995,
+      "loss": 1.246,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883470023083244,
+      "loss": 1.4632,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882559644146945,
+      "loss": 1.3554,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019881645743915428,
+      "loss": 1.2644,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019880728322714333,
+      "loss": 1.4274,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987980738087055,
+      "loss": 1.5054,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878882918712218,
+      "loss": 1.3347,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877954936568744,
+      "loss": 1.2321,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877023434770774,
+      "loss": 1.1806,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987608841365022,
+      "loss": 1.411,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987514987354024,
+      "loss": 1.2936,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019874207814775253,
+      "loss": 1.1896,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019873262237690923,
+      "loss": 1.3398,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872313142624176,
+      "loss": 1.3694,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871360529913186,
+      "loss": 1.2843,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019870404399897383,
+      "loss": 1.3487,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019869444752917452,
+      "loss": 1.3347,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019868481589315325,
+      "loss": 1.4285,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019867514909434193,
+      "loss": 1.1552,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866544713618493,
+      "loss": 1.4068,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019865571002213928,
+      "loss": 1.2267,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019864593775567436,
+      "loss": 1.3524,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019863613034027224,
+      "loss": 1.1046,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986262877794274,
+      "loss": 1.3049,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986164100766469,
+      "loss": 1.3798,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986064972354503,
+      "loss": 1.2179,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019859654925936967,
+      "loss": 1.312,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019858656615194965,
+      "loss": 1.3714,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857654791674732,
+      "loss": 1.61,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019856649455733232,
+      "loss": 1.3262,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019855640607728683,
+      "loss": 1.4056,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854628248020553,
+      "loss": 1.2483,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019853612376969558,
+      "loss": 1.3335,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852592994937663,
+      "loss": 1.2708,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019851570102288096,
+      "loss": 1.2169,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985054369938532,
+      "loss": 1.4164,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849513786595067,
+      "loss": 1.2357,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198484803642843,
+      "loss": 1.3957,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019847443432821244,
+      "loss": 1.2552,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846402992575377,
+      "loss": 1.324,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019845359043917416,
+      "loss": 1.3609,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984431158721934,
+      "loss": 1.1403,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984326062285437,
+      "loss": 1.4419,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019842206151196977,
+      "loss": 1.3143,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001984114817262289,
+      "loss": 1.4272,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019840086687509078,
+      "loss": 1.3789,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019839021696233764,
+      "loss": 1.3452,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983795319917642,
+      "loss": 1.2198,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983688119671776,
+      "loss": 1.3062,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019835805689239768,
+      "loss": 1.3758,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019834726677125648,
+      "loss": 1.2731,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833644160759874,
+      "loss": 1.482,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019832558140528162,
+      "loss": 1.3052,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831468616817471,
+      "loss": 1.2674,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019830375590016022,
+      "loss": 1.1693,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019829279060513272,
+      "loss": 1.2817,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019828179028699925,
+      "loss": 1.3387,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827075494967944,
+      "loss": 1.2878,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019825968459710532,
+      "loss": 1.3464,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982485792332214,
+      "loss": 1.2953,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982374388619847,
+      "loss": 1.4282,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982262634873647,
+      "loss": 1.3335,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019821505311334333,
+      "loss": 1.2368,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019820380774391495,
+      "loss": 1.2242,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981925273830865,
+      "loss": 1.3069,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818121203487736,
+      "loss": 1.1543,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981698617033193,
+      "loss": 1.4138,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019815847639245658,
+      "loss": 1.2415,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000198147056106346,
+      "loss": 1.2266,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019813560084905676,
+      "loss": 1.2534,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019812411062467052,
+      "loss": 1.3184,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981125854372814,
+      "loss": 1.3385,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810102529099597,
+      "loss": 1.2096,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019808943018993332,
+      "loss": 1.2634,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019807780013822492,
+      "loss": 1.3767,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019806613514001474,
+      "loss": 1.4014,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019805443519945915,
+      "loss": 1.3266,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019804270032072703,
+      "loss": 1.3097,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019803093050799967,
+      "loss": 1.3006,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801912576547083,
+      "loss": 1.3819,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980072860973467,
+      "loss": 1.373,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019799541150784594,
+      "loss": 1.3092,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979835020011996,
+      "loss": 1.3512,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979715575816512,
+      "loss": 1.3284,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795957825345678,
+      "loss": 1.4825,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794756402088462,
+      "loss": 1.2665,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979355148882157,
+      "loss": 1.2601,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019792343085974317,
+      "loss": 1.3904,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979113119397728,
+      "loss": 1.3562,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019789915813262274,
+      "loss": 1.2801,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019788696944262357,
+      "loss": 1.2923,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019787474587411823,
+      "loss": 1.4362,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978624874314622,
+      "loss": 1.2619,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785019411902336,
+      "loss": 1.2265,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019783786594118193,
+      "loss": 1.3166,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019782550290233064,
+      "loss": 1.2106,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019781310500687464,
+      "loss": 1.4332,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019780067225923142,
+      "loss": 1.3165,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019778820466383098,
+      "loss": 1.2276,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977757022251157,
+      "loss": 1.331,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019776316494754035,
+      "loss": 1.2565,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019775059283557217,
+      "loss": 1.4366,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977379858936908,
+      "loss": 1.2748,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977253441263882,
+      "loss": 1.3397,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019771266753816885,
+      "loss": 1.2467,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019769995613354958,
+      "loss": 1.4536,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976872099170597,
+      "loss": 1.2744,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976744288932408,
+      "loss": 1.3113,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019766161306664694,
+      "loss": 1.26,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764876244184466,
+      "loss": 1.2703,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976358770234127,
+      "loss": 1.3892,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976229568159424,
+      "loss": 1.2105,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976100018240374,
+      "loss": 1.2751,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019759701205231372,
+      "loss": 1.3075,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019758398750539984,
+      "loss": 1.1783,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757092818793655,
+      "loss": 1.2705,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019755783410457705,
+      "loss": 1.3734,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754470525998704,
+      "loss": 1.2772,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975315416588444,
+      "loss": 1.1862,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019751834330583956,
+      "loss": 1.0466,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019750511020567532,
+      "loss": 1.342,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019749184236306672,
+      "loss": 1.385,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019747853978274135,
+      "loss": 1.3489,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974652024694391,
+      "loss": 1.2117,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019745183042791227,
+      "loss": 1.3268,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974384236629254,
+      "loss": 1.2534,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019742498217925565,
+      "loss": 1.2625,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019741150598169232,
+      "loss": 1.2415,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973979950750372,
+      "loss": 1.1915,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019738444946410436,
+      "loss": 1.1787,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019737086915372037,
+      "loss": 1.288,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735725414872404,
+      "loss": 1.3228,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973436044539666,
+      "loss": 1.2734,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019732992007431168,
+      "loss": 1.384,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019731620101463512,
+      "loss": 1.341,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019730244727982527,
+      "loss": 1.2446,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019728865887478278,
+      "loss": 1.3869,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019727483580442065,
+      "loss": 1.2327,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972609780736642,
+      "loss": 1.2164,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019724708568745117,
+      "loss": 1.2732,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019723315865073163,
+      "loss": 1.3267,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019721919696846795,
+      "loss": 1.2916,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972052006456349,
+      "loss": 1.4113,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971911696872195,
+      "loss": 1.3024,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717710409822127,
+      "loss": 1.1392,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019716300388365193,
+      "loss": 1.2854,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019714886904853554,
+      "loss": 1.2892,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713469959790864,
+      "loss": 1.3296,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019712049553681992,
+      "loss": 1.3822,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019710625687033054,
+      "loss": 1.1701,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019709198360351392,
+      "loss": 1.3691,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001970776757414558,
+      "loss": 1.2264,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019706333328925433,
+      "loss": 1.256,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704895625201986,
+      "loss": 1.4763,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019703454463487521,
+      "loss": 1.1871,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019702009844295537,
+      "loss": 1.2319,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019700561768140775,
+      "loss": 1.3715,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019699110235539205,
+      "loss": 1.246,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019697655247008028,
+      "loss": 1.2041,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019696196803065677,
+      "loss": 1.4268,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019694734904231818,
+      "loss": 1.3034,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019693269551027346,
+      "loss": 1.2512,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019691800743974384,
+      "loss": 1.3925,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019690328483596288,
+      "loss": 1.3427,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968885277041765,
+      "loss": 1.2747,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019687373604964284,
+      "loss": 1.2887,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968589098776324,
+      "loss": 1.2217,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968440491934279,
+      "loss": 1.3337,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019682915400232447,
+      "loss": 1.1409,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681422430962943,
+      "loss": 1.3202,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019679926012066247,
+      "loss": 1.1211,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019678426144075558,
+      "loss": 1.3302,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967692282752529,
+      "loss": 1.3204,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019675416062951104,
+      "loss": 1.1841,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967390585088988,
+      "loss": 1.2738,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019672392191879724,
+      "loss": 1.2795,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967087508645998,
+      "loss": 1.2633,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019669354535171207,
+      "loss": 1.2653,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019667830538555204,
+      "loss": 1.3595,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019666303097154992,
+      "loss": 1.1885,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019664772211514818,
+      "loss": 1.2292,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966323788218016,
+      "loss": 1.3721,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966170010969772,
+      "loss": 1.3337,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966015889461543,
+      "loss": 1.2437,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965861423748244,
+      "loss": 1.3784,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019657066138849142,
+      "loss": 1.4655,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965551459926714,
+      "loss": 1.2058,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965395961928927,
+      "loss": 1.3213,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019652401199469596,
+      "loss": 1.3299,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019650839340363402,
+      "loss": 1.3567,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019649274042527203,
+      "loss": 1.1036,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964770530651873,
+      "loss": 1.206,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019646133132896957,
+      "loss": 1.1957,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964455752222206,
+      "loss": 1.2791,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964297847505546,
+      "loss": 1.1006,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964139599195979,
+      "loss": 1.449,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963981007349891,
+      "loss": 1.3149,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019638220720237908,
+      "loss": 1.1924,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963662793274309,
+      "loss": 1.2912,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963503171158199,
+      "loss": 1.2306,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019633432057323367,
+      "loss": 1.2028,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019631828970537196,
+      "loss": 1.275,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963022245179468,
+      "loss": 1.1875,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019628612501668244,
+      "loss": 1.2931,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962699912073154,
+      "loss": 1.3897,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019625382309559433,
+      "loss": 1.3887,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962376206872802,
+      "loss": 1.342,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001962213839881461,
+      "loss": 1.1804,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019620511300397748,
+      "loss": 1.2132,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019618880774057186,
+      "loss": 1.2959,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019617246820373904,
+      "loss": 1.1326,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019615609439930102,
+      "loss": 1.2853,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019613968633309204,
+      "loss": 1.2793,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019612324401095855,
+      "loss": 1.2479,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019610676743875912,
+      "loss": 1.1642,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960902566223646,
+      "loss": 1.2391,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019607371156765805,
+      "loss": 1.1625,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960571322805347,
+      "loss": 1.3174,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019604051876690197,
+      "loss": 1.192,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960238710326795,
+      "loss": 1.135,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019600718908379912,
+      "loss": 1.2268,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019599047292620482,
+      "loss": 1.2769,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959737225658528,
+      "loss": 1.3075,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019595693800871148,
+      "loss": 1.3346,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019594011926076145,
+      "loss": 1.191,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959232663279954,
+      "loss": 1.184,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019590637921641834,
+      "loss": 1.4132,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958894579320473,
+      "loss": 1.2753,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019587250248091165,
+      "loss": 1.4092,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019585551286905282,
+      "loss": 1.2548,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019583848910252445,
+      "loss": 1.3317,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019582143118739234,
+      "loss": 1.3346,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958043391297345,
+      "loss": 1.2209,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019578721293564104,
+      "loss": 1.3885,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019577005261121427,
+      "loss": 1.32,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019575285816256867,
+      "loss": 1.1568,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019573562959583084,
+      "loss": 1.2728,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019571836691713957,
+      "loss": 1.1899,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019570107013264581,
+      "loss": 1.373,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019568373924851265,
+      "loss": 1.2485,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019566637427091533,
+      "loss": 1.2093,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019564897520604117,
+      "loss": 1.2889,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956315420600898,
+      "loss": 1.3359,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019561407483927284,
+      "loss": 1.3748,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019559657354981412,
+      "loss": 1.1845,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019557903819794956,
+      "loss": 1.1793,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019556146878992728,
+      "loss": 1.3056,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019554386533200755,
+      "loss": 1.201,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019552622783046268,
+      "loss": 1.2386,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019550855629157717,
+      "loss": 1.2733,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019549085072164765,
+      "loss": 1.4233,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019547311112698282,
+      "loss": 1.198,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019545533751390363,
+      "loss": 1.2365,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195437529888743,
+      "loss": 1.3078,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019541968825784607,
+      "loss": 1.1676,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019540181262757005,
+      "loss": 1.1725,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019538390300428426,
+      "loss": 1.2591,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019536595939437018,
+      "loss": 1.2647,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019534798180422138,
+      "loss": 1.3211,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953299702402435,
+      "loss": 1.2914,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019531192470885434,
+      "loss": 1.3499,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019529384521648377,
+      "loss": 1.2062,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019527573176957375,
+      "loss": 1.2547,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019525758437457835,
+      "loss": 1.1982,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952394030379638,
+      "loss": 1.3225,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019522118776620827,
+      "loss": 1.252,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019520293856580223,
+      "loss": 1.3186,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019518465544324806,
+      "loss": 1.4751,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951663384050603,
+      "loss": 1.217,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514798745776557,
+      "loss": 1.2782,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019512960260790256,
+      "loss": 1.1716,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951111838620221,
+      "loss": 1.2439,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019509273122668698,
+      "loss": 1.234,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950742447084722,
+      "loss": 1.3358,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950557243139647,
+      "loss": 1.3954,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019503717004976364,
+      "loss": 1.2893,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019501858192248007,
+      "loss": 1.1996,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949999599387373,
+      "loss": 1.2046,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019498130410517054,
+      "loss": 1.2776,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019496261442842713,
+      "loss": 1.2805,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949438909151665,
+      "loss": 1.3144,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019492513357206005,
+      "loss": 1.3624,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019490634240579134,
+      "loss": 1.3143,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019488751742305588,
+      "loss": 1.1949,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948686586305613,
+      "loss": 1.2804,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019484976603502727,
+      "loss": 1.3076,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019483083964318544,
+      "loss": 1.2584,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948118794617796,
+      "loss": 1.2261,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019479288549756555,
+      "loss": 1.2492,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000194773857757311,
+      "loss": 1.2525,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947547962477959,
+      "loss": 1.2207,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947357009758121,
+      "loss": 1.2,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947165719481635,
+      "loss": 1.306,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019469740917166608,
+      "loss": 1.3034,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019467821265314777,
+      "loss": 1.3047,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946589823994486,
+      "loss": 1.2443,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019463971841742056,
+      "loss": 1.4226,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019462042071392762,
+      "loss": 1.3659,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019460108929584592,
+      "loss": 1.2171,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019458172417006347,
+      "loss": 1.2141,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019456232534348033,
+      "loss": 1.2808,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945428928230086,
+      "loss": 1.3332,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019452342661557237,
+      "loss": 1.3255,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945039267281077,
+      "loss": 1.3136,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944843931675626,
+      "loss": 1.2888,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944648259408973,
+      "loss": 1.2373,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944452250550838,
+      "loss": 1.228,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019442559051710618,
+      "loss": 1.1468,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944059223339605,
+      "loss": 1.3023,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943862205126548,
+      "loss": 1.4126,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019436648506020916,
+      "loss": 1.3094,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019434671598365553,
+      "loss": 1.241,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019432691329003795,
+      "loss": 1.2634,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019430707698641242,
+      "loss": 1.2991,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019428720707984686,
+      "loss": 1.2839,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019426730357742123,
+      "loss": 1.3583,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942473664862274,
+      "loss": 1.1762,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019422739581336922,
+      "loss": 1.0533,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019420739156596258,
+      "loss": 1.1889,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019418735375113523,
+      "loss": 1.3905,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019416728237602694,
+      "loss": 1.3503,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941471774477894,
+      "loss": 1.2366,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019412703897358634,
+      "loss": 1.2518,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941068669605933,
+      "loss": 1.2818,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019408666141599792,
+      "loss": 1.3562,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019406642234699967,
+      "loss": 1.3772,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019404614976081008,
+      "loss": 1.3198,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019402584366465248,
+      "loss": 1.3195,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019400550406576227,
+      "loss": 1.415,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019398513097138676,
+      "loss": 1.3455,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019396472438878507,
+      "loss": 1.3489,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019394428432522847,
+      "loss": 1.1975,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019392381078799998,
+      "loss": 1.0715,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019390330378439458,
+      "loss": 1.1901,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938827633217193,
+      "loss": 1.2514,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019386218940729288,
+      "loss": 1.1846,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938415820484462,
+      "loss": 1.2421,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019382094125252187,
+      "loss": 1.1659,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019380026702687454,
+      "loss": 1.2843,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019377955937887071,
+      "loss": 1.2954,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019375881831588886,
+      "loss": 1.3021,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019373804384531927,
+      "loss": 1.2438,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019371723597456415,
+      "loss": 1.2876,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019369639471103768,
+      "loss": 1.2641,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019367552006216592,
+      "loss": 1.2703,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019365461203538675,
+      "loss": 1.2585,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019363367063815004,
+      "loss": 1.1882,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019361269587791745,
+      "loss": 1.3171,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019359168776216263,
+      "loss": 1.2472,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019357064629837103,
+      "loss": 1.1021,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019354957149404008,
+      "loss": 1.3433,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019352846335667895,
+      "loss": 1.3782,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019350732189380882,
+      "loss": 1.1554,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934861471129627,
+      "loss": 1.2893,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934649390216854,
+      "loss": 1.118,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019344369762753374,
+      "loss": 1.3013,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019342242293807628,
+      "loss": 1.3452,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934011149608935,
+      "loss": 1.4487,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019337977370357772,
+      "loss": 1.3262,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019335839917373316,
+      "loss": 1.2729,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019333699137897585,
+      "loss": 1.2489,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019331555032693366,
+      "loss": 1.3548,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001932940760252464,
+      "loss": 1.3048,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019327256848156564,
+      "loss": 1.2581,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001932510277035548,
+      "loss": 1.1131,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001932294536988892,
+      "loss": 1.3438,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019320784647525595,
+      "loss": 1.3383,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019318620604035394,
+      "loss": 1.3406,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019316453240189406,
+      "loss": 1.4445,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001931428255675989,
+      "loss": 1.2059,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001931210855452029,
+      "loss": 1.1649,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019309931234245232,
+      "loss": 1.2059,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019307750596710528,
+      "loss": 1.2254,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930556664269317,
+      "loss": 1.3368,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019303379372971334,
+      "loss": 1.277,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930118878832437,
+      "loss": 1.2679,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019298994889532816,
+      "loss": 1.301,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929679767737839,
+      "loss": 1.1645,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929459715264399,
+      "loss": 1.2618,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019292393316113694,
+      "loss": 1.1408,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019290186168572756,
+      "loss": 1.3571,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019287975710807618,
+      "loss": 1.3265,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019285761943605898,
+      "loss": 1.305,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019283544867756389,
+      "loss": 1.2608,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928132448404907,
+      "loss": 1.2952,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019279100793275088,
+      "loss": 1.3283,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019276873796226785,
+      "loss": 1.3343,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019274643493697667,
+      "loss": 1.2268,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927240988648242,
+      "loss": 1.1538,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019270172975376912,
+      "loss": 1.2076,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019267932761178185,
+      "loss": 1.247,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926568924468446,
+      "loss": 1.2655,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019263442426695134,
+      "loss": 1.3585,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019261192308010777,
+      "loss": 1.3747,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925893888943314,
+      "loss": 1.2516,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925668217176515,
+      "loss": 1.2097,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019254422155810902,
+      "loss": 1.2374,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019252158842375676,
+      "loss": 1.3137,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001924989223226592,
+      "loss": 1.2894,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001924762232628926,
+      "loss": 1.1627,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019245349125254497,
+      "loss": 1.2759,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019243072629971603,
+      "loss": 1.2083,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019240792841251724,
+      "loss": 1.2526,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019238509759907183,
+      "loss": 1.2212,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019236223386751475,
+      "loss": 1.1826,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019233933722599263,
+      "loss": 1.367,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019231640768266388,
+      "loss": 1.3257,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019229344524569864,
+      "loss": 1.3204,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019227044992327873,
+      "loss": 1.2533,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019224742172359774,
+      "loss": 1.2269,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019222436065486088,
+      "loss": 1.2668,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922012667252852,
+      "loss": 1.2467,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019217813994309933,
+      "loss": 1.2551,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019215498031654375,
+      "loss": 1.1791,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019213178785387048,
+      "loss": 1.3202,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019210856256334337,
+      "loss": 1.2334,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920853044532379,
+      "loss": 1.3027,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019206201353184123,
+      "loss": 1.2659,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920386898074523,
+      "loss": 1.383,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019201533328838167,
+      "loss": 1.2923,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019199194398295157,
+      "loss": 1.2057,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019196852189949594,
+      "loss": 1.3711,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019194506704636047,
+      "loss": 1.2203,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019192157943190234,
+      "loss": 1.2881,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019189805906449057,
+      "loss": 1.1851,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019187450595250586,
+      "loss": 1.2847,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019185092010434043,
+      "loss": 1.2685,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019182730152839827,
+      "loss": 1.2061,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019180365023309504,
+      "loss": 1.3032,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019177996622685799,
+      "loss": 1.2469,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019175624951812606,
+      "loss": 1.1961,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019173250011534994,
+      "loss": 1.3513,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019170871802699177,
+      "loss": 1.2342,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916849032615255,
+      "loss": 1.2685,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019166105582743662,
+      "loss": 1.2194,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019163717573322236,
+      "loss": 1.1521,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916132629873915,
+      "loss": 1.2553,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019158931759846447,
+      "loss": 1.3083,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019156533957497343,
+      "loss": 1.3656,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000191541328925462,
+      "loss": 1.2475,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019151728565848557,
+      "loss": 1.2832,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019149320978261106,
+      "loss": 1.3392,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019146910130641705,
+      "loss": 1.2165,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019144496023849373,
+      "loss": 1.3949,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019142078658744292,
+      "loss": 1.2742,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019139658036187803,
+      "loss": 1.3465,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019137234157042405,
+      "loss": 1.2548,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019134807022171764,
+      "loss": 1.4378,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019132376632440695,
+      "loss": 1.3388,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019129942988715192,
+      "loss": 1.3212,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019127506091862388,
+      "loss": 1.1798,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019125065942750586,
+      "loss": 1.3099,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019122622542249248,
+      "loss": 1.213,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019120175891228986,
+      "loss": 1.1397,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911772599056158,
+      "loss": 1.2269,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019115272841119967,
+      "loss": 1.3249,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911281644377823,
+      "loss": 1.1296,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911035679941163,
+      "loss": 1.207,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019107893908896562,
+      "loss": 1.3753,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019105427773110593,
+      "loss": 1.1528,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910295839293244,
+      "loss": 1.3237,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019100485769241984,
+      "loss": 1.2555,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909800990292025,
+      "loss": 1.2762,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019095530794849426,
+      "loss": 1.2333,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909304844591285,
+      "loss": 1.2961,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019090562856995024,
+      "loss": 1.2277,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019088074028981595,
+      "loss": 1.1122,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019085581962759366,
+      "loss": 1.1174,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019083086659216295,
+      "loss": 1.1855,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000190805881192415,
+      "loss": 1.271,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019078086343725237,
+      "loss": 1.3975,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907558133355893,
+      "loss": 1.3431,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019073073089635148,
+      "loss": 1.3291,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070561612847612,
+      "loss": 1.1924,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000190680469040912,
+      "loss": 1.2245,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019065528964261935,
+      "loss": 1.113,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019063007794256994,
+      "loss": 1.1071,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019060483394974707,
+      "loss": 1.3126,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019057955767314556,
+      "loss": 1.184,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019055424912177164,
+      "loss": 1.1464,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019052890830464317,
+      "loss": 1.2027,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905035352307894,
+      "loss": 1.237,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019047812990925108,
+      "loss": 1.3356,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019045269234908057,
+      "loss": 1.272,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019042722255934155,
+      "loss": 1.1261,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019040172054910932,
+      "loss": 1.3207,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001903761863274706,
+      "loss": 1.3079,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019035061990352355,
+      "loss": 1.2828,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019032502128637788,
+      "loss": 1.2624,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019029939048515474,
+      "loss": 1.2862,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019027372750898673,
+      "loss": 1.2695,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019024803236701793,
+      "loss": 1.2618,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001902223050684039,
+      "loss": 1.3428,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019019654562231163,
+      "loss": 1.2756,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901707540379196,
+      "loss": 1.3188,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019014493032441766,
+      "loss": 1.2962,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019011907449100718,
+      "loss": 1.1797,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019009318654690098,
+      "loss": 1.4172,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900672665013233,
+      "loss": 1.2961,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019004131436350978,
+      "loss": 1.1933,
+      "step": 917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019001533014270755,
+      "loss": 1.365,
+      "step": 918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899893138481752,
+      "loss": 1.1964,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899632654891826,
+      "loss": 1.1528,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018993718507501125,
+      "loss": 1.3675,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018991107261495394,
+      "loss": 1.2004,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018988492811831487,
+      "loss": 1.1461,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018985875159440973,
+      "loss": 1.2243,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018983254305256558,
+      "loss": 1.4954,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898063025021209,
+      "loss": 1.3015,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018978002995242548,
+      "loss": 1.2666,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018975372541284074,
+      "loss": 1.2546,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897273888927393,
+      "loss": 1.1852,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018970102040150516,
+      "loss": 1.2762,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896746199485339,
+      "loss": 1.2589,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018964818754323227,
+      "loss": 1.3692,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018962172319501863,
+      "loss": 1.1857,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895952269133225,
+      "loss": 1.2242,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018956869870758488,
+      "loss": 1.3045,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895421385872582,
+      "loss": 1.2754,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895155465618062,
+      "loss": 1.2937,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894889226407039,
+      "loss": 1.238,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894622668334379,
+      "loss": 1.213,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018943557914950597,
+      "loss": 1.2474,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018940885959841736,
+      "loss": 1.0714,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018938210818969254,
+      "loss": 1.2348,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018935532493286349,
+      "loss": 1.1682,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018932850983747343,
+      "loss": 1.2185,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018930166291307695,
+      "loss": 1.2796,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018927478416924003,
+      "loss": 1.2764,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018924787361553988,
+      "loss": 1.1953,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018922093126156516,
+      "loss": 1.129,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018919395711691577,
+      "loss": 1.1844,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000189166951191203,
+      "loss": 1.3591,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018913991349404948,
+      "loss": 1.2302,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018911284403508904,
+      "loss": 1.2529,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018908574282396697,
+      "loss": 1.2677,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018905860987033978,
+      "loss": 1.2345,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018903144518387537,
+      "loss": 1.1932,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018900424877425287,
+      "loss": 1.2449,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018897702065116275,
+      "loss": 1.3848,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001889497608243067,
+      "loss": 1.1801,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018892246930339793,
+      "loss": 1.248,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018889514609816068,
+      "loss": 1.2639,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018886779121833063,
+      "loss": 1.3458,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018884040467365468,
+      "loss": 1.2077,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888129864738911,
+      "loss": 1.3178,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018878553662880933,
+      "loss": 1.2055,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018875805514819013,
+      "loss": 1.17,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018873054204182559,
+      "loss": 1.1826,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188702997319519,
+      "loss": 1.3194,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886754209910849,
+      "loss": 1.1484,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018864781306634915,
+      "loss": 1.2449,
+      "step": 969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018862017355514884,
+      "loss": 1.252,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018859250246733232,
+      "loss": 1.3488,
+      "step": 971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885647998127592,
+      "loss": 1.2597,
+      "step": 972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018853706560130026,
+      "loss": 1.2552,
+      "step": 973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018850929984283766,
+      "loss": 1.2241,
+      "step": 974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018848150254726473,
+      "loss": 1.1664,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018845367372448602,
+      "loss": 1.2568,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018842581338441731,
+      "loss": 1.1739,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018839792153698567,
+      "loss": 1.3899,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018836999819212932,
+      "loss": 1.3008,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018834204335979777,
+      "loss": 1.2617,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018831405704995173,
+      "loss": 1.37,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018828603927256309,
+      "loss": 1.3474,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018825799003761498,
+      "loss": 1.2575,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018822990935510173,
+      "loss": 1.4526,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018820179723502893,
+      "loss": 1.2221,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018817365368741328,
+      "loss": 1.2262,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018814547872228273,
+      "loss": 1.267,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881172723496764,
+      "loss": 1.1593,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018808903457964468,
+      "loss": 1.1345,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018806076542224903,
+      "loss": 1.2518,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018803246488756215,
+      "loss": 1.2693,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018800413298566794,
+      "loss": 1.3409,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018797576972666145,
+      "loss": 1.2885,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879473751206489,
+      "loss": 1.2941,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018791894917774774,
+      "loss": 1.2552,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018789049190808644,
+      "loss": 1.2884,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018786200332180476,
+      "loss": 1.2446,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018783348342905364,
+      "loss": 1.2551,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018780493223999507,
+      "loss": 1.383,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018777634976480226,
+      "loss": 1.2988,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018774773601365951,
+      "loss": 1.2892,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018771909099676238,
+      "loss": 1.3437,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018769041472431737,
+      "loss": 1.2857,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018766170720654237,
+      "loss": 1.1264,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018763296845366621,
+      "loss": 1.279,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876041984759289,
+      "loss": 1.4246,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018757539728358162,
+      "loss": 1.2674,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018754656488688664,
+      "loss": 1.3967,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018751770129611732,
+      "loss": 1.2218,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018748880652155818,
+      "loss": 1.2084,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018745988057350486,
+      "loss": 1.2892,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018743092346226403,
+      "loss": 1.2815,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018740193519815355,
+      "loss": 1.3005,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018737291579150239,
+      "loss": 1.2216,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018734386525265045,
+      "loss": 1.2859,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000187314783591949,
+      "loss": 1.2176,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018728567081976013,
+      "loss": 1.2058,
+      "step": 1017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018725652694645716,
+      "loss": 1.231,
+      "step": 1018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001872273519824245,
+      "loss": 1.3593,
+      "step": 1019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018719814593805753,
+      "loss": 1.3171,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018716890882376284,
+      "loss": 1.2533,
+      "step": 1021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000187139640649958,
+      "loss": 1.2249,
+      "step": 1022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018711034142707172,
+      "loss": 1.1695,
+      "step": 1023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001870810111655436,
+      "loss": 1.1618,
+      "step": 1024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018705164987582458,
+      "loss": 1.2022,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001870222575683764,
+      "loss": 1.0753,
+      "step": 1026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018699283425367195,
+      "loss": 1.1242,
+      "step": 1027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018696337994219525,
+      "loss": 1.1503,
+      "step": 1028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018693389464444123,
+      "loss": 1.2885,
+      "step": 1029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001869043783709159,
+      "loss": 1.3526,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018687483113213635,
+      "loss": 1.2479,
+      "step": 1031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868452529386307,
+      "loss": 1.3107,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018681564380093802,
+      "loss": 1.3351,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018678600372960854,
+      "loss": 1.2709,
+      "step": 1034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018675633273520338,
+      "loss": 1.148,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018672663082829473,
+      "loss": 1.2832,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018669689801946584,
+      "loss": 1.3468,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001866671343193109,
+      "loss": 1.0989,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001866373397384351,
+      "loss": 1.3311,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018660751428745473,
+      "loss": 1.2225,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018657765797699697,
+      "loss": 1.2158,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018654777081770013,
+      "loss": 1.3245,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865178528202133,
+      "loss": 1.2162,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018648790399519678,
+      "loss": 1.3197,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018645792435332173,
+      "loss": 1.1523,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001864279139052703,
+      "loss": 1.2849,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018639787266173566,
+      "loss": 1.2873,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018636780063342192,
+      "loss": 1.2718,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018633769783104416,
+      "loss": 1.226,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018630756426532842,
+      "loss": 1.2631,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018627739994701175,
+      "loss": 1.1056,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018624720488684212,
+      "loss": 1.3471,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018621697909557844,
+      "loss": 1.1636,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018618672258399058,
+      "loss": 1.2094,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018615643536285933,
+      "loss": 1.2115,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001861261174429765,
+      "loss": 1.3,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018609576883514483,
+      "loss": 1.088,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018606538955017788,
+      "loss": 1.1683,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018603497959890025,
+      "loss": 1.2028,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001860045389921474,
+      "loss": 1.2272,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001859740677407658,
+      "loss": 1.1537,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018594356585561276,
+      "loss": 1.2668,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018591303334755655,
+      "loss": 1.1853,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018588247022747632,
+      "loss": 1.182,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858518765062621,
+      "loss": 1.1811,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858212521948149,
+      "loss": 1.3762,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018579059730404663,
+      "loss": 1.2155,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018575991184488,
+      "loss": 1.1876,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018572919582824867,
+      "loss": 1.1471,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018569844926509728,
+      "loss": 1.2537,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018566767216638116,
+      "loss": 1.1901,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018563686454306665,
+      "loss": 1.2109,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018560602640613097,
+      "loss": 1.3883,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018557515776656218,
+      "loss": 1.2913,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018554425863535917,
+      "loss": 1.1764,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018551332902353175,
+      "loss": 1.1987,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018548236894210064,
+      "loss": 1.1131,
+      "step": 1077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018545137840209725,
+      "loss": 1.2708,
+      "step": 1078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000185420357414564,
+      "loss": 1.1919,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853893059905541,
+      "loss": 1.218,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018535822414113165,
+      "loss": 1.1838,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018532711187737149,
+      "loss": 1.1812,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018529596921035936,
+      "loss": 1.301,
+      "step": 1083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018526479615119188,
+      "loss": 1.0555,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001852335927109764,
+      "loss": 1.3771,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001852023589008312,
+      "loss": 1.235,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018517109473188527,
+      "loss": 1.1915,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018513980021527847,
+      "loss": 1.2075,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018510847536216154,
+      "loss": 1.1858,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850771201836959,
+      "loss": 1.3074,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850457346910539,
+      "loss": 1.1771,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850143188954186,
+      "loss": 1.2522,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018498287280798389,
+      "loss": 1.212,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018495139643995447,
+      "loss": 1.3738,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849198898025458,
+      "loss": 1.2066,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018488835290698412,
+      "loss": 1.2706,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018485678576450653,
+      "loss": 1.2891,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018482518838636077,
+      "loss": 1.2844,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847935607838055,
+      "loss": 1.3723,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018476190296811003,
+      "loss": 1.2969,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018473021495055454,
+      "loss": 1.1631,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018469849674242986,
+      "loss": 1.2794,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018466674835503763,
+      "loss": 1.225,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846349697996903,
+      "loss": 1.2541,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000184603161087711,
+      "loss": 1.2703,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018457132223043363,
+      "loss": 1.3169,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018453945323920276,
+      "loss": 1.0742,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018450755412537383,
+      "loss": 1.15,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018447562490031297,
+      "loss": 1.2721,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001844436655753969,
+      "loss": 1.2389,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018441167616201328,
+      "loss": 1.2704,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001843796566715603,
+      "loss": 1.243,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018434760711544707,
+      "loss": 1.1161,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001843155275050932,
+      "loss": 1.3412,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018428341785192915,
+      "loss": 1.368,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018425127816739604,
+      "loss": 1.2702,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842191084629457,
+      "loss": 1.2876,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018418690875004058,
+      "loss": 1.3158,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000184154679040154,
+      "loss": 1.172,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001841224193447698,
+      "loss": 1.2542,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018409012967538253,
+      "loss": 1.2386,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018405781004349753,
+      "loss": 1.3701,
+      "step": 1122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018402546046063068,
+      "loss": 1.2703,
+      "step": 1123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018399308093830863,
+      "loss": 1.2645,
+      "step": 1124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018396067148806865,
+      "loss": 1.2255,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001839282321214587,
+      "loss": 1.3434,
+      "step": 1126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018389576285003732,
+      "loss": 1.2211,
+      "step": 1127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018386326368537384,
+      "loss": 1.329,
+      "step": 1128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018383073463904811,
+      "loss": 1.1497,
+      "step": 1129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018379817572265075,
+      "loss": 1.2968,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018376558694778292,
+      "loss": 1.1448,
+      "step": 1131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018373296832605647,
+      "loss": 1.3169,
+      "step": 1132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018370031986909385,
+      "loss": 1.2524,
+      "step": 1133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018366764158852825,
+      "loss": 1.2234,
+      "step": 1134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836349334960033,
+      "loss": 1.2797,
+      "step": 1135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836021956031734,
+      "loss": 1.2387,
+      "step": 1136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018356942792170351,
+      "loss": 1.3,
+      "step": 1137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018353663046326922,
+      "loss": 1.1838,
+      "step": 1138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018350380323955674,
+      "loss": 1.243,
+      "step": 1139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018347094626226283,
+      "loss": 1.3304,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000183438059543095,
+      "loss": 1.1256,
+      "step": 1141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018340514309377114,
+      "loss": 1.3568,
+      "step": 1142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018337219692601985,
+      "loss": 1.1598,
+      "step": 1143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001833392210515804,
+      "loss": 1.3103,
+      "step": 1144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018330621548220247,
+      "loss": 1.3106,
+      "step": 1145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018327318022964644,
+      "loss": 1.1189,
+      "step": 1146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018324011530568326,
+      "loss": 1.1275,
+      "step": 1147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018320702072209438,
+      "loss": 1.3179,
+      "step": 1148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018317389649067192,
+      "loss": 1.2611,
+      "step": 1149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018314074262321846,
+      "loss": 1.1585,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018310755913154725,
+      "loss": 1.17,
+      "step": 1151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018307434602748198,
+      "loss": 1.1732,
+      "step": 1152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018304110332285694,
+      "loss": 1.2953,
+      "step": 1153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018300783102951704,
+      "loss": 1.2144,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018297452915931763,
+      "loss": 1.1426,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001829411977241246,
+      "loss": 1.2511,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018290783673581445,
+      "loss": 1.2518,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018287444620627418,
+      "loss": 1.273,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828410261474013,
+      "loss": 1.2094,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018280757657110382,
+      "loss": 1.2331,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827740974893003,
+      "loss": 1.1651,
+      "step": 1161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018274058891391986,
+      "loss": 1.1944,
+      "step": 1162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000182707050856902,
+      "loss": 1.1929,
+      "step": 1163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018267348333019686,
+      "loss": 1.2302,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018263988634576497,
+      "loss": 1.2613,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018260625991557748,
+      "loss": 1.2728,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001825726040516159,
+      "loss": 1.1563,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001825389187658723,
+      "loss": 1.1817,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018250520407034926,
+      "loss": 1.0805,
+      "step": 1169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018247145997705975,
+      "loss": 1.3775,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018243768649802732,
+      "loss": 1.3108,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018240388364528586,
+      "loss": 1.1651,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018237005143087988,
+      "loss": 1.312,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018233618986686424,
+      "loss": 1.1943,
+      "step": 1174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001823022989653043,
+      "loss": 1.2351,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018226837873827585,
+      "loss": 1.2237,
+      "step": 1176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018223442919786516,
+      "loss": 1.1805,
+      "step": 1177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001822004503561689,
+      "loss": 1.2117,
+      "step": 1178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018216644222529427,
+      "loss": 1.1895,
+      "step": 1179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001821324048173588,
+      "loss": 1.2199,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018209833814449052,
+      "loss": 1.3299,
+      "step": 1181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018206424221882782,
+      "loss": 1.1756,
+      "step": 1182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018203011705251963,
+      "loss": 1.214,
+      "step": 1183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001819959626577252,
+      "loss": 1.2791,
+      "step": 1184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018196177904661415,
+      "loss": 1.343,
+      "step": 1185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001819275662313667,
+      "loss": 1.2274,
+      "step": 1186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018189332422417325,
+      "loss": 1.1298,
+      "step": 1187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018185905303723482,
+      "loss": 1.2298,
+      "step": 1188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018182475268276263,
+      "loss": 1.3459,
+      "step": 1189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018179042317297844,
+      "loss": 1.3471,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018175606452011425,
+      "loss": 1.1728,
+      "step": 1191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018172167673641263,
+      "loss": 1.2208,
+      "step": 1192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018168725983412637,
+      "loss": 1.1564,
+      "step": 1193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018165281382551872,
+      "loss": 1.2529,
+      "step": 1194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001816183387228633,
+      "loss": 1.2068,
+      "step": 1195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018158383453844402,
+      "loss": 1.2711,
+      "step": 1196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018154930128455523,
+      "loss": 1.2721,
+      "step": 1197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018151473897350164,
+      "loss": 1.2419,
+      "step": 1198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018148014761759827,
+      "loss": 1.3283,
+      "step": 1199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001814455272291705,
+      "loss": 1.2221,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001814108778205541,
+      "loss": 1.341,
+      "step": 1201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018137619940409506,
+      "loss": 1.3237,
+      "step": 1202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018134149199214988,
+      "loss": 1.3324,
+      "step": 1203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018130675559708525,
+      "loss": 1.1857,
+      "step": 1204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018127199023127823,
+      "loss": 1.3635,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018123719590711624,
+      "loss": 1.28,
+      "step": 1206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181202372636997,
+      "loss": 1.2417,
+      "step": 1207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018116752043332847,
+      "loss": 1.1998,
+      "step": 1208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018113263930852902,
+      "loss": 1.2496,
+      "step": 1209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810977292750273,
+      "loss": 1.2237,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810627903452622,
+      "loss": 1.2938,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181027822531683,
+      "loss": 1.1901,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001809928258467492,
+      "loss": 1.2315,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018095780030293062,
+      "loss": 1.1738,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001809227459127074,
+      "loss": 1.1581,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018088766268856984,
+      "loss": 1.2853,
+      "step": 1216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018085255064301858,
+      "loss": 1.2228,
+      "step": 1217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001808174097885646,
+      "loss": 1.1582,
+      "step": 1218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018078224013772903,
+      "loss": 1.3397,
+      "step": 1219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018074704170304334,
+      "loss": 1.1083,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018071181449704927,
+      "loss": 1.2055,
+      "step": 1221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018067655853229865,
+      "loss": 1.1628,
+      "step": 1222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001806412738213538,
+      "loss": 1.1761,
+      "step": 1223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018060596037678707,
+      "loss": 1.1952,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001805706182111812,
+      "loss": 1.2856,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018053524733712908,
+      "loss": 1.0367,
+      "step": 1226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018049984776723384,
+      "loss": 1.1357,
+      "step": 1227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018046441951410883,
+      "loss": 1.1493,
+      "step": 1228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001804289625903777,
+      "loss": 1.2594,
+      "step": 1229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803934770086742,
+      "loss": 1.2872,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018035796278164233,
+      "loss": 1.2739,
+      "step": 1231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018032241992193635,
+      "loss": 1.1775,
+      "step": 1232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018028684844222066,
+      "loss": 1.2807,
+      "step": 1233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018025124835516988,
+      "loss": 1.2762,
+      "step": 1234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018021561967346882,
+      "loss": 1.2427,
+      "step": 1235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018017996240981253,
+      "loss": 1.2255,
+      "step": 1236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801442765769061,
+      "loss": 1.2736,
+      "step": 1237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018010856218746496,
+      "loss": 0.9902,
+      "step": 1238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018007281925421463,
+      "loss": 1.1292,
+      "step": 1239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018003704778989086,
+      "loss": 1.12,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001800012478072395,
+      "loss": 1.334,
+      "step": 1241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017996541931901652,
+      "loss": 1.2327,
+      "step": 1242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017992956233798825,
+      "loss": 1.1804,
+      "step": 1243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798936768769309,
+      "loss": 1.1307,
+      "step": 1244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798577629486311,
+      "loss": 1.1742,
+      "step": 1245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017982182056588535,
+      "loss": 1.2913,
+      "step": 1246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017978584974150054,
+      "loss": 1.1641,
+      "step": 1247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017974985048829352,
+      "loss": 1.2154,
+      "step": 1248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017971382281909134,
+      "loss": 1.1288,
+      "step": 1249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001796777667467312,
+      "loss": 1.265,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017964168228406035,
+      "loss": 1.3487,
+      "step": 1251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001796055694439362,
+      "loss": 1.206,
+      "step": 1252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017956942823922625,
+      "loss": 1.199,
+      "step": 1253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001795332586828082,
+      "loss": 1.1803,
+      "step": 1254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017949706078756967,
+      "loss": 1.0641,
+      "step": 1255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017946083456640851,
+      "loss": 1.1758,
+      "step": 1256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017942458003223266,
+      "loss": 1.2883,
+      "step": 1257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793882971979601,
+      "loss": 1.3222,
+      "step": 1258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793519860765189,
+      "loss": 1.2017,
+      "step": 1259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017931564668084726,
+      "loss": 1.1392,
+      "step": 1260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017927927902389342,
+      "loss": 1.0438,
+      "step": 1261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017924288311861565,
+      "loss": 1.2606,
+      "step": 1262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017920645897798236,
+      "loss": 1.1746,
+      "step": 1263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017917000661497195,
+      "loss": 1.2884,
+      "step": 1264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001791335260425729,
+      "loss": 1.1281,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017909701727378385,
+      "loss": 1.171,
+      "step": 1266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017906048032161328,
+      "loss": 1.3053,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017902391519907987,
+      "loss": 1.2621,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001789873219192123,
+      "loss": 1.3831,
+      "step": 1269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001789507004950492,
+      "loss": 1.3112,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.2407,
+      "step": 1271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017887737326604156,
+      "loss": 1.2135,
+      "step": 1272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001788406674873245,
+      "loss": 1.3374,
+      "step": 1273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000178803933616567,
+      "loss": 1.2058,
+      "step": 1274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017876717166685787,
+      "loss": 1.2151,
+      "step": 1275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017873038165129587,
+      "loss": 1.2119,
+      "step": 1276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017869356358298988,
+      "loss": 1.4366,
+      "step": 1277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017865671747505858,
+      "loss": 1.1781,
+      "step": 1278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017861984334063084,
+      "loss": 1.1427,
+      "step": 1279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017858294119284544,
+      "loss": 1.1903,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001785460110448511,
+      "loss": 1.2264,
+      "step": 1281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017850905290980656,
+      "loss": 1.2178,
+      "step": 1282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001784720668008805,
+      "loss": 1.2567,
+      "step": 1283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017843505273125164,
+      "loss": 1.2121,
+      "step": 1284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001783980107141086,
+      "loss": 1.3689,
+      "step": 1285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017836094076264996,
+      "loss": 1.2936,
+      "step": 1286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017832384289008423,
+      "loss": 1.2547,
+      "step": 1287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017828671710963,
+      "loss": 1.0993,
+      "step": 1288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001782495634345156,
+      "loss": 1.1734,
+      "step": 1289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017821238187797948,
+      "loss": 1.2159,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017817517245326996,
+      "loss": 1.3679,
+      "step": 1291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017813793517364523,
+      "loss": 1.2212,
+      "step": 1292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017810067005237347,
+      "loss": 1.2879,
+      "step": 1293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001780633771027328,
+      "loss": 1.3256,
+      "step": 1294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017802605633801118,
+      "loss": 1.1026,
+      "step": 1295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001779887077715066,
+      "loss": 1.2733,
+      "step": 1296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017795133141652677,
+      "loss": 1.1758,
+      "step": 1297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001779139272863895,
+      "loss": 1.2666,
+      "step": 1298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017787649539442236,
+      "loss": 1.318,
+      "step": 1299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001778390357539629,
+      "loss": 1.1325,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017780154837835848,
+      "loss": 1.3479,
+      "step": 1301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017776403328096643,
+      "loss": 1.2396,
+      "step": 1302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017772649047515383,
+      "loss": 1.3131,
+      "step": 1303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017768891997429776,
+      "loss": 1.209,
+      "step": 1304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017765132179178514,
+      "loss": 1.3803,
+      "step": 1305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017761369594101266,
+      "loss": 1.2069,
+      "step": 1306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017757604243538703,
+      "loss": 1.3159,
+      "step": 1307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017753836128832465,
+      "loss": 1.3416,
+      "step": 1308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001775006525132519,
+      "loss": 1.1981,
+      "step": 1309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001774629161236049,
+      "loss": 1.1581,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001774251521328297,
+      "loss": 1.2708,
+      "step": 1311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773873605543821,
+      "loss": 1.1515,
+      "step": 1312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773495414017278,
+      "loss": 1.2559,
+      "step": 1313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773116946883423,
+      "loss": 1.1202,
+      "step": 1314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001772738204277109,
+      "loss": 1.196,
+      "step": 1315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017723591863332873,
+      "loss": 1.2173,
+      "step": 1316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017719798931870075,
+      "loss": 1.2556,
+      "step": 1317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001771600324973417,
+      "loss": 1.2175,
+      "step": 1318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017712204818277614,
+      "loss": 1.1328,
+      "step": 1319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017708403638853838,
+      "loss": 1.2068,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017704599712817262,
+      "loss": 1.2144,
+      "step": 1321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017700793041523273,
+      "loss": 1.2606,
+      "step": 1322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017696983626328244,
+      "loss": 1.132,
+      "step": 1323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017693171468589523,
+      "loss": 1.2381,
+      "step": 1324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017689356569665433,
+      "loss": 1.2181,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017685538930915283,
+      "loss": 1.3792,
+      "step": 1326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017681718553699341,
+      "loss": 1.1832,
+      "step": 1327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017677895439378874,
+      "loss": 1.2582,
+      "step": 1328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000176740695893161,
+      "loss": 1.3206,
+      "step": 1329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017670241004874232,
+      "loss": 1.2381,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001766640968741744,
+      "loss": 1.2651,
+      "step": 1331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017662575638310887,
+      "loss": 1.0618,
+      "step": 1332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017658738858920687,
+      "loss": 1.1737,
+      "step": 1333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017654899350613948,
+      "loss": 1.297,
+      "step": 1334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017651057114758734,
+      "loss": 1.1929,
+      "step": 1335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764721215272409,
+      "loss": 1.2204,
+      "step": 1336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017643364465880036,
+      "loss": 1.3038,
+      "step": 1337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017639514055597548,
+      "loss": 1.2502,
+      "step": 1338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763566092324859,
+      "loss": 1.2261,
+      "step": 1339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763180507020608,
+      "loss": 1.0371,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017627946497843916,
+      "loss": 1.2524,
+      "step": 1341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762408520753696,
+      "loss": 1.2008,
+      "step": 1342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017620221200661046,
+      "loss": 1.3663,
+      "step": 1343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017616354478592973,
+      "loss": 1.2386,
+      "step": 1344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001761248504271051,
+      "loss": 1.1968,
+      "step": 1345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017608612894392392,
+      "loss": 1.1832,
+      "step": 1346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017604738035018316,
+      "loss": 1.2211,
+      "step": 1347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017600860465968952,
+      "loss": 1.216,
+      "step": 1348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017596980188625932,
+      "loss": 1.2222,
+      "step": 1349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017593097204371852,
+      "loss": 1.1297,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001758921151459028,
+      "loss": 1.1722,
+      "step": 1351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017585323120665732,
+      "loss": 1.1123,
+      "step": 1352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017581432023983706,
+      "loss": 1.2245,
+      "step": 1353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017577538225930654,
+      "loss": 1.1896,
+      "step": 1354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017573641727893985,
+      "loss": 1.1446,
+      "step": 1355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001756974253126208,
+      "loss": 1.1731,
+      "step": 1356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001756584063742428,
+      "loss": 1.1536,
+      "step": 1357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017561936047770879,
+      "loss": 1.2709,
+      "step": 1358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001755802876369314,
+      "loss": 1.1872,
+      "step": 1359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001755411878658329,
+      "loss": 1.1587,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017550206117834498,
+      "loss": 1.1885,
+      "step": 1361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017546290758840906,
+      "loss": 1.3322,
+      "step": 1362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017542372710997617,
+      "loss": 1.1306,
+      "step": 1363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001753845197570068,
+      "loss": 1.201,
+      "step": 1364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017534528554347115,
+      "loss": 1.0762,
+      "step": 1365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017530602448334888,
+      "loss": 1.2592,
+      "step": 1366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017526673659062922,
+      "loss": 1.1118,
+      "step": 1367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017522742187931108,
+      "loss": 1.2199,
+      "step": 1368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017518808036340279,
+      "loss": 1.1314,
+      "step": 1369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017514871205692232,
+      "loss": 1.2075,
+      "step": 1370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017510931697389714,
+      "loss": 1.2119,
+      "step": 1371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017506989512836425,
+      "loss": 1.229,
+      "step": 1372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017503044653437025,
+      "loss": 1.1953,
+      "step": 1373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001749909712059712,
+      "loss": 1.0281,
+      "step": 1374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017495146915723272,
+      "loss": 1.1775,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017491194040222994,
+      "loss": 1.2288,
+      "step": 1376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748723849550476,
+      "loss": 1.2656,
+      "step": 1377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748328028297797,
+      "loss": 1.2991,
+      "step": 1378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017479319404053003,
+      "loss": 1.3449,
+      "step": 1379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017475355860141172,
+      "loss": 1.2125,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017471389652654745,
+      "loss": 1.3109,
+      "step": 1381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017467420783006933,
+      "loss": 1.4411,
+      "step": 1382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017463449252611905,
+      "loss": 1.1428,
+      "step": 1383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017459475062884776,
+      "loss": 1.1977,
+      "step": 1384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017455498215241598,
+      "loss": 1.2016,
+      "step": 1385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001745151871109938,
+      "loss": 1.1913,
+      "step": 1386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744753655187608,
+      "loss": 1.1875,
+      "step": 1387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017443551738990595,
+      "loss": 1.1636,
+      "step": 1388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001743956427386277,
+      "loss": 1.1997,
+      "step": 1389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017435574157913394,
+      "loss": 1.2893,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017431581392564197,
+      "loss": 1.18,
+      "step": 1391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017427585979237867,
+      "loss": 1.1265,
+      "step": 1392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017423587919358023,
+      "loss": 1.1473,
+      "step": 1393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017419587214349225,
+      "loss": 1.3176,
+      "step": 1394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017415583865636983,
+      "loss": 1.1197,
+      "step": 1395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001741157787464775,
+      "loss": 1.2416,
+      "step": 1396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017407569242808917,
+      "loss": 1.2736,
+      "step": 1397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740355797154881,
+      "loss": 1.0422,
+      "step": 1398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017399544062296707,
+      "loss": 1.1918,
+      "step": 1399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017395527516482818,
+      "loss": 1.2197,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017391508335538294,
+      "loss": 1.1656,
+      "step": 1401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017387486520895228,
+      "loss": 1.2237,
+      "step": 1402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001738346207398665,
+      "loss": 1.3257,
+      "step": 1403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017379434996246524,
+      "loss": 1.2784,
+      "step": 1404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017375405289109757,
+      "loss": 1.2915,
+      "step": 1405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017371372954012187,
+      "loss": 1.0518,
+      "step": 1406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017367337992390595,
+      "loss": 1.2375,
+      "step": 1407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017363300405682697,
+      "loss": 1.2821,
+      "step": 1408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017359260195327135,
+      "loss": 1.1144,
+      "step": 1409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173552173627635,
+      "loss": 1.1843,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017351171909432304,
+      "loss": 1.3609,
+      "step": 1411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017347123836775006,
+      "loss": 1.1684,
+      "step": 1412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017343073146233983,
+      "loss": 1.1361,
+      "step": 1413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733901983925256,
+      "loss": 1.1992,
+      "step": 1414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017334963917274985,
+      "loss": 1.0418,
+      "step": 1415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733090538174644,
+      "loss": 1.1364,
+      "step": 1416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017326844234113038,
+      "loss": 1.2021,
+      "step": 1417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017322780475821824,
+      "loss": 1.1642,
+      "step": 1418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001731871410832077,
+      "loss": 1.2481,
+      "step": 1419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001731464513305878,
+      "loss": 1.1703,
+      "step": 1420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017310573551485692,
+      "loss": 1.1008,
+      "step": 1421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017306499365052266,
+      "loss": 1.2727,
+      "step": 1422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017302422575210186,
+      "loss": 1.2382,
+      "step": 1423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017298343183412077,
+      "loss": 1.1945,
+      "step": 1424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017294261191111484,
+      "loss": 1.2835,
+      "step": 1425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017290176599762873,
+      "loss": 1.3709,
+      "step": 1426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017286089410821641,
+      "loss": 1.2991,
+      "step": 1427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017281999625744118,
+      "loss": 1.3137,
+      "step": 1428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727790724598755,
+      "loss": 1.1907,
+      "step": 1429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017273812273010104,
+      "loss": 1.1794,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017269714708270884,
+      "loss": 1.1233,
+      "step": 1431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017265614553229903,
+      "loss": 1.2167,
+      "step": 1432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017261511809348105,
+      "loss": 1.2112,
+      "step": 1433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725740647808736,
+      "loss": 1.1812,
+      "step": 1434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017253298560910457,
+      "loss": 1.1797,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017249188059281098,
+      "loss": 1.0878,
+      "step": 1436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017245074974663914,
+      "loss": 1.191,
+      "step": 1437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017240959308524457,
+      "loss": 1.214,
+      "step": 1438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017236841062329196,
+      "loss": 1.2253,
+      "step": 1439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017232720237545524,
+      "loss": 1.1965,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001722859683564174,
+      "loss": 1.2632,
+      "step": 1441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017224470858087076,
+      "loss": 1.2249,
+      "step": 1442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017220342306351679,
+      "loss": 1.0468,
+      "step": 1443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017216211181906606,
+      "loss": 1.276,
+      "step": 1444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001721207748622383,
+      "loss": 1.2347,
+      "step": 1445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017207941220776256,
+      "loss": 1.158,
+      "step": 1446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017203802387037688,
+      "loss": 0.9859,
+      "step": 1447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001719966098648285,
+      "loss": 1.1001,
+      "step": 1448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017195517020587378,
+      "loss": 1.3223,
+      "step": 1449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017191370490827836,
+      "loss": 1.1533,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017187221398681684,
+      "loss": 1.2613,
+      "step": 1451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017183069745627298,
+      "loss": 1.1942,
+      "step": 1452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017178915533143979,
+      "loss": 1.1766,
+      "step": 1453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017174758762711925,
+      "loss": 1.1639,
+      "step": 1454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017170599435812253,
+      "loss": 1.1171,
+      "step": 1455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017166437553926993,
+      "loss": 1.2142,
+      "step": 1456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017162273118539082,
+      "loss": 1.258,
+      "step": 1457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001715810613113236,
+      "loss": 1.12,
+      "step": 1458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017153936593191594,
+      "loss": 1.2543,
+      "step": 1459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001714976450620244,
+      "loss": 1.1931,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017145589871651477,
+      "loss": 1.2134,
+      "step": 1461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001714141269102618,
+      "loss": 1.1229,
+      "step": 1462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017137232965814948,
+      "loss": 1.0308,
+      "step": 1463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017133050697507068,
+      "loss": 1.1187,
+      "step": 1464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017128865887592739,
+      "loss": 1.1222,
+      "step": 1465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017124678537563072,
+      "loss": 1.2002,
+      "step": 1466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001712048864891008,
+      "loss": 1.2559,
+      "step": 1467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017116296223126676,
+      "loss": 1.1725,
+      "step": 1468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001711210126170668,
+      "loss": 1.2989,
+      "step": 1469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001710790376614482,
+      "loss": 1.2134,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017103703737936718,
+      "loss": 1.2327,
+      "step": 1471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017099501178578907,
+      "loss": 0.9165,
+      "step": 1472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017095296089568817,
+      "loss": 1.1552,
+      "step": 1473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001709108847240478,
+      "loss": 1.329,
+      "step": 1474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001708687832858603,
+      "loss": 1.1479,
+      "step": 1475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000170826656596127,
+      "loss": 1.1453,
+      "step": 1476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707845046698582,
+      "loss": 1.2917,
+      "step": 1477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017074232752207328,
+      "loss": 1.2175,
+      "step": 1478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017070012516780055,
+      "loss": 1.0365,
+      "step": 1479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017065789762207726,
+      "loss": 1.278,
+      "step": 1480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017061564489994968,
+      "loss": 1.1648,
+      "step": 1481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001705733670164731,
+      "loss": 1.3297,
+      "step": 1482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001705310639867117,
+      "loss": 1.403,
+      "step": 1483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017048873582573864,
+      "loss": 1.1626,
+      "step": 1484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017044638254863604,
+      "loss": 1.1389,
+      "step": 1485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017040400417049495,
+      "loss": 1.3136,
+      "step": 1486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017036160070641544,
+      "loss": 1.3039,
+      "step": 1487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001703191721715064,
+      "loss": 1.1937,
+      "step": 1488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017027671858088576,
+      "loss": 1.0943,
+      "step": 1489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001702342399496803,
+      "loss": 1.1182,
+      "step": 1490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017019173629302577,
+      "loss": 1.2138,
+      "step": 1491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017014920762606687,
+      "loss": 1.1032,
+      "step": 1492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017010665396395704,
+      "loss": 1.0061,
+      "step": 1493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017006407532185887,
+      "loss": 1.2154,
+      "step": 1494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017002147171494365,
+      "loss": 1.2172,
+      "step": 1495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001699788431583917,
+      "loss": 1.2804,
+      "step": 1496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016993618966739215,
+      "loss": 1.3163,
+      "step": 1497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016989351125714305,
+      "loss": 1.2119,
+      "step": 1498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016985080794285128,
+      "loss": 1.2323,
+      "step": 1499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001698080797397327,
+      "loss": 1.1655,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001697653266630119,
+      "loss": 1.1004,
+      "step": 1501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016972254872792245,
+      "loss": 1.3874,
+      "step": 1502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016967974594970673,
+      "loss": 1.2388,
+      "step": 1503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016963691834361597,
+      "loss": 1.275,
+      "step": 1504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016959406592491024,
+      "loss": 1.2606,
+      "step": 1505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016955118870885843,
+      "loss": 1.3397,
+      "step": 1506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016950828671073834,
+      "loss": 1.2047,
+      "step": 1507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016946535994583656,
+      "loss": 1.1473,
+      "step": 1508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016942240842944848,
+      "loss": 1.2658,
+      "step": 1509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016937943217687832,
+      "loss": 1.0781,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016933643120343916,
+      "loss": 1.1663,
+      "step": 1511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016929340552445282,
+      "loss": 1.1983,
+      "step": 1512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016925035515524996,
+      "loss": 1.2973,
+      "step": 1513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016920728011117007,
+      "loss": 1.2403,
+      "step": 1514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016916418040756134,
+      "loss": 1.1292,
+      "step": 1515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001691210560597808,
+      "loss": 1.2362,
+      "step": 1516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016907790708319434,
+      "loss": 1.2901,
+      "step": 1517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016903473349317652,
+      "loss": 1.2024,
+      "step": 1518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016899153530511063,
+      "loss": 1.1289,
+      "step": 1519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016894831253438884,
+      "loss": 1.2576,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016890506519641203,
+      "loss": 1.2019,
+      "step": 1521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016886179330658985,
+      "loss": 1.2431,
+      "step": 1522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016881849688034066,
+      "loss": 1.1562,
+      "step": 1523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687751759330916,
+      "loss": 1.2901,
+      "step": 1524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016873183048027852,
+      "loss": 1.2413,
+      "step": 1525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016868846053734606,
+      "loss": 1.1445,
+      "step": 1526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001686450661197475,
+      "loss": 1.1726,
+      "step": 1527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016860164724294486,
+      "loss": 1.2321,
+      "step": 1528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016855820392240896,
+      "loss": 1.2365,
+      "step": 1529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016851473617361922,
+      "loss": 1.3249,
+      "step": 1530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016847124401206384,
+      "loss": 1.2852,
+      "step": 1531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016842772745323967,
+      "loss": 1.2638,
+      "step": 1532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016838418651265227,
+      "loss": 1.247,
+      "step": 1533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016834062120581588,
+      "loss": 1.3268,
+      "step": 1534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682970315482535,
+      "loss": 1.1704,
+      "step": 1535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016825341755549665,
+      "loss": 1.2463,
+      "step": 1536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016820977924308567,
+      "loss": 1.2014,
+      "step": 1537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016816611662656946,
+      "loss": 1.2113,
+      "step": 1538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016812242972150568,
+      "loss": 1.2345,
+      "step": 1539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001680787185434606,
+      "loss": 1.2886,
+      "step": 1540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016803498310800902,
+      "loss": 1.2227,
+      "step": 1541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679912234307346,
+      "loss": 1.2133,
+      "step": 1542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679474395272295,
+      "loss": 1.2004,
+      "step": 1543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016790363141309448,
+      "loss": 1.1993,
+      "step": 1544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001678597991039391,
+      "loss": 1.1738,
+      "step": 1545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016781594261538132,
+      "loss": 1.2293,
+      "step": 1546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001677720619630479,
+      "loss": 1.2308,
+      "step": 1547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016772815716257412,
+      "loss": 1.2348,
+      "step": 1548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016768422822960387,
+      "loss": 1.0569,
+      "step": 1549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016764027517978961,
+      "loss": 1.2729,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016759629802879248,
+      "loss": 1.1908,
+      "step": 1551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016755229679228217,
+      "loss": 1.1799,
+      "step": 1552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016750827148593692,
+      "loss": 1.2052,
+      "step": 1553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016746422212544353,
+      "loss": 1.1969,
+      "step": 1554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016742014872649745,
+      "loss": 1.235,
+      "step": 1555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016737605130480264,
+      "loss": 1.3365,
+      "step": 1556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016733192987607165,
+      "loss": 1.2668,
+      "step": 1557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016728778445602554,
+      "loss": 1.2502,
+      "step": 1558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016724361506039397,
+      "loss": 1.174,
+      "step": 1559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016719942170491508,
+      "loss": 1.1723,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016715520440533564,
+      "loss": 1.22,
+      "step": 1561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016711096317741087,
+      "loss": 1.1611,
+      "step": 1562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016706669803690449,
+      "loss": 1.1629,
+      "step": 1563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016702240899958888,
+      "loss": 1.1538,
+      "step": 1564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016697809608124482,
+      "loss": 1.2002,
+      "step": 1565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669337592976616,
+      "loss": 1.0641,
+      "step": 1566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016688939866463706,
+      "loss": 1.3104,
+      "step": 1567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668450141979775,
+      "loss": 1.2248,
+      "step": 1568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016680060591349775,
+      "loss": 1.3583,
+      "step": 1569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016675617382702114,
+      "loss": 1.1808,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016671171795437938,
+      "loss": 1.1572,
+      "step": 1571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016666723831141277,
+      "loss": 1.1311,
+      "step": 1572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016662273491397004,
+      "loss": 1.257,
+      "step": 1573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016657820777790834,
+      "loss": 1.0092,
+      "step": 1574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016653365691909333,
+      "loss": 1.2538,
+      "step": 1575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664890823533991,
+      "loss": 1.1146,
+      "step": 1576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016644448409670822,
+      "loss": 1.0686,
+      "step": 1577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016639986216491167,
+      "loss": 1.1499,
+      "step": 1578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016635521657390885,
+      "loss": 1.2007,
+      "step": 1579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016631054733960764,
+      "loss": 1.1455,
+      "step": 1580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001662658544779243,
+      "loss": 1.1934,
+      "step": 1581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016622113800478355,
+      "loss": 1.1172,
+      "step": 1582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016617639793611847,
+      "loss": 1.2024,
+      "step": 1583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016613163428787063,
+      "loss": 1.21,
+      "step": 1584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016608684707598986,
+      "loss": 1.2116,
+      "step": 1585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016604203631643454,
+      "loss": 1.2464,
+      "step": 1586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016599720202517136,
+      "loss": 1.1444,
+      "step": 1587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001659523442181754,
+      "loss": 1.1333,
+      "step": 1588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016590746291143013,
+      "loss": 1.17,
+      "step": 1589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001658625581209274,
+      "loss": 1.3383,
+      "step": 1590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016581762986266744,
+      "loss": 1.0725,
+      "step": 1591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016577267815265877,
+      "loss": 1.2019,
+      "step": 1592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016572770300691837,
+      "loss": 1.2065,
+      "step": 1593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001656827044414715,
+      "loss": 1.1116,
+      "step": 1594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016563768247235174,
+      "loss": 1.2632,
+      "step": 1595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001655926371156011,
+      "loss": 1.1505,
+      "step": 1596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001655475683872699,
+      "loss": 1.1906,
+      "step": 1597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016550247630341667,
+      "loss": 1.2502,
+      "step": 1598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016545736088010844,
+      "loss": 1.1832,
+      "step": 1599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016541222213342044,
+      "loss": 1.3324,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016536706007943625,
+      "loss": 1.2838,
+      "step": 1601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016532187473424772,
+      "loss": 1.1723,
+      "step": 1602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001652766661139551,
+      "loss": 1.1632,
+      "step": 1603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016523143423466673,
+      "loss": 1.0828,
+      "step": 1604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016518617911249946,
+      "loss": 1.2907,
+      "step": 1605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016514090076357834,
+      "loss": 1.2458,
+      "step": 1606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001650955992040366,
+      "loss": 1.3055,
+      "step": 1607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016505027445001594,
+      "loss": 1.1279,
+      "step": 1608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001650049265176661,
+      "loss": 1.2035,
+      "step": 1609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001649595554231452,
+      "loss": 1.207,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016491416118261966,
+      "loss": 1.1223,
+      "step": 1611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016486874381226407,
+      "loss": 1.2268,
+      "step": 1612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016482330332826125,
+      "loss": 1.1599,
+      "step": 1613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001647778397468023,
+      "loss": 1.2854,
+      "step": 1614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016473235308408653,
+      "loss": 1.2155,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016468684335632148,
+      "loss": 1.3172,
+      "step": 1616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016464131057972292,
+      "loss": 1.2471,
+      "step": 1617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001645957547705148,
+      "loss": 1.1826,
+      "step": 1618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016455017594492932,
+      "loss": 1.1543,
+      "step": 1619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016450457411920684,
+      "loss": 1.1779,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016445894930959591,
+      "loss": 1.2497,
+      "step": 1621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016441330153235334,
+      "loss": 1.2055,
+      "step": 1622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016436763080374407,
+      "loss": 1.1475,
+      "step": 1623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016432193714004117,
+      "loss": 1.2126,
+      "step": 1624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016427622055752603,
+      "loss": 1.191,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016423048107248802,
+      "loss": 1.0712,
+      "step": 1626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016418471870122483,
+      "loss": 1.1962,
+      "step": 1627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016413893346004215,
+      "loss": 1.2409,
+      "step": 1628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016409312536525401,
+      "loss": 1.2057,
+      "step": 1629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016404729443318243,
+      "loss": 1.0753,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640014406801576,
+      "loss": 1.1432,
+      "step": 1631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001639555641225179,
+      "loss": 1.2125,
+      "step": 1632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016390966477660977,
+      "loss": 1.1986,
+      "step": 1633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016386374265878778,
+      "loss": 1.1974,
+      "step": 1634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016381779778541468,
+      "loss": 1.2197,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637718301728612,
+      "loss": 1.1791,
+      "step": 1636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637258398375063,
+      "loss": 1.165,
+      "step": 1637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016367982679573698,
+      "loss": 1.1384,
+      "step": 1638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001636337910639483,
+      "loss": 1.1868,
+      "step": 1639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635877326585435,
+      "loss": 1.2002,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635416515959338,
+      "loss": 1.0799,
+      "step": 1641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016349554789253851,
+      "loss": 1.1965,
+      "step": 1642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016344942156478506,
+      "loss": 1.1711,
+      "step": 1643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016340327262910894,
+      "loss": 1.1503,
+      "step": 1644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001633571011019536,
+      "loss": 1.255,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016331090699977062,
+      "loss": 1.3944,
+      "step": 1646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001632646903390197,
+      "loss": 1.1169,
+      "step": 1647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016321845113616835,
+      "loss": 1.1392,
+      "step": 1648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016317218940769233,
+      "loss": 1.326,
+      "step": 1649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001631259051700753,
+      "loss": 1.3237,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016307959843980905,
+      "loss": 1.0904,
+      "step": 1651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630332692333933,
+      "loss": 1.1922,
+      "step": 1652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016298691756733573,
+      "loss": 1.2966,
+      "step": 1653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016294054345815216,
+      "loss": 1.2367,
+      "step": 1654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016289414692236635,
+      "loss": 1.2741,
+      "step": 1655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016284772797651,
+      "loss": 1.2113,
+      "step": 1656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016280128663712285,
+      "loss": 1.1456,
+      "step": 1657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001627548229207526,
+      "loss": 1.2241,
+      "step": 1658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016270833684395492,
+      "loss": 1.1227,
+      "step": 1659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016266182842329349,
+      "loss": 1.1471,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016261529767533985,
+      "loss": 1.1612,
+      "step": 1661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016256874461667358,
+      "loss": 1.2026,
+      "step": 1662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016252216926388228,
+      "loss": 1.2526,
+      "step": 1663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016247557163356127,
+      "loss": 1.3482,
+      "step": 1664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162428951742314,
+      "loss": 1.2425,
+      "step": 1665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016238230960675183,
+      "loss": 1.1133,
+      "step": 1666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162335645243494,
+      "loss": 1.2609,
+      "step": 1667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001622889586691676,
+      "loss": 1.2487,
+      "step": 1668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016224224990040782,
+      "loss": 1.1242,
+      "step": 1669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016219551895385758,
+      "loss": 1.1746,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016214876584616782,
+      "loss": 1.292,
+      "step": 1671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001621019905939973,
+      "loss": 1.1323,
+      "step": 1672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016205519321401267,
+      "loss": 1.2474,
+      "step": 1673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016200837372288863,
+      "loss": 1.2454,
+      "step": 1674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016196153213730746,
+      "loss": 1.2006,
+      "step": 1675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016191466847395954,
+      "loss": 1.2506,
+      "step": 1676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001618677827495431,
+      "loss": 1.258,
+      "step": 1677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016182087498076416,
+      "loss": 1.1712,
+      "step": 1678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016177394518433654,
+      "loss": 1.0774,
+      "step": 1679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617269933769821,
+      "loss": 1.2573,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016168001957543037,
+      "loss": 1.1031,
+      "step": 1681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001616330237964188,
+      "loss": 1.1632,
+      "step": 1682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016158600605669263,
+      "loss": 1.3699,
+      "step": 1683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016153896637300497,
+      "loss": 1.1484,
+      "step": 1684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016149190476211667,
+      "loss": 1.1303,
+      "step": 1685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016144482124079649,
+      "loss": 1.1778,
+      "step": 1686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016139771582582093,
+      "loss": 1.207,
+      "step": 1687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016135058853397434,
+      "loss": 1.2598,
+      "step": 1688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016130343938204882,
+      "loss": 1.153,
+      "step": 1689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016125626838684425,
+      "loss": 1.2401,
+      "step": 1690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016120907556516837,
+      "loss": 1.2112,
+      "step": 1691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016116186093383662,
+      "loss": 1.2133,
+      "step": 1692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016111462450967226,
+      "loss": 1.3208,
+      "step": 1693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016106736630950628,
+      "loss": 1.2315,
+      "step": 1694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001610200863501774,
+      "loss": 1.1142,
+      "step": 1695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001609727846485322,
+      "loss": 1.159,
+      "step": 1696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001609254612214249,
+      "loss": 1.2663,
+      "step": 1697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016087811608571752,
+      "loss": 1.261,
+      "step": 1698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608307492582798,
+      "loss": 1.1375,
+      "step": 1699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016078336075598925,
+      "loss": 1.1515,
+      "step": 1700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016073595059573097,
+      "loss": 1.2158,
+      "step": 1701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016068851879439793,
+      "loss": 1.0893,
+      "step": 1702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016064106536889074,
+      "loss": 1.1549,
+      "step": 1703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016059359033611772,
+      "loss": 1.1901,
+      "step": 1704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016054609371299487,
+      "loss": 1.2144,
+      "step": 1705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016049857551644596,
+      "loss": 1.2192,
+      "step": 1706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016045103576340237,
+      "loss": 1.0749,
+      "step": 1707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016040347447080313,
+      "loss": 1.1549,
+      "step": 1708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016035589165559508,
+      "loss": 1.1835,
+      "step": 1709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016030828733473257,
+      "loss": 1.144,
+      "step": 1710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016026066152517775,
+      "loss": 1.2309,
+      "step": 1711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016021301424390036,
+      "loss": 1.2118,
+      "step": 1712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001601653455078778,
+      "loss": 1.3033,
+      "step": 1713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016011765533409505,
+      "loss": 1.2568,
+      "step": 1714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600699437395449,
+      "loss": 1.2009,
+      "step": 1715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600222107412275,
+      "loss": 1.2614,
+      "step": 1716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015997445635615099,
+      "loss": 1.3483,
+      "step": 1717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001599266806013308,
+      "loss": 1.2348,
+      "step": 1718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001598788834937901,
+      "loss": 1.0866,
+      "step": 1719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015983106505055976,
+      "loss": 1.2245,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015978322528867808,
+      "loss": 1.1653,
+      "step": 1721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001597353642251911,
+      "loss": 1.2378,
+      "step": 1722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015968748187715235,
+      "loss": 1.2658,
+      "step": 1723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000159639578261623,
+      "loss": 1.1331,
+      "step": 1724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015959165339567174,
+      "loss": 1.3096,
+      "step": 1725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015954370729637498,
+      "loss": 1.2427,
+      "step": 1726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001594957399808165,
+      "loss": 1.193,
+      "step": 1727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015944775146608777,
+      "loss": 1.1624,
+      "step": 1728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015939974176928777,
+      "loss": 1.1353,
+      "step": 1729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015935171090752302,
+      "loss": 1.2333,
+      "step": 1730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015930365889790758,
+      "loss": 1.2503,
+      "step": 1731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001592555857575631,
+      "loss": 1.198,
+      "step": 1732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015920749150361867,
+      "loss": 1.2043,
+      "step": 1733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015915937615321098,
+      "loss": 1.149,
+      "step": 1734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015911123972348417,
+      "loss": 1.1039,
+      "step": 1735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015906308223158998,
+      "loss": 1.089,
+      "step": 1736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015901490369468762,
+      "loss": 1.2054,
+      "step": 1737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015896670412994366,
+      "loss": 1.2139,
+      "step": 1738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589184835545324,
+      "loss": 1.1971,
+      "step": 1739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015887024198563552,
+      "loss": 1.2483,
+      "step": 1740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015882197944044207,
+      "loss": 1.2866,
+      "step": 1741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015877369593614874,
+      "loss": 1.1243,
+      "step": 1742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015872539148995964,
+      "loss": 1.2749,
+      "step": 1743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015867706611908625,
+      "loss": 1.2976,
+      "step": 1744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001586287198407477,
+      "loss": 1.114,
+      "step": 1745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015858035267217032,
+      "loss": 1.107,
+      "step": 1746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585319646305881,
+      "loss": 1.015,
+      "step": 1747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015848355573324234,
+      "loss": 1.1707,
+      "step": 1748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001584351259973818,
+      "loss": 1.183,
+      "step": 1749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015838667544026275,
+      "loss": 1.0872,
+      "step": 1750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015833820407914874,
+      "loss": 1.202,
+      "step": 1751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001582897119313108,
+      "loss": 1.1932,
+      "step": 1752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001582411990140274,
+      "loss": 1.2403,
+      "step": 1753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015819266534458434,
+      "loss": 1.149,
+      "step": 1754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001581441109402749,
+      "loss": 1.3746,
+      "step": 1755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015809553581839966,
+      "loss": 1.2211,
+      "step": 1756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580469399962666,
+      "loss": 1.0727,
+      "step": 1757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015799832349119112,
+      "loss": 1.0982,
+      "step": 1758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000157949686320496,
+      "loss": 1.1591,
+      "step": 1759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015790102850151127,
+      "loss": 1.1572,
+      "step": 1760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015785235005157447,
+      "loss": 1.185,
+      "step": 1761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015780365098803042,
+      "loss": 1.0971,
+      "step": 1762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001577549313282312,
+      "loss": 1.1482,
+      "step": 1763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015770619108953642,
+      "loss": 1.2177,
+      "step": 1764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015765743028931285,
+      "loss": 1.136,
+      "step": 1765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001576086489449346,
+      "loss": 1.2006,
+      "step": 1766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015755984707378326,
+      "loss": 1.3344,
+      "step": 1767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015751102469324755,
+      "loss": 1.1215,
+      "step": 1768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015746218182072357,
+      "loss": 1.1343,
+      "step": 1769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001574133184736148,
+      "loss": 1.0403,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001573644346693319,
+      "loss": 1.2373,
+      "step": 1771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015731553042529277,
+      "loss": 1.2421,
+      "step": 1772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001572666057589228,
+      "loss": 1.1018,
+      "step": 1773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015721766068765448,
+      "loss": 1.2736,
+      "step": 1774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015716869522892766,
+      "loss": 1.1442,
+      "step": 1775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001571197094001894,
+      "loss": 1.2853,
+      "step": 1776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015707070321889404,
+      "loss": 1.1599,
+      "step": 1777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001570216767025032,
+      "loss": 1.2125,
+      "step": 1778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015697262986848564,
+      "loss": 1.2547,
+      "step": 1779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001569235627343175,
+      "loss": 1.2284,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015687447531748214,
+      "loss": 1.2231,
+      "step": 1781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015682536763547003,
+      "loss": 1.1945,
+      "step": 1782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015677623970577893,
+      "loss": 1.177,
+      "step": 1783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567270915459138,
+      "loss": 1.308,
+      "step": 1784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001566779231733869,
+      "loss": 1.2833,
+      "step": 1785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015662873460571753,
+      "loss": 1.1298,
+      "step": 1786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001565795258604323,
+      "loss": 1.3203,
+      "step": 1787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000156530296955065,
+      "loss": 1.1169,
+      "step": 1788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015648104790715658,
+      "loss": 1.364,
+      "step": 1789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015643177873425507,
+      "loss": 1.2208,
+      "step": 1790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015638248945391592,
+      "loss": 1.232,
+      "step": 1791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015633318008370151,
+      "loss": 1.2005,
+      "step": 1792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015628385064118147,
+      "loss": 1.2992,
+      "step": 1793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001562345011439326,
+      "loss": 1.216,
+      "step": 1794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015618513160953882,
+      "loss": 1.1715,
+      "step": 1795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015613574205559116,
+      "loss": 1.1994,
+      "step": 1796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015608633249968783,
+      "loss": 1.1454,
+      "step": 1797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015603690295943413,
+      "loss": 1.1363,
+      "step": 1798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015598745345244257,
+      "loss": 1.1356,
+      "step": 1799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015593798399633264,
+      "loss": 1.1786,
+      "step": 1800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015588849460873103,
+      "loss": 1.2953,
+      "step": 1801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015583898530727148,
+      "loss": 1.0148,
+      "step": 1802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001557894561095949,
+      "loss": 1.0244,
+      "step": 1803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015573990703334918,
+      "loss": 1.258,
+      "step": 1804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001556903380961894,
+      "loss": 1.2378,
+      "step": 1805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015564074931577766,
+      "loss": 1.1281,
+      "step": 1806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001555911407097831,
+      "loss": 1.1018,
+      "step": 1807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015554151229588205,
+      "loss": 1.132,
+      "step": 1808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015549186409175776,
+      "loss": 1.1643,
+      "step": 1809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015544219611510058,
+      "loss": 1.2201,
+      "step": 1810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015539250838360794,
+      "loss": 1.3515,
+      "step": 1811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001553428009149842,
+      "loss": 1.1936,
+      "step": 1812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015529307372694095,
+      "loss": 1.2234,
+      "step": 1813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015524332683719663,
+      "loss": 1.1958,
+      "step": 1814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001551935602634767,
+      "loss": 1.2116,
+      "step": 1815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015514377402351377,
+      "loss": 1.1441,
+      "step": 1816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001550939681350474,
+      "loss": 0.9673,
+      "step": 1817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000155044142615824,
+      "loss": 1.242,
+      "step": 1818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015499429748359726,
+      "loss": 1.184,
+      "step": 1819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015494443275612763,
+      "loss": 1.1994,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015489454845118257,
+      "loss": 1.0545,
+      "step": 1821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015484464458653664,
+      "loss": 1.0699,
+      "step": 1822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015479472117997122,
+      "loss": 1.2067,
+      "step": 1823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001547447782492748,
+      "loss": 1.3628,
+      "step": 1824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015469481581224272,
+      "loss": 1.186,
+      "step": 1825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015464483388667724,
+      "loss": 1.1259,
+      "step": 1826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001545948324903877,
+      "loss": 1.1251,
+      "step": 1827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015454481164119027,
+      "loss": 1.0714,
+      "step": 1828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015449477135690812,
+      "loss": 1.2438,
+      "step": 1829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015444471165537124,
+      "loss": 1.1487,
+      "step": 1830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015439463255441667,
+      "loss": 1.1651,
+      "step": 1831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015434453407188825,
+      "loss": 1.1959,
+      "step": 1832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015429441622563683,
+      "loss": 1.1575,
+      "step": 1833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015424427903352003,
+      "loss": 1.1268,
+      "step": 1834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001541941225134025,
+      "loss": 1.0749,
+      "step": 1835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001541439466831557,
+      "loss": 1.2632,
+      "step": 1836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015409375156065798,
+      "loss": 1.1152,
+      "step": 1837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015404353716379454,
+      "loss": 1.1925,
+      "step": 1838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015399330351045754,
+      "loss": 1.4314,
+      "step": 1839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015394305061854592,
+      "loss": 1.3242,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015389277850596546,
+      "loss": 1.2713,
+      "step": 1841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015384248719062883,
+      "loss": 1.2198,
+      "step": 1842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015379217669045554,
+      "loss": 1.2433,
+      "step": 1843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015374184702337198,
+      "loss": 1.1114,
+      "step": 1844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015369149820731122,
+      "loss": 1.2474,
+      "step": 1845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015364113026021333,
+      "loss": 1.17,
+      "step": 1846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001535907432000251,
+      "loss": 1.2861,
+      "step": 1847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015354033704470012,
+      "loss": 1.2025,
+      "step": 1848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015348991181219886,
+      "loss": 1.1277,
+      "step": 1849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015343946752048853,
+      "loss": 1.0464,
+      "step": 1850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015338900418754309,
+      "loss": 1.0817,
+      "step": 1851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001533385218313434,
+      "loss": 1.3188,
+      "step": 1852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000153288020469877,
+      "loss": 1.1344,
+      "step": 1853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015323750012113827,
+      "loss": 1.1315,
+      "step": 1854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015318696080312828,
+      "loss": 1.224,
+      "step": 1855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015313640253385494,
+      "loss": 1.3144,
+      "step": 1856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015308582533133283,
+      "loss": 1.2467,
+      "step": 1857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015303522921358335,
+      "loss": 1.1719,
+      "step": 1858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001529846141986346,
+      "loss": 1.1511,
+      "step": 1859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015293398030452143,
+      "loss": 1.2501,
+      "step": 1860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015288332754928542,
+      "loss": 1.0516,
+      "step": 1861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015283265595097481,
+      "loss": 1.1917,
+      "step": 1862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001527819655276447,
+      "loss": 1.0388,
+      "step": 1863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001527312562973567,
+      "loss": 1.0899,
+      "step": 1864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015268052827817927,
+      "loss": 1.2662,
+      "step": 1865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015262978148818755,
+      "loss": 1.2239,
+      "step": 1866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015257901594546326,
+      "loss": 1.239,
+      "step": 1867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001525282316680949,
+      "loss": 1.1392,
+      "step": 1868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015247742867417768,
+      "loss": 1.1724,
+      "step": 1869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015242660698181336,
+      "loss": 1.1661,
+      "step": 1870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015237576660911042,
+      "loss": 1.1015,
+      "step": 1871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015232490757418406,
+      "loss": 1.229,
+      "step": 1872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015227402989515608,
+      "loss": 1.2928,
+      "step": 1873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015222313359015483,
+      "loss": 1.1976,
+      "step": 1874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015217221867731544,
+      "loss": 1.0774,
+      "step": 1875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015212128517477956,
+      "loss": 1.1569,
+      "step": 1876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015207033310069557,
+      "loss": 1.2913,
+      "step": 1877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015201936247321841,
+      "loss": 1.0955,
+      "step": 1878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015196837331050959,
+      "loss": 1.1342,
+      "step": 1879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001519173656307373,
+      "loss": 1.1871,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001518663394520763,
+      "loss": 1.1208,
+      "step": 1881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015181529479270787,
+      "loss": 1.2869,
+      "step": 1882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015176423167082007,
+      "loss": 1.0946,
+      "step": 1883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015171315010460728,
+      "loss": 1.213,
+      "step": 1884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015166205011227061,
+      "loss": 1.18,
+      "step": 1885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015161093171201778,
+      "loss": 1.3339,
+      "step": 1886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515597949220629,
+      "loss": 1.3212,
+      "step": 1887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515086397606268,
+      "loss": 1.0859,
+      "step": 1888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514574662459368,
+      "loss": 1.2132,
+      "step": 1889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015140627439622667,
+      "loss": 1.198,
+      "step": 1890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001513550642297368,
+      "loss": 1.1644,
+      "step": 1891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015130383576471415,
+      "loss": 1.2114,
+      "step": 1892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015125258901941207,
+      "loss": 1.2877,
+      "step": 1893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015120132401209055,
+      "loss": 1.1249,
+      "step": 1894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015115004076101608,
+      "loss": 1.0511,
+      "step": 1895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015109873928446148,
+      "loss": 1.0507,
+      "step": 1896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015104741960070625,
+      "loss": 1.224,
+      "step": 1897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015099608172803634,
+      "loss": 1.1339,
+      "step": 1898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015094472568474415,
+      "loss": 1.2023,
+      "step": 1899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001508933514891285,
+      "loss": 1.2233,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015084195915949486,
+      "loss": 1.1744,
+      "step": 1901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015079054871415494,
+      "loss": 1.1341,
+      "step": 1902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015073912017142703,
+      "loss": 1.1534,
+      "step": 1903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015068767354963584,
+      "loss": 1.1108,
+      "step": 1904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015063620886711254,
+      "loss": 1.1263,
+      "step": 1905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015058472614219473,
+      "loss": 1.1693,
+      "step": 1906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015053322539322644,
+      "loss": 1.1911,
+      "step": 1907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015048170663855802,
+      "loss": 1.1529,
+      "step": 1908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015043016989654647,
+      "loss": 1.2447,
+      "step": 1909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015037861518555493,
+      "loss": 1.1881,
+      "step": 1910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015032704252395315,
+      "loss": 1.118,
+      "step": 1911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015027545193011717,
+      "loss": 1.2974,
+      "step": 1912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015022384342242945,
+      "loss": 1.1459,
+      "step": 1913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001501722170192788,
+      "loss": 1.2728,
+      "step": 1914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015012057273906048,
+      "loss": 1.2525,
+      "step": 1915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015006891060017607,
+      "loss": 1.2262,
+      "step": 1916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015001723062103348,
+      "loss": 1.1223,
+      "step": 1917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014996553282004704,
+      "loss": 1.0908,
+      "step": 1918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001499138172156374,
+      "loss": 1.0805,
+      "step": 1919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014986208382623156,
+      "loss": 1.2395,
+      "step": 1920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014981033267026284,
+      "loss": 1.1868,
+      "step": 1921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014975856376617093,
+      "loss": 1.1752,
+      "step": 1922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001497067771324018,
+      "loss": 1.17,
+      "step": 1923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001496549727874078,
+      "loss": 1.2026,
+      "step": 1924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001496031507496475,
+      "loss": 1.2408,
+      "step": 1925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014955131103758582,
+      "loss": 1.0918,
+      "step": 1926
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000149499453669694,
+      "loss": 1.2437,
+      "step": 1927
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014944757866444956,
+      "loss": 1.1693,
+      "step": 1928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001493956860403363,
+      "loss": 1.1655,
+      "step": 1929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014934377581584424,
+      "loss": 1.198,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014929184800946976,
+      "loss": 1.1534,
+      "step": 1931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001492399026397155,
+      "loss": 1.2215,
+      "step": 1932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014918793972509026,
+      "loss": 1.2677,
+      "step": 1933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001491359592841092,
+      "loss": 1.1819,
+      "step": 1934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014908396133529373,
+      "loss": 1.2181,
+      "step": 1935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014903194589717134,
+      "loss": 1.2934,
+      "step": 1936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014897991298827595,
+      "loss": 1.1545,
+      "step": 1937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014892786262714756,
+      "loss": 1.2464,
+      "step": 1938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488757948323325,
+      "loss": 1.2632,
+      "step": 1939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014882370962238323,
+      "loss": 1.1012,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014877160701585847,
+      "loss": 1.1842,
+      "step": 1941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014871948703132305,
+      "loss": 1.2288,
+      "step": 1942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014866734968734815,
+      "loss": 1.0926,
+      "step": 1943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014861519500251096,
+      "loss": 1.1781,
+      "step": 1944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014856302299539494,
+      "loss": 1.1532,
+      "step": 1945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001485108336845898,
+      "loss": 1.0978,
+      "step": 1946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001484586270886912,
+      "loss": 1.2654,
+      "step": 1947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014840640322630113,
+      "loss": 1.2137,
+      "step": 1948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483541621160277,
+      "loss": 1.0873,
+      "step": 1949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014830190377648517,
+      "loss": 1.2027,
+      "step": 1950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001482496282262939,
+      "loss": 1.1771,
+      "step": 1951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014819733548408038,
+      "loss": 1.0771,
+      "step": 1952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014814502556847732,
+      "loss": 1.1594,
+      "step": 1953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014809269849812344,
+      "loss": 1.1016,
+      "step": 1954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014804035429166358,
+      "loss": 1.1045,
+      "step": 1955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014798799296774877,
+      "loss": 1.0824,
+      "step": 1956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001479356145450361,
+      "loss": 1.139,
+      "step": 1957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014788321904218865,
+      "loss": 1.1904,
+      "step": 1958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014783080647787575,
+      "loss": 1.069,
+      "step": 1959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014777837687077273,
+      "loss": 1.1481,
+      "step": 1960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014772593023956097,
+      "loss": 1.2412,
+      "step": 1961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014767346660292797,
+      "loss": 1.1434,
+      "step": 1962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014762098597956725,
+      "loss": 1.11,
+      "step": 1963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014756848838817836,
+      "loss": 1.2569,
+      "step": 1964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014751597384746702,
+      "loss": 1.3318,
+      "step": 1965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014746344237614482,
+      "loss": 1.1718,
+      "step": 1966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014741089399292952,
+      "loss": 1.0122,
+      "step": 1967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001473583287165448,
+      "loss": 1.1295,
+      "step": 1968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014730574656572043,
+      "loss": 1.1801,
+      "step": 1969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014725314755919218,
+      "loss": 1.118,
+      "step": 1970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001472005317157018,
+      "loss": 1.1714,
+      "step": 1971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014714789905399708,
+      "loss": 1.2643,
+      "step": 1972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014709524959283175,
+      "loss": 1.152,
+      "step": 1973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014704258335096557,
+      "loss": 1.2183,
+      "step": 1974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014698990034716424,
+      "loss": 1.3027,
+      "step": 1975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014693720060019954,
+      "loss": 1.2305,
+      "step": 1976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000146884484128849,
+      "loss": 1.2388,
+      "step": 1977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014683175095189633,
+      "loss": 1.0385,
+      "step": 1978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014677900108813112,
+      "loss": 1.0735,
+      "step": 1979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467262345563488,
+      "loss": 1.1125,
+      "step": 1980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014667345137535094,
+      "loss": 1.0643,
+      "step": 1981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014662065156394484,
+      "loss": 1.2534,
+      "step": 1982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014656783514094385,
+      "loss": 1.1415,
+      "step": 1983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014651500212516723,
+      "loss": 1.2329,
+      "step": 1984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014646215253544007,
+      "loss": 1.0278,
+      "step": 1985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001464092863905935,
+      "loss": 1.1046,
+      "step": 1986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001463564037094644,
+      "loss": 1.097,
+      "step": 1987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014630350451089562,
+      "loss": 1.3219,
+      "step": 1988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014625058881373595,
+      "loss": 1.3264,
+      "step": 1989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014619765663683996,
+      "loss": 1.262,
+      "step": 1990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001461447079990681,
+      "loss": 1.2117,
+      "step": 1991
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001460917429192868,
+      "loss": 1.164,
+      "step": 1992
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001460387614163682,
+      "loss": 1.1638,
+      "step": 1993
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014598576350919034,
+      "loss": 1.2814,
+      "step": 1994
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014593274921663717,
+      "loss": 1.1794,
+      "step": 1995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001458797185575984,
+      "loss": 1.1822,
+      "step": 1996
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001458266715509696,
+      "loss": 1.1143,
+      "step": 1997
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001457736082156522,
+      "loss": 1.1284,
+      "step": 1998
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014572052857055333,
+      "loss": 1.2232,
+      "step": 1999
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001456674326345861,
+      "loss": 1.1815,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014561432042666935,
+      "loss": 1.1345,
+      "step": 2001
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014556119196572763,
+      "loss": 1.0504,
+      "step": 2002
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014550804727069143,
+      "loss": 1.2016,
+      "step": 2003
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001454548863604969,
+      "loss": 1.2157,
+      "step": 2004
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014540170925408607,
+      "loss": 1.2171,
+      "step": 2005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014534851597040665,
+      "loss": 1.2835,
+      "step": 2006
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014529530652841217,
+      "loss": 1.2192,
+      "step": 2007
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014524208094706192,
+      "loss": 1.1174,
+      "step": 2008
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001451888392453209,
+      "loss": 1.2283,
+      "step": 2009
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001451355814421599,
+      "loss": 1.2266,
+      "step": 2010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014508230755655538,
+      "loss": 1.2042,
+      "step": 2011
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014502901760748965,
+      "loss": 1.1187,
+      "step": 2012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014497571161395061,
+      "loss": 1.2553,
+      "step": 2013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000144922389594932,
+      "loss": 1.1884,
+      "step": 2014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001448690515694331,
+      "loss": 1.1478,
+      "step": 2015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001448156975564591,
+      "loss": 1.151,
+      "step": 2016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014476232757502076,
+      "loss": 1.2284,
+      "step": 2017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001447089416441345,
+      "loss": 1.265,
+      "step": 2018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014465553978282257,
+      "loss": 1.2493,
+      "step": 2019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014460212201011276,
+      "loss": 1.1118,
+      "step": 2020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014454868834503853,
+      "loss": 1.2118,
+      "step": 2021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014449523880663914,
+      "loss": 1.1419,
+      "step": 2022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014444177341395935,
+      "loss": 1.0892,
+      "step": 2023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014438829218604962,
+      "loss": 1.0627,
+      "step": 2024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014433479514196616,
+      "loss": 1.0854,
+      "step": 2025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001442812823007706,
+      "loss": 1.1566,
+      "step": 2026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014422775368153038,
+      "loss": 1.1685,
+      "step": 2027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014417420930331852,
+      "loss": 1.3243,
+      "step": 2028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014412064918521358,
+      "loss": 1.2063,
+      "step": 2029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014406707334629982,
+      "loss": 1.2416,
+      "step": 2030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001440134818056671,
+      "loss": 1.2353,
+      "step": 2031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001439598745824108,
+      "loss": 1.1128,
+      "step": 2032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014390625169563188,
+      "loss": 1.1628,
+      "step": 2033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000143852613164437,
+      "loss": 1.2055,
+      "step": 2034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014379895900793832,
+      "loss": 1.1262,
+      "step": 2035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014374528924525354,
+      "loss": 1.0998,
+      "step": 2036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000143691603895506,
+      "loss": 1.0355,
+      "step": 2037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001436379029778245,
+      "loss": 1.2082,
+      "step": 2038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001435841865113434,
+      "loss": 1.1736,
+      "step": 2039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014353045451520277,
+      "loss": 1.16,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014347670700854797,
+      "loss": 1.2142,
+      "step": 2041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014342294401052997,
+      "loss": 1.1933,
+      "step": 2042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014336916554030538,
+      "loss": 1.203,
+      "step": 2043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001433153716170361,
+      "loss": 1.1386,
+      "step": 2044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001432615622598898,
+      "loss": 1.2912,
+      "step": 2045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014320773748803943,
+      "loss": 1.1502,
+      "step": 2046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014315389732066352,
+      "loss": 1.2381,
+      "step": 2047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014310004177694611,
+      "loss": 1.1459,
+      "step": 2048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001430461708760767,
+      "loss": 1.1796,
+      "step": 2049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014299228463725017,
+      "loss": 1.0897,
+      "step": 2050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014293838307966703,
+      "loss": 1.1646,
+      "step": 2051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014288446622253312,
+      "loss": 1.2261,
+      "step": 2052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001428305340850598,
+      "loss": 1.3285,
+      "step": 2053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014277658668646382,
+      "loss": 1.1648,
+      "step": 2054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014272262404596746,
+      "loss": 1.0389,
+      "step": 2055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014266864618279828,
+      "loss": 1.153,
+      "step": 2056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001426146531161894,
+      "loss": 1.0344,
+      "step": 2057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014256064486537935,
+      "loss": 1.1058,
+      "step": 2058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014250662144961197,
+      "loss": 1.1649,
+      "step": 2059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014245258288813655,
+      "loss": 1.3766,
+      "step": 2060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014239852920020787,
+      "loss": 1.1357,
+      "step": 2061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014234446040508597,
+      "loss": 1.1765,
+      "step": 2062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014229037652203628,
+      "loss": 1.1883,
+      "step": 2063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014223627757032977,
+      "loss": 1.3059,
+      "step": 2064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014218216356924254,
+      "loss": 1.0891,
+      "step": 2065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014212803453805624,
+      "loss": 1.0636,
+      "step": 2066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014207389049605777,
+      "loss": 1.1118,
+      "step": 2067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014201973146253945,
+      "loss": 1.1708,
+      "step": 2068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014196555745679888,
+      "loss": 1.3085,
+      "step": 2069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000141911368498139,
+      "loss": 1.1189,
+      "step": 2070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014185716460586814,
+      "loss": 1.1788,
+      "step": 2071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014180294579929992,
+      "loss": 1.1917,
+      "step": 2072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417487120977532,
+      "loss": 1.1913,
+      "step": 2073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014169446352055228,
+      "loss": 1.0902,
+      "step": 2074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014164020008702665,
+      "loss": 1.078,
+      "step": 2075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014158592181651115,
+      "loss": 1.2039,
+      "step": 2076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001415316287283459,
+      "loss": 1.2646,
+      "step": 2077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001414773208418763,
+      "loss": 1.3399,
+      "step": 2078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014142299817645297,
+      "loss": 1.0572,
+      "step": 2079
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001413686607514319,
+      "loss": 1.1533,
+      "step": 2080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014131430858617424,
+      "loss": 1.2233,
+      "step": 2081
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014125994170004644,
+      "loss": 1.2068,
+      "step": 2082
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014120556011242022,
+      "loss": 1.1038,
+      "step": 2083
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014115116384267243,
+      "loss": 1.1185,
+      "step": 2084
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001410967529101853,
+      "loss": 1.2832,
+      "step": 2085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014104232733434617,
+      "loss": 1.1352,
+      "step": 2086
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014098788713454768,
+      "loss": 1.2236,
+      "step": 2087
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001409334323301876,
+      "loss": 1.3005,
+      "step": 2088
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014087896294066892,
+      "loss": 1.1701,
+      "step": 2089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014082447898539993,
+      "loss": 1.1391,
+      "step": 2090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014076998048379402,
+      "loss": 1.2062,
+      "step": 2091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001407154674552697,
+      "loss": 1.2521,
+      "step": 2092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014066093991925077,
+      "loss": 1.1474,
+      "step": 2093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014060639789516619,
+      "loss": 1.2104,
+      "step": 2094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014055184140245,
+      "loss": 1.1568,
+      "step": 2095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014049727046054148,
+      "loss": 1.2198,
+      "step": 2096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014044268508888504,
+      "loss": 1.3989,
+      "step": 2097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014038808530693017,
+      "loss": 1.296,
+      "step": 2098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014033347113413157,
+      "loss": 1.2515,
+      "step": 2099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014027884258994904,
+      "loss": 1.2295,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014022419969384747,
+      "loss": 1.0794,
+      "step": 2101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014016954246529696,
+      "loss": 1.2837,
+      "step": 2102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014011487092377257,
+      "loss": 1.1345,
+      "step": 2103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001400601850887546,
+      "loss": 1.2329,
+      "step": 2104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014000548497972837,
+      "loss": 1.1129,
+      "step": 2105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013995077061618427,
+      "loss": 1.2025,
+      "step": 2106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013989604201761782,
+      "loss": 1.1743,
+      "step": 2107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001398412992035296,
+      "loss": 1.1322,
+      "step": 2108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013978654219342523,
+      "loss": 1.1265,
+      "step": 2109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013973177100681542,
+      "loss": 1.0796,
+      "step": 2110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001396769856632159,
+      "loss": 1.1701,
+      "step": 2111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013962218618214747,
+      "loss": 1.166,
+      "step": 2112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013956737258313595,
+      "loss": 1.0792,
+      "step": 2113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013951254488571218,
+      "loss": 1.3418,
+      "step": 2114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013945770310941209,
+      "loss": 1.2416,
+      "step": 2115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013940284727377656,
+      "loss": 1.1036,
+      "step": 2116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013934797739835144,
+      "loss": 1.1451,
+      "step": 2117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001392930935026877,
+      "loss": 1.1262,
+      "step": 2118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013923819560634126,
+      "loss": 1.2163,
+      "step": 2119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013918328372887294,
+      "loss": 1.0765,
+      "step": 2120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001391283578898487,
+      "loss": 1.213,
+      "step": 2121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013907341810883933,
+      "loss": 1.2957,
+      "step": 2122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013901846440542064,
+      "loss": 1.2726,
+      "step": 2123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001389634967991735,
+      "loss": 1.2049,
+      "step": 2124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013890851530968355,
+      "loss": 1.212,
+      "step": 2125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013885351995654148,
+      "loss": 1.0843,
+      "step": 2126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013879851075934295,
+      "loss": 1.2288,
+      "step": 2127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013874348773768846,
+      "loss": 1.237,
+      "step": 2128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001386884509111835,
+      "loss": 1.0991,
+      "step": 2129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013863340029943857,
+      "loss": 1.1858,
+      "step": 2130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013857833592206882,
+      "loss": 1.1367,
+      "step": 2131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013852325779869457,
+      "loss": 1.1011,
+      "step": 2132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013846816594894086,
+      "loss": 1.2287,
+      "step": 2133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013841306039243776,
+      "loss": 1.1327,
+      "step": 2134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001383579411488201,
+      "loss": 1.0768,
+      "step": 2135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013830280823772767,
+      "loss": 1.074,
+      "step": 2136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001382476616788051,
+      "loss": 1.2397,
+      "step": 2137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381925014917019,
+      "loss": 1.1157,
+      "step": 2138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013813732769607238,
+      "loss": 1.1127,
+      "step": 2139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001380821403115758,
+      "loss": 1.2264,
+      "step": 2140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013802693935787618,
+      "loss": 1.2599,
+      "step": 2141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013797172485464237,
+      "loss": 1.1145,
+      "step": 2142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001379164968215481,
+      "loss": 0.9969,
+      "step": 2143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013786125527827188,
+      "loss": 1.1128,
+      "step": 2144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013780600024449702,
+      "loss": 1.2391,
+      "step": 2145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001377507317399118,
+      "loss": 1.2668,
+      "step": 2146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000137695449784209,
+      "loss": 1.0677,
+      "step": 2147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013764015439708641,
+      "loss": 1.2158,
+      "step": 2148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013758484559824665,
+      "loss": 0.9452,
+      "step": 2149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001375295234073969,
+      "loss": 1.2596,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013747418784424933,
+      "loss": 1.1894,
+      "step": 2151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013741883892852072,
+      "loss": 1.0541,
+      "step": 2152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001373634766799327,
+      "loss": 1.23,
+      "step": 2153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013730810111821156,
+      "loss": 1.0779,
+      "step": 2154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001372527122630885,
+      "loss": 1.0471,
+      "step": 2155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013719731013429924,
+      "loss": 1.1311,
+      "step": 2156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013714189475158442,
+      "loss": 1.1803,
+      "step": 2157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013708646613468925,
+      "loss": 1.0539,
+      "step": 2158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001370310243033638,
+      "loss": 1.1649,
+      "step": 2159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013697556927736275,
+      "loss": 1.1642,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013692010107644546,
+      "loss": 1.1245,
+      "step": 2161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013686461972037609,
+      "loss": 1.1506,
+      "step": 2162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013680912522892342,
+      "loss": 1.2176,
+      "step": 2163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013675361762186086,
+      "loss": 1.096,
+      "step": 2164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013669809691896668,
+      "loss": 0.9321,
+      "step": 2165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013664256314002354,
+      "loss": 1.1143,
+      "step": 2166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013658701630481898,
+      "loss": 1.1739,
+      "step": 2167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013653145643314514,
+      "loss": 1.179,
+      "step": 2168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013647588354479876,
+      "loss": 1.1639,
+      "step": 2169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013642029765958125,
+      "loss": 1.1338,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013636469879729863,
+      "loss": 1.2208,
+      "step": 2171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013630908697776154,
+      "loss": 1.3334,
+      "step": 2172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013625346222078533,
+      "loss": 1.0992,
+      "step": 2173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001361978245461898,
+      "loss": 1.1338,
+      "step": 2174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013614217397379946,
+      "loss": 1.2534,
+      "step": 2175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013608651052344342,
+      "loss": 1.1597,
+      "step": 2176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013603083421495534,
+      "loss": 1.0833,
+      "step": 2177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013597514506817347,
+      "loss": 1.2514,
+      "step": 2178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013591944310294065,
+      "loss": 1.3631,
+      "step": 2179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013586372833910423,
+      "loss": 1.0257,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001358080007965162,
+      "loss": 1.1199,
+      "step": 2181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013575226049503312,
+      "loss": 1.1184,
+      "step": 2182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013569650745451592,
+      "loss": 1.1917,
+      "step": 2183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001356407416948303,
+      "loss": 1.1305,
+      "step": 2184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013558496323584632,
+      "loss": 1.1415,
+      "step": 2185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013552917209743869,
+      "loss": 1.1899,
+      "step": 2186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013547336829948653,
+      "loss": 1.1525,
+      "step": 2187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013541755186187356,
+      "loss": 1.2563,
+      "step": 2188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013536172280448794,
+      "loss": 1.168,
+      "step": 2189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001353058811472223,
+      "loss": 1.2115,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013525002690997393,
+      "loss": 1.044,
+      "step": 2191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013519416011264435,
+      "loss": 1.0153,
+      "step": 2192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001351382807751398,
+      "loss": 1.1562,
+      "step": 2193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001350823889173708,
+      "loss": 1.1503,
+      "step": 2194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013502648455925243,
+      "loss": 1.1598,
+      "step": 2195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013497056772070418,
+      "loss": 1.2215,
+      "step": 2196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013491463842165008,
+      "loss": 1.2096,
+      "step": 2197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001348586966820184,
+      "loss": 1.1895,
+      "step": 2198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013480274252174206,
+      "loss": 1.2395,
+      "step": 2199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013474677596075836,
+      "loss": 1.1184,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013469079701900887,
+      "loss": 1.1917,
+      "step": 2201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001346348057164397,
+      "loss": 1.2257,
+      "step": 2202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001345788020730014,
+      "loss": 1.1542,
+      "step": 2203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013452278610864878,
+      "loss": 1.3057,
+      "step": 2204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013446675784334116,
+      "loss": 1.0866,
+      "step": 2205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013441071729704225,
+      "loss": 1.1352,
+      "step": 2206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013435466448971997,
+      "loss": 1.1386,
+      "step": 2207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001342985994413468,
+      "loss": 1.1704,
+      "step": 2208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001342425221718995,
+      "loss": 1.4093,
+      "step": 2209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013418643270135915,
+      "loss": 1.2635,
+      "step": 2210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013413033104971126,
+      "loss": 1.1196,
+      "step": 2211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013407421723694567,
+      "loss": 1.2798,
+      "step": 2212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013401809128305643,
+      "loss": 1.2037,
+      "step": 2213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013396195320804208,
+      "loss": 1.2052,
+      "step": 2214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001339058030319054,
+      "loss": 1.2119,
+      "step": 2215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013384964077465346,
+      "loss": 1.2989,
+      "step": 2216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013379346645629768,
+      "loss": 1.1364,
+      "step": 2217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013373728009685376,
+      "loss": 1.2124,
+      "step": 2218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013368108171634167,
+      "loss": 1.3454,
+      "step": 2219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013362487133478573,
+      "loss": 1.2772,
+      "step": 2220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013356864897221446,
+      "loss": 1.1861,
+      "step": 2221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001335124146486607,
+      "loss": 1.1008,
+      "step": 2222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013345616838416152,
+      "loss": 1.1255,
+      "step": 2223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001333999101987582,
+      "loss": 1.1178,
+      "step": 2224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013334364011249642,
+      "loss": 1.0621,
+      "step": 2225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013328735814542597,
+      "loss": 1.2475,
+      "step": 2226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013323106431760087,
+      "loss": 1.3147,
+      "step": 2227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013317475864907947,
+      "loss": 1.0659,
+      "step": 2228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001331184411599242,
+      "loss": 1.1849,
+      "step": 2229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001330621118702018,
+      "loss": 1.2221,
+      "step": 2230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013300577079998325,
+      "loss": 1.1645,
+      "step": 2231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001329494179693436,
+      "loss": 1.2079,
+      "step": 2232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013289305339836215,
+      "loss": 1.2521,
+      "step": 2233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013283667710712243,
+      "loss": 1.1597,
+      "step": 2234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001327802891157121,
+      "loss": 1.1276,
+      "step": 2235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013272388944422298,
+      "loss": 1.0833,
+      "step": 2236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001326674781127511,
+      "loss": 1.3674,
+      "step": 2237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013261105514139656,
+      "loss": 1.1939,
+      "step": 2238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013255462055026369,
+      "loss": 1.179,
+      "step": 2239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013249817435946092,
+      "loss": 1.1928,
+      "step": 2240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013244171658910084,
+      "loss": 1.0918,
+      "step": 2241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013238524725930014,
+      "loss": 1.1268,
+      "step": 2242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013232876639017964,
+      "loss": 1.1262,
+      "step": 2243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013227227400186426,
+      "loss": 1.2725,
+      "step": 2244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013221577011448304,
+      "loss": 1.2306,
+      "step": 2245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321592547481691,
+      "loss": 1.1869,
+      "step": 2246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013210272792305968,
+      "loss": 1.1312,
+      "step": 2247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013204618965929608,
+      "loss": 1.2226,
+      "step": 2248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013198963997702366,
+      "loss": 1.1567,
+      "step": 2249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013193307889639193,
+      "loss": 1.1627,
+      "step": 2250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001318765064375543,
+      "loss": 1.1909,
+      "step": 2251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001318199226206684,
+      "loss": 0.9967,
+      "step": 2252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013176332746589586,
+      "loss": 1.048,
+      "step": 2253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013170672099340226,
+      "loss": 1.205,
+      "step": 2254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013165010322335733,
+      "loss": 1.2069,
+      "step": 2255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001315934741759348,
+      "loss": 1.3226,
+      "step": 2256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013153683387131233,
+      "loss": 1.1234,
+      "step": 2257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013148018232967171,
+      "loss": 1.1628,
+      "step": 2258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001314235195711987,
+      "loss": 1.2717,
+      "step": 2259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013136684561608296,
+      "loss": 1.2405,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013131016048451832,
+      "loss": 1.2467,
+      "step": 2261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001312534641967024,
+      "loss": 0.9505,
+      "step": 2262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013119675677283695,
+      "loss": 1.2074,
+      "step": 2263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001311400382331276,
+      "loss": 1.1934,
+      "step": 2264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013108330859778393,
+      "loss": 1.1102,
+      "step": 2265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013102656788701954,
+      "loss": 1.1612,
+      "step": 2266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000130969816121052,
+      "loss": 1.2512,
+      "step": 2267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013091305332010267,
+      "loss": 1.1914,
+      "step": 2268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013085627950439697,
+      "loss": 1.1964,
+      "step": 2269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013079949469416423,
+      "loss": 1.1138,
+      "step": 2270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013074269890963767,
+      "loss": 1.1724,
+      "step": 2271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013068589217105441,
+      "loss": 1.1047,
+      "step": 2272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013062907449865548,
+      "loss": 1.0098,
+      "step": 2273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013057224591268588,
+      "loss": 1.1356,
+      "step": 2274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001305154064333944,
+      "loss": 1.1933,
+      "step": 2275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001304585560810337,
+      "loss": 1.1862,
+      "step": 2276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013040169487586043,
+      "loss": 1.2482,
+      "step": 2277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013034482283813501,
+      "loss": 1.2311,
+      "step": 2278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013028793998812174,
+      "loss": 1.1315,
+      "step": 2279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013023104634608883,
+      "loss": 1.1464,
+      "step": 2280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001301741419323082,
+      "loss": 1.1436,
+      "step": 2281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013011722676705575,
+      "loss": 1.1881,
+      "step": 2282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013006030087061118,
+      "loss": 1.1082,
+      "step": 2283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001300033642632579,
+      "loss": 1.2172,
+      "step": 2284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012994641696528332,
+      "loss": 1.1622,
+      "step": 2285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012988945899697853,
+      "loss": 1.1371,
+      "step": 2286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012983249037863844,
+      "loss": 1.1989,
+      "step": 2287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012977551113056178,
+      "loss": 1.0517,
+      "step": 2288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012971852127305108,
+      "loss": 1.1667,
+      "step": 2289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001296615208264126,
+      "loss": 1.1243,
+      "step": 2290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012960450981095643,
+      "loss": 1.193,
+      "step": 2291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012954748824699634,
+      "loss": 1.1945,
+      "step": 2292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012949045615484996,
+      "loss": 1.2369,
+      "step": 2293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012943341355483865,
+      "loss": 1.2569,
+      "step": 2294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012937636046728748,
+      "loss": 1.2559,
+      "step": 2295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001293192969125252,
+      "loss": 1.1525,
+      "step": 2296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001292622229108845,
+      "loss": 1.2324,
+      "step": 2297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012920513848270148,
+      "loss": 1.2584,
+      "step": 2298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012914804364831623,
+      "loss": 1.1816,
+      "step": 2299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012909093842807247,
+      "loss": 1.2349,
+      "step": 2300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001290338228423175,
+      "loss": 1.0238,
+      "step": 2301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001289766969114025,
+      "loss": 1.1712,
+      "step": 2302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012891956065568216,
+      "loss": 1.1654,
+      "step": 2303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012886241409551497,
+      "loss": 1.2209,
+      "step": 2304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001288052572512631,
+      "loss": 1.27,
+      "step": 2305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012874809014329228,
+      "loss": 1.1358,
+      "step": 2306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012869091279197196,
+      "loss": 1.2087,
+      "step": 2307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012863372521767526,
+      "loss": 1.0351,
+      "step": 2308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012857652744077892,
+      "loss": 1.1844,
+      "step": 2309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012851931948166327,
+      "loss": 1.2499,
+      "step": 2310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012846210136071236,
+      "loss": 1.1632,
+      "step": 2311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001284048730983138,
+      "loss": 1.075,
+      "step": 2312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001283476347148588,
+      "loss": 1.1557,
+      "step": 2313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012829038623074222,
+      "loss": 0.9654,
+      "step": 2314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012823312766636254,
+      "loss": 1.103,
+      "step": 2315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012817585904212174,
+      "loss": 1.1527,
+      "step": 2316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012811858037842542,
+      "loss": 1.184,
+      "step": 2317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012806129169568283,
+      "loss": 1.1292,
+      "step": 2318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012800399301430676,
+      "loss": 1.1345,
+      "step": 2319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012794668435471345,
+      "loss": 1.3219,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012788936573732281,
+      "loss": 1.1224,
+      "step": 2321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001278320371825583,
+      "loss": 1.1546,
+      "step": 2322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001277746987108469,
+      "loss": 1.2905,
+      "step": 2323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001277173503426191,
+      "loss": 1.1957,
+      "step": 2324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012765999209830888,
+      "loss": 1.127,
+      "step": 2325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012760262399835384,
+      "loss": 1.2173,
+      "step": 2326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012754524606319506,
+      "loss": 1.1525,
+      "step": 2327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012748785831327708,
+      "loss": 1.2898,
+      "step": 2328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012743046076904793,
+      "loss": 1.1994,
+      "step": 2329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012737305345095925,
+      "loss": 1.1015,
+      "step": 2330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000127315636379466,
+      "loss": 1.1919,
+      "step": 2331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001272582095750267,
+      "loss": 1.1455,
+      "step": 2332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012720077305810337,
+      "loss": 1.1077,
+      "step": 2333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012714332684916145,
+      "loss": 1.1889,
+      "step": 2334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012708587096866975,
+      "loss": 1.0604,
+      "step": 2335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012702840543710073,
+      "loss": 1.1839,
+      "step": 2336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012697093027493007,
+      "loss": 1.175,
+      "step": 2337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012691344550263701,
+      "loss": 1.146,
+      "step": 2338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012685595114070422,
+      "loss": 1.1522,
+      "step": 2339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012679844720961774,
+      "loss": 1.2195,
+      "step": 2340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012674093372986697,
+      "loss": 1.3952,
+      "step": 2341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012668341072194484,
+      "loss": 1.1336,
+      "step": 2342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001266258782063476,
+      "loss": 1.0566,
+      "step": 2343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012656833620357485,
+      "loss": 1.0786,
+      "step": 2344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012651078473412968,
+      "loss": 1.1571,
+      "step": 2345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012645322381851844,
+      "loss": 1.1982,
+      "step": 2346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001263956534772509,
+      "loss": 1.1808,
+      "step": 2347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001263380737308402,
+      "loss": 1.2683,
+      "step": 2348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012628048459980282,
+      "loss": 1.1929,
+      "step": 2349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012622288610465856,
+      "loss": 1.2234,
+      "step": 2350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012616527826593058,
+      "loss": 1.1556,
+      "step": 2351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001261076611041454,
+      "loss": 1.1436,
+      "step": 2352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012605003463983272,
+      "loss": 1.1422,
+      "step": 2353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001259923988935258,
+      "loss": 1.3158,
+      "step": 2354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000125934753885761,
+      "loss": 1.2412,
+      "step": 2355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012587709963707804,
+      "loss": 1.0814,
+      "step": 2356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012581943616801996,
+      "loss": 1.0459,
+      "step": 2357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012576176349913303,
+      "loss": 1.2729,
+      "step": 2358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012570408165096689,
+      "loss": 1.21,
+      "step": 2359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001256463906440744,
+      "loss": 1.192,
+      "step": 2360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012558869049901158,
+      "loss": 1.1511,
+      "step": 2361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001255309812363379,
+      "loss": 1.1182,
+      "step": 2362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012547326287661598,
+      "loss": 1.1768,
+      "step": 2363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012541553544041163,
+      "loss": 1.2487,
+      "step": 2364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012535779894829398,
+      "loss": 1.2886,
+      "step": 2365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012530005342083537,
+      "loss": 1.2132,
+      "step": 2366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012524229887861133,
+      "loss": 1.2176,
+      "step": 2367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012518453534220064,
+      "loss": 1.1987,
+      "step": 2368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012512676283218518,
+      "loss": 1.1494,
+      "step": 2369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012506898136915019,
+      "loss": 1.2431,
+      "step": 2370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012501119097368398,
+      "loss": 1.168,
+      "step": 2371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012495339166637812,
+      "loss": 1.1926,
+      "step": 2372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012489558346782728,
+      "loss": 1.2016,
+      "step": 2373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012483776639862938,
+      "loss": 1.254,
+      "step": 2374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001247799404793854,
+      "loss": 1.0338,
+      "step": 2375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012472210573069955,
+      "loss": 1.1688,
+      "step": 2376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012466426217317921,
+      "loss": 1.3834,
+      "step": 2377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001246064098274348,
+      "loss": 1.1624,
+      "step": 2378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012454854871407994,
+      "loss": 1.3367,
+      "step": 2379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012449067885373133,
+      "loss": 1.1309,
+      "step": 2380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001244328002670089,
+      "loss": 1.1796,
+      "step": 2381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012437491297453553,
+      "loss": 1.1653,
+      "step": 2382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001243170169969373,
+      "loss": 1.1182,
+      "step": 2383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012425911235484333,
+      "loss": 1.1529,
+      "step": 2384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012420119906888597,
+      "loss": 1.2708,
+      "step": 2385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001241432771597004,
+      "loss": 1.1637,
+      "step": 2386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012408534664792512,
+      "loss": 1.238,
+      "step": 2387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001240274075542015,
+      "loss": 1.119,
+      "step": 2388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012396945989917411,
+      "loss": 1.1852,
+      "step": 2389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012391150370349055,
+      "loss": 1.1691,
+      "step": 2390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012385353898780135,
+      "loss": 1.2312,
+      "step": 2391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012379556577276016,
+      "loss": 1.1804,
+      "step": 2392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012373758407902374,
+      "loss": 1.0994,
+      "step": 2393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001236795939272517,
+      "loss": 1.1909,
+      "step": 2394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012362159533810672,
+      "loss": 1.1752,
+      "step": 2395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012356358833225464,
+      "loss": 1.1164,
+      "step": 2396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012350557293036406,
+      "loss": 1.1759,
+      "step": 2397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012344754915310678,
+      "loss": 1.0387,
+      "step": 2398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012338951702115737,
+      "loss": 1.2502,
+      "step": 2399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001233314765551936,
+      "loss": 1.0502,
+      "step": 2400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012327342777589606,
+      "loss": 1.0923,
+      "step": 2401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012321537070394836,
+      "loss": 1.1018,
+      "step": 2402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012315730536003702,
+      "loss": 1.2402,
+      "step": 2403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001230992317648516,
+      "loss": 1.1782,
+      "step": 2404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012304114993908448,
+      "loss": 1.2401,
+      "step": 2405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012298305990343108,
+      "loss": 1.1519,
+      "step": 2406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012292496167858966,
+      "loss": 1.1616,
+      "step": 2407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012286685528526146,
+      "loss": 1.107,
+      "step": 2408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012280874074415064,
+      "loss": 1.2418,
+      "step": 2409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012275061807596416,
+      "loss": 1.1631,
+      "step": 2410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012269248730141193,
+      "loss": 1.0946,
+      "step": 2411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012263434844120687,
+      "loss": 1.1494,
+      "step": 2412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001225762015160646,
+      "loss": 1.215,
+      "step": 2413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001225180465467037,
+      "loss": 1.2676,
+      "step": 2414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012245988355384562,
+      "loss": 1.1574,
+      "step": 2415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001224017125582146,
+      "loss": 1.0766,
+      "step": 2416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012234353358053784,
+      "loss": 1.199,
+      "step": 2417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012228534664154536,
+      "loss": 1.1849,
+      "step": 2418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012222715176196991,
+      "loss": 1.1376,
+      "step": 2419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001221689489625472,
+      "loss": 1.0291,
+      "step": 2420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012211073826401566,
+      "loss": 1.041,
+      "step": 2421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012205251968711657,
+      "loss": 1.1085,
+      "step": 2422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012199429325259412,
+      "loss": 1.1747,
+      "step": 2423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012193605898119514,
+      "loss": 1.0536,
+      "step": 2424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012187781689366933,
+      "loss": 1.099,
+      "step": 2425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012181956701076918,
+      "loss": 1.2654,
+      "step": 2426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012176130935324991,
+      "loss": 1.1345,
+      "step": 2427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001217030439418696,
+      "loss": 1.2354,
+      "step": 2428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012164477079738901,
+      "loss": 1.0145,
+      "step": 2429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012158648994057167,
+      "loss": 1.2244,
+      "step": 2430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012152820139218389,
+      "loss": 1.0166,
+      "step": 2431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012146990517299466,
+      "loss": 1.3502,
+      "step": 2432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012141160130377578,
+      "loss": 1.1571,
+      "step": 2433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012135328980530176,
+      "loss": 0.9717,
+      "step": 2434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012129497069834977,
+      "loss": 1.0889,
+      "step": 2435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012123664400369972,
+      "loss": 1.2496,
+      "step": 2436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012117830974213428,
+      "loss": 1.0968,
+      "step": 2437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012111996793443874,
+      "loss": 1.1994,
+      "step": 2438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001210616186014011,
+      "loss": 1.0901,
+      "step": 2439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012100326176381208,
+      "loss": 1.2122,
+      "step": 2440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000120944897442465,
+      "loss": 1.2938,
+      "step": 2441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012088652565815594,
+      "loss": 1.1401,
+      "step": 2442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012082814643168357,
+      "loss": 1.1358,
+      "step": 2443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012076975978384917,
+      "loss": 1.1072,
+      "step": 2444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012071136573545683,
+      "loss": 1.0365,
+      "step": 2445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001206529643073131,
+      "loss": 1.2713,
+      "step": 2446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012059455552022725,
+      "loss": 1.0882,
+      "step": 2447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012053613939501119,
+      "loss": 1.1342,
+      "step": 2448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012047771595247934,
+      "loss": 1.1563,
+      "step": 2449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012041928521344883,
+      "loss": 1.1356,
+      "step": 2450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012036084719873936,
+      "loss": 1.1777,
+      "step": 2451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012030240192917321,
+      "loss": 1.2123,
+      "step": 2452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012024394942557529,
+      "loss": 1.1274,
+      "step": 2453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012018548970877299,
+      "loss": 1.2835,
+      "step": 2454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012012702279959633,
+      "loss": 1.1401,
+      "step": 2455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012006854871887798,
+      "loss": 1.2389,
+      "step": 2456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012001006748745299,
+      "loss": 1.1614,
+      "step": 2457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011995157912615905,
+      "loss": 1.1671,
+      "step": 2458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011989308365583645,
+      "loss": 1.1242,
+      "step": 2459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011983458109732791,
+      "loss": 1.1436,
+      "step": 2460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011977607147147869,
+      "loss": 1.1157,
+      "step": 2461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011971755479913665,
+      "loss": 1.0812,
+      "step": 2462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011965903110115206,
+      "loss": 1.1483,
+      "step": 2463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011960050039837778,
+      "loss": 1.1711,
+      "step": 2464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011954196271166906,
+      "loss": 1.1187,
+      "step": 2465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011948341806188377,
+      "loss": 1.0817,
+      "step": 2466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011942486646988219,
+      "loss": 1.1482,
+      "step": 2467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011936630795652699,
+      "loss": 1.1816,
+      "step": 2468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011930774254268348,
+      "loss": 1.0892,
+      "step": 2469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011924917024921935,
+      "loss": 1.1977,
+      "step": 2470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011919059109700465,
+      "loss": 1.086,
+      "step": 2471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011913200510691203,
+      "loss": 1.1638,
+      "step": 2472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001190734122998165,
+      "loss": 1.1315,
+      "step": 2473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011901481269659547,
+      "loss": 1.1125,
+      "step": 2474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011895620631812883,
+      "loss": 1.0866,
+      "step": 2475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011889759318529882,
+      "loss": 1.1153,
+      "step": 2476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011883897331899016,
+      "loss": 1.1876,
+      "step": 2477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011878034674008993,
+      "loss": 1.2004,
+      "step": 2478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001187217134694876,
+      "loss": 1.1519,
+      "step": 2479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011866307352807503,
+      "loss": 1.1258,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011860442693674647,
+      "loss": 1.1298,
+      "step": 2481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011854577371639848,
+      "loss": 1.1432,
+      "step": 2482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011848711388793008,
+      "loss": 1.1132,
+      "step": 2483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001184284474722426,
+      "loss": 1.0624,
+      "step": 2484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011836977449023967,
+      "loss": 0.9779,
+      "step": 2485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001183110949628273,
+      "loss": 1.1777,
+      "step": 2486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001182524089109139,
+      "loss": 1.0849,
+      "step": 2487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011819371635541003,
+      "loss": 1.1167,
+      "step": 2488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011813501731722875,
+      "loss": 1.1642,
+      "step": 2489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011807631181728536,
+      "loss": 1.2354,
+      "step": 2490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011801759987649741,
+      "loss": 1.2188,
+      "step": 2491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001179588815157848,
+      "loss": 0.9772,
+      "step": 2492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011790015675606976,
+      "loss": 1.0416,
+      "step": 2493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011784142561827669,
+      "loss": 1.0628,
+      "step": 2494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011778268812333233,
+      "loss": 1.1692,
+      "step": 2495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011772394429216571,
+      "loss": 1.1143,
+      "step": 2496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011766519414570805,
+      "loss": 1.1468,
+      "step": 2497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011760643770489288,
+      "loss": 0.9961,
+      "step": 2498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011754767499065593,
+      "loss": 1.0917,
+      "step": 2499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011748890602393521,
+      "loss": 1.0489,
+      "step": 2500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011743013082567087,
+      "loss": 1.1545,
+      "step": 2501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011737134941680544,
+      "loss": 1.0226,
+      "step": 2502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011731256181828346,
+      "loss": 1.2031,
+      "step": 2503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011725376805105182,
+      "loss": 1.0091,
+      "step": 2504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011719496813605962,
+      "loss": 1.2643,
+      "step": 2505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011713616209425799,
+      "loss": 1.1435,
+      "step": 2506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011707734994660044,
+      "loss": 1.2261,
+      "step": 2507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011701853171404256,
+      "loss": 1.2295,
+      "step": 2508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011695970741754206,
+      "loss": 1.1587,
+      "step": 2509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001169008770780589,
+      "loss": 1.3099,
+      "step": 2510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011684204071655519,
+      "loss": 1.0988,
+      "step": 2511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011678319835399509,
+      "loss": 1.2338,
+      "step": 2512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011672435001134501,
+      "loss": 1.1452,
+      "step": 2513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011666549570957341,
+      "loss": 1.0466,
+      "step": 2514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011660663546965094,
+      "loss": 1.0122,
+      "step": 2515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011654776931255033,
+      "loss": 1.2486,
+      "step": 2516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001164888972592464,
+      "loss": 1.1329,
+      "step": 2517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011643001933071609,
+      "loss": 1.3244,
+      "step": 2518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011637113554793846,
+      "loss": 1.29,
+      "step": 2519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011631224593189461,
+      "loss": 1.0814,
+      "step": 2520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011625335050356771,
+      "loss": 1.0645,
+      "step": 2521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011619444928394312,
+      "loss": 1.0967,
+      "step": 2522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011613554229400809,
+      "loss": 1.1293,
+      "step": 2523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011607662955475204,
+      "loss": 1.0466,
+      "step": 2524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011601771108716646,
+      "loss": 1.1332,
+      "step": 2525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011595878691224473,
+      "loss": 1.1365,
+      "step": 2526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011589985705098245,
+      "loss": 1.119,
+      "step": 2527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011584092152437711,
+      "loss": 1.0807,
+      "step": 2528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011578198035342826,
+      "loss": 1.0951,
+      "step": 2529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011572303355913754,
+      "loss": 1.1518,
+      "step": 2530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011566408116250842,
+      "loss": 1.096,
+      "step": 2531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011560512318454655,
+      "loss": 1.0141,
+      "step": 2532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011554615964625948,
+      "loss": 1.092,
+      "step": 2533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011548719056865671,
+      "loss": 1.2266,
+      "step": 2534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011542821597274978,
+      "loss": 1.1931,
+      "step": 2535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011536923587955218,
+      "loss": 1.1408,
+      "step": 2536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001153102503100793,
+      "loss": 1.161,
+      "step": 2537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011525125928534859,
+      "loss": 1.2919,
+      "step": 2538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011519226282637934,
+      "loss": 1.0823,
+      "step": 2539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011513326095419281,
+      "loss": 1.1829,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011507425368981228,
+      "loss": 0.8794,
+      "step": 2541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011501524105426278,
+      "loss": 1.2194,
+      "step": 2542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011495622306857136,
+      "loss": 1.2199,
+      "step": 2543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000114897199753767,
+      "loss": 1.0803,
+      "step": 2544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011483817113088052,
+      "loss": 1.1077,
+      "step": 2545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011477913722094462,
+      "loss": 1.1179,
+      "step": 2546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011472009804499396,
+      "loss": 1.2917,
+      "step": 2547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011466105362406498,
+      "loss": 1.1322,
+      "step": 2548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001146020039791961,
+      "loss": 0.9531,
+      "step": 2549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011454294913142747,
+      "loss": 1.0986,
+      "step": 2550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011448388910180121,
+      "loss": 1.1656,
+      "step": 2551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011442482391136124,
+      "loss": 1.2494,
+      "step": 2552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011436575358115331,
+      "loss": 1.209,
+      "step": 2553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000114306678132225,
+      "loss": 1.2455,
+      "step": 2554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011424759758562575,
+      "loss": 1.1988,
+      "step": 2555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011418851196240677,
+      "loss": 0.9714,
+      "step": 2556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001141294212836211,
+      "loss": 1.1598,
+      "step": 2557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011407032557032363,
+      "loss": 1.1304,
+      "step": 2558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001140112248435709,
+      "loss": 1.1281,
+      "step": 2559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011395211912442141,
+      "loss": 1.2328,
+      "step": 2560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011389300843393533,
+      "loss": 1.076,
+      "step": 2561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011383389279317463,
+      "loss": 1.1091,
+      "step": 2562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011377477222320308,
+      "loss": 1.189,
+      "step": 2563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001137156467450861,
+      "loss": 1.1025,
+      "step": 2564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011365651637989099,
+      "loss": 1.1351,
+      "step": 2565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011359738114868674,
+      "loss": 1.1474,
+      "step": 2566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000113538241072544,
+      "loss": 1.1408,
+      "step": 2567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001134790961725353,
+      "loss": 1.2802,
+      "step": 2568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011341994646973475,
+      "loss": 1.2954,
+      "step": 2569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011336079198521821,
+      "loss": 1.2128,
+      "step": 2570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011330163274006328,
+      "loss": 1.101,
+      "step": 2571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011324246875534923,
+      "loss": 1.2731,
+      "step": 2572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011318330005215701,
+      "loss": 1.3144,
+      "step": 2573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011312412665156929,
+      "loss": 1.1286,
+      "step": 2574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011306494857467034,
+      "loss": 1.2133,
+      "step": 2575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011300576584254617,
+      "loss": 1.1499,
+      "step": 2576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011294657847628445,
+      "loss": 1.181,
+      "step": 2577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011288738649697442,
+      "loss": 1.1862,
+      "step": 2578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011282818992570702,
+      "loss": 1.2001,
+      "step": 2579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011276898878357489,
+      "loss": 1.1466,
+      "step": 2580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011270978309167215,
+      "loss": 1.0682,
+      "step": 2581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011265057287109466,
+      "loss": 1.3249,
+      "step": 2582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011259135814293986,
+      "loss": 1.1241,
+      "step": 2583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011253213892830676,
+      "loss": 1.1139,
+      "step": 2584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011247291524829605,
+      "loss": 1.0491,
+      "step": 2585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011241368712400991,
+      "loss": 1.2558,
+      "step": 2586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011235445457655218,
+      "loss": 1.083,
+      "step": 2587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011229521762702826,
+      "loss": 1.2925,
+      "step": 2588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001122359762965451,
+      "loss": 1.2176,
+      "step": 2589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001121767306062112,
+      "loss": 1.1209,
+      "step": 2590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011211748057713666,
+      "loss": 1.1409,
+      "step": 2591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001120582262304331,
+      "loss": 1.0563,
+      "step": 2592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011199896758721369,
+      "loss": 1.1717,
+      "step": 2593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011193970466859308,
+      "loss": 1.1957,
+      "step": 2594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011188043749568751,
+      "loss": 1.0123,
+      "step": 2595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001118211660896147,
+      "loss": 1.0432,
+      "step": 2596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011176189047149388,
+      "loss": 1.09,
+      "step": 2597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001117026106624458,
+      "loss": 1.1628,
+      "step": 2598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011164332668359272,
+      "loss": 1.149,
+      "step": 2599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011158403855605828,
+      "loss": 1.0062,
+      "step": 2600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011152474630096775,
+      "loss": 1.0715,
+      "step": 2601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011146544993944776,
+      "loss": 1.2313,
+      "step": 2602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011140614949262642,
+      "loss": 1.1178,
+      "step": 2603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001113468449816334,
+      "loss": 1.1836,
+      "step": 2604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011128753642759961,
+      "loss": 1.2046,
+      "step": 2605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011122822385165761,
+      "loss": 1.2259,
+      "step": 2606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001111689072749413,
+      "loss": 0.9961,
+      "step": 2607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011110958671858596,
+      "loss": 1.2199,
+      "step": 2608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011105026220372836,
+      "loss": 1.1327,
+      "step": 2609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011099093375150673,
+      "loss": 1.2387,
+      "step": 2610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011093160138306054,
+      "loss": 1.1808,
+      "step": 2611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011087226511953077,
+      "loss": 1.191,
+      "step": 2612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011081292498205981,
+      "loss": 1.0799,
+      "step": 2613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011075358099179137,
+      "loss": 1.217,
+      "step": 2614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011069423316987053,
+      "loss": 1.2011,
+      "step": 2615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011063488153744375,
+      "loss": 1.09,
+      "step": 2616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011057552611565887,
+      "loss": 1.1407,
+      "step": 2617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011051616692566506,
+      "loss": 1.0737,
+      "step": 2618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011045680398861284,
+      "loss": 1.1981,
+      "step": 2619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011039743732565404,
+      "loss": 1.1588,
+      "step": 2620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011033806695794189,
+      "loss": 1.0824,
+      "step": 2621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011027869290663083,
+      "loss": 1.1839,
+      "step": 2622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011021931519287669,
+      "loss": 1.092,
+      "step": 2623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011015993383783658,
+      "loss": 1.1764,
+      "step": 2624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011010054886266894,
+      "loss": 1.2329,
+      "step": 2625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011004116028853346,
+      "loss": 1.13,
+      "step": 2626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010998176813659111,
+      "loss": 1.0489,
+      "step": 2627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010992237242800416,
+      "loss": 1.0591,
+      "step": 2628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010986297318393615,
+      "loss": 1.1447,
+      "step": 2629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010980357042555182,
+      "loss": 1.2702,
+      "step": 2630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010974416417401729,
+      "loss": 1.3087,
+      "step": 2631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001096847544504998,
+      "loss": 1.1993,
+      "step": 2632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010962534127616784,
+      "loss": 1.1942,
+      "step": 2633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001095659246721912,
+      "loss": 1.1287,
+      "step": 2634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010950650465974088,
+      "loss": 1.021,
+      "step": 2635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000109447081259989,
+      "loss": 1.1218,
+      "step": 2636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010938765449410899,
+      "loss": 1.2724,
+      "step": 2637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010932822438327546,
+      "loss": 1.026,
+      "step": 2638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010926879094866416,
+      "loss": 1.1191,
+      "step": 2639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010920935421145206,
+      "loss": 1.0635,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010914991419281738,
+      "loss": 1.1212,
+      "step": 2641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010909047091393931,
+      "loss": 1.1606,
+      "step": 2642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010903102439599844,
+      "loss": 1.2255,
+      "step": 2643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010897157466017636,
+      "loss": 1.1133,
+      "step": 2644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010891212172765583,
+      "loss": 1.1296,
+      "step": 2645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001088526656196208,
+      "loss": 1.1277,
+      "step": 2646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010879320635725632,
+      "loss": 1.1991,
+      "step": 2647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010873374396174852,
+      "loss": 1.0973,
+      "step": 2648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001086742784542847,
+      "loss": 1.2157,
+      "step": 2649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001086148098560533,
+      "loss": 1.3286,
+      "step": 2650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010855533818824379,
+      "loss": 1.2743,
+      "step": 2651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010849586347204676,
+      "loss": 1.1059,
+      "step": 2652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010843638572865391,
+      "loss": 1.2247,
+      "step": 2653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010837690497925796,
+      "loss": 1.2212,
+      "step": 2654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010831742124505277,
+      "loss": 1.074,
+      "step": 2655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.123,
+      "step": 2656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010819844490699529,
+      "loss": 1.0697,
+      "step": 2657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010813895234553593,
+      "loss": 1.178,
+      "step": 2658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010807945688405321,
+      "loss": 1.2779,
+      "step": 2659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010801995854374616,
+      "loss": 1.1658,
+      "step": 2660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010796045734581492,
+      "loss": 1.1164,
+      "step": 2661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001079009533114606,
+      "loss": 1.1152,
+      "step": 2662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001078414464618853,
+      "loss": 1.2467,
+      "step": 2663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010778193681829216,
+      "loss": 1.2853,
+      "step": 2664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010772242440188534,
+      "loss": 1.1149,
+      "step": 2665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001076629092338699,
+      "loss": 1.2443,
+      "step": 2666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010760339133545196,
+      "loss": 1.2211,
+      "step": 2667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010754387072783858,
+      "loss": 1.0768,
+      "step": 2668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010748434743223777,
+      "loss": 1.1974,
+      "step": 2669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010742482146985857,
+      "loss": 1.2262,
+      "step": 2670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010736529286191086,
+      "loss": 1.1593,
+      "step": 2671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010730576162960555,
+      "loss": 1.2016,
+      "step": 2672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001072462277941545,
+      "loss": 1.1254,
+      "step": 2673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010718669137677038,
+      "loss": 1.1648,
+      "step": 2674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010712715239866687,
+      "loss": 1.0671,
+      "step": 2675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010706761088105858,
+      "loss": 1.1664,
+      "step": 2676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010700806684516096,
+      "loss": 1.1348,
+      "step": 2677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010694852031219042,
+      "loss": 1.1004,
+      "step": 2678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010688897130336419,
+      "loss": 1.1889,
+      "step": 2679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010682941983990045,
+      "loss": 1.1503,
+      "step": 2680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010676986594301823,
+      "loss": 1.2464,
+      "step": 2681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001067103096339374,
+      "loss": 1.1898,
+      "step": 2682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010665075093387871,
+      "loss": 1.0809,
+      "step": 2683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001065911898640638,
+      "loss": 1.0276,
+      "step": 2684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010653162644571506,
+      "loss": 1.1817,
+      "step": 2685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010647206070005581,
+      "loss": 1.0228,
+      "step": 2686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010641249264831019,
+      "loss": 1.1234,
+      "step": 2687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010635292231170309,
+      "loss": 1.1554,
+      "step": 2688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010629334971146028,
+      "loss": 1.0802,
+      "step": 2689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001062337748688083,
+      "loss": 1.1457,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010617419780497452,
+      "loss": 1.0177,
+      "step": 2691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010611461854118714,
+      "loss": 1.1415,
+      "step": 2692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010605503709867499,
+      "loss": 1.2212,
+      "step": 2693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010599545349866782,
+      "loss": 1.225,
+      "step": 2694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010593586776239614,
+      "loss": 1.2349,
+      "step": 2695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010587627991109112,
+      "loss": 1.1879,
+      "step": 2696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010581668996598482,
+      "loss": 1.1246,
+      "step": 2697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010575709794830994,
+      "loss": 1.0883,
+      "step": 2698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010569750387929998,
+      "loss": 1.1192,
+      "step": 2699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010563790778018912,
+      "loss": 1.0926,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010557830967221227,
+      "loss": 1.201,
+      "step": 2701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010551870957660513,
+      "loss": 1.1383,
+      "step": 2702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010545910751460401,
+      "loss": 1.2806,
+      "step": 2703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010539950350744598,
+      "loss": 1.233,
+      "step": 2704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001053398975763688,
+      "loss": 1.0361,
+      "step": 2705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010528028974261091,
+      "loss": 1.074,
+      "step": 2706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010522068002741136,
+      "loss": 1.0866,
+      "step": 2707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010516106845200999,
+      "loss": 1.1422,
+      "step": 2708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010510145503764726,
+      "loss": 1.1416,
+      "step": 2709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010504183980556419,
+      "loss": 1.0377,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010498222277700261,
+      "loss": 1.1315,
+      "step": 2711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010492260397320483,
+      "loss": 1.0963,
+      "step": 2712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010486298341541393,
+      "loss": 1.1728,
+      "step": 2713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010480336112487357,
+      "loss": 1.0906,
+      "step": 2714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010474373712282793,
+      "loss": 1.1162,
+      "step": 2715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010468411143052194,
+      "loss": 1.0348,
+      "step": 2716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010462448406920108,
+      "loss": 1.1426,
+      "step": 2717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010456485506011138,
+      "loss": 1.0521,
+      "step": 2718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010450522442449952,
+      "loss": 1.1511,
+      "step": 2719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010444559218361277,
+      "loss": 1.236,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010438595835869885,
+      "loss": 1.1441,
+      "step": 2721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010432632297100621,
+      "loss": 1.2428,
+      "step": 2722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010426668604178376,
+      "loss": 1.0873,
+      "step": 2723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010420704759228093,
+      "loss": 1.1771,
+      "step": 2724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010414740764374782,
+      "loss": 1.1065,
+      "step": 2725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010408776621743493,
+      "loss": 1.1559,
+      "step": 2726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010402812333459333,
+      "loss": 0.982,
+      "step": 2727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010396847901647468,
+      "loss": 1.1638,
+      "step": 2728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010390883328433103,
+      "loss": 1.1073,
+      "step": 2729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010384918615941502,
+      "loss": 1.1431,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010378953766297976,
+      "loss": 1.1682,
+      "step": 2731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010372988781627884,
+      "loss": 0.9678,
+      "step": 2732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010367023664056638,
+      "loss": 1.1385,
+      "step": 2733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010361058415709687,
+      "loss": 1.2598,
+      "step": 2734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010355093038712538,
+      "loss": 1.2064,
+      "step": 2735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010349127535190737,
+      "loss": 1.1375,
+      "step": 2736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010343161907269876,
+      "loss": 1.1605,
+      "step": 2737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010337196157075593,
+      "loss": 1.1404,
+      "step": 2738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010331230286733573,
+      "loss": 1.2019,
+      "step": 2739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010325264298369532,
+      "loss": 1.0539,
+      "step": 2740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010319298194109241,
+      "loss": 1.0801,
+      "step": 2741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010313331976078505,
+      "loss": 1.0284,
+      "step": 2742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010307365646403171,
+      "loss": 1.1523,
+      "step": 2743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001030139920720913,
+      "loss": 1.038,
+      "step": 2744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010295432660622304,
+      "loss": 1.0257,
+      "step": 2745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001028946600876866,
+      "loss": 1.0239,
+      "step": 2746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000102834992537742,
+      "loss": 1.1311,
+      "step": 2747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010277532397764959,
+      "loss": 1.1001,
+      "step": 2748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001027156544286702,
+      "loss": 1.0652,
+      "step": 2749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010265598391206486,
+      "loss": 1.256,
+      "step": 2750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010259631244909502,
+      "loss": 1.0919,
+      "step": 2751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001025366400610225,
+      "loss": 1.0199,
+      "step": 2752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010247696676910939,
+      "loss": 1.0619,
+      "step": 2753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010241729259461812,
+      "loss": 1.0408,
+      "step": 2754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010235761755881148,
+      "loss": 1.068,
+      "step": 2755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010229794168295245,
+      "loss": 1.0859,
+      "step": 2756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010223826498830444,
+      "loss": 1.2821,
+      "step": 2757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001021785874961311,
+      "loss": 1.1864,
+      "step": 2758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010211890922769631,
+      "loss": 1.1406,
+      "step": 2759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010205923020426434,
+      "loss": 1.1202,
+      "step": 2760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010199955044709965,
+      "loss": 1.2381,
+      "step": 2761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010193986997746697,
+      "loss": 1.1271,
+      "step": 2762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010188018881663126,
+      "loss": 1.1877,
+      "step": 2763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010182050698585781,
+      "loss": 1.1282,
+      "step": 2764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010176082450641207,
+      "loss": 1.1597,
+      "step": 2765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010170114139955975,
+      "loss": 1.0115,
+      "step": 2766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010164145768656679,
+      "loss": 1.3792,
+      "step": 2767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010158177338869932,
+      "loss": 1.2977,
+      "step": 2768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010152208852722374,
+      "loss": 1.2069,
+      "step": 2769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010146240312340656,
+      "loss": 1.1965,
+      "step": 2770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010140271719851453,
+      "loss": 1.1649,
+      "step": 2771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010134303077381464,
+      "loss": 1.132,
+      "step": 2772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010128334387057394,
+      "loss": 1.1908,
+      "step": 2773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010122365651005977,
+      "loss": 1.3693,
+      "step": 2774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001011639687135395,
+      "loss": 1.1898,
+      "step": 2775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001011042805022808,
+      "loss": 1.0139,
+      "step": 2776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010104459189755142,
+      "loss": 1.3464,
+      "step": 2777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001009849029206192,
+      "loss": 1.0817,
+      "step": 2778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001009252135927522,
+      "loss": 1.1611,
+      "step": 2779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010086552393521859,
+      "loss": 1.2593,
+      "step": 2780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001008058339692866,
+      "loss": 1.1097,
+      "step": 2781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010074614371622459,
+      "loss": 1.2347,
+      "step": 2782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001006864531973011,
+      "loss": 1.0658,
+      "step": 2783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010062676243378465,
+      "loss": 1.1776,
+      "step": 2784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001005670714469439,
+      "loss": 1.1776,
+      "step": 2785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010050738025804764,
+      "loss": 1.2015,
+      "step": 2786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010044768888836462,
+      "loss": 1.1661,
+      "step": 2787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010038799735916374,
+      "loss": 1.134,
+      "step": 2788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010032830569171396,
+      "loss": 1.3523,
+      "step": 2789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010026861390728422,
+      "loss": 1.1579,
+      "step": 2790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010020892202714356,
+      "loss": 1.0133,
+      "step": 2791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010014923007256105,
+      "loss": 1.1,
+      "step": 2792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010008953806480575,
+      "loss": 1.0181,
+      "step": 2793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010002984602514678,
+      "loss": 1.0146,
+      "step": 2794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997015397485326e-05,
+      "loss": 1.1794,
+      "step": 2795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991046193519428e-05,
+      "loss": 1.117,
+      "step": 2796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985076992743897e-05,
+      "loss": 1.1827,
+      "step": 2797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979107797285648e-05,
+      "loss": 1.0759,
+      "step": 2798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97313860927158e-05,
+      "loss": 1.2887,
+      "step": 2799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967169430828608e-05,
+      "loss": 1.1825,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961200264083627e-05,
+      "loss": 1.0948,
+      "step": 2801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95523111116354e-05,
+      "loss": 1.0327,
+      "step": 2802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949261974195238e-05,
+      "loss": 1.1337,
+      "step": 2803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943292855305612e-05,
+      "loss": 1.1318,
+      "step": 2804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937323756621537e-05,
+      "loss": 1.0624,
+      "step": 2805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931354680269893e-05,
+      "loss": 1.183,
+      "step": 2806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925385628377543e-05,
+      "loss": 1.1029,
+      "step": 2807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919416603071341e-05,
+      "loss": 1.2817,
+      "step": 2808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.913447606478142e-05,
+      "loss": 1.1089,
+      "step": 2809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907478640724781e-05,
+      "loss": 1.051,
+      "step": 2810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901509707938084e-05,
+      "loss": 1.3055,
+      "step": 2811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895540810244863e-05,
+      "loss": 1.1214,
+      "step": 2812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889571949771922e-05,
+      "loss": 1.2766,
+      "step": 2813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883603128646052e-05,
+      "loss": 1.1544,
+      "step": 2814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877634348994027e-05,
+      "loss": 1.1618,
+      "step": 2815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87166561294261e-05,
+      "loss": 1.1811,
+      "step": 2816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86569692261854e-05,
+      "loss": 1.1813,
+      "step": 2817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85972828014855e-05,
+      "loss": 1.06,
+      "step": 2818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853759687659346e-05,
+      "loss": 1.0418,
+      "step": 2819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847791147277627e-05,
+      "loss": 1.242,
+      "step": 2820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.841822661130067e-05,
+      "loss": 1.1247,
+      "step": 2821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835854231343325e-05,
+      "loss": 1.2734,
+      "step": 2822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829885860044028e-05,
+      "loss": 1.0058,
+      "step": 2823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823917549358797e-05,
+      "loss": 1.1912,
+      "step": 2824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817949301414222e-05,
+      "loss": 1.1374,
+      "step": 2825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811981118336875e-05,
+      "loss": 1.1321,
+      "step": 2826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806013002253304e-05,
+      "loss": 1.0034,
+      "step": 2827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800044955290038e-05,
+      "loss": 1.2428,
+      "step": 2828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794076979573567e-05,
+      "loss": 1.0916,
+      "step": 2829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78810907723037e-05,
+      "loss": 1.2001,
+      "step": 2830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.782141250386892e-05,
+      "loss": 1.0784,
+      "step": 2831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776173501169557e-05,
+      "loss": 1.0897,
+      "step": 2832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77020583170476e-05,
+      "loss": 1.0899,
+      "step": 2833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764238244118858e-05,
+      "loss": 0.9825,
+      "step": 2834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75827074053819e-05,
+      "loss": 1.1774,
+      "step": 2835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752303323089062e-05,
+      "loss": 1.0733,
+      "step": 2836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746335993897751e-05,
+      "loss": 1.2107,
+      "step": 2837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740368755090498e-05,
+      "loss": 1.2573,
+      "step": 2838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734401608793519e-05,
+      "loss": 1.0779,
+      "step": 2839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728434557132984e-05,
+      "loss": 1.1124,
+      "step": 2840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.722467602235042e-05,
+      "loss": 1.1838,
+      "step": 2841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716500746225802e-05,
+      "loss": 1.0211,
+      "step": 2842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710533991231341e-05,
+      "loss": 1.2617,
+      "step": 2843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704567339377701e-05,
+      "loss": 1.1832,
+      "step": 2844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698600792790875e-05,
+      "loss": 1.2816,
+      "step": 2845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69263435359683e-05,
+      "loss": 1.1304,
+      "step": 2846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686668023921496e-05,
+      "loss": 1.1654,
+      "step": 2847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.680701805890761e-05,
+      "loss": 1.0638,
+      "step": 2848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.674735701630469e-05,
+      "loss": 0.9831,
+      "step": 2849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668769713266432e-05,
+      "loss": 1.1964,
+      "step": 2850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662803842924408e-05,
+      "loss": 1.1528,
+      "step": 2851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656838092730126e-05,
+      "loss": 1.276,
+      "step": 2852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650872464809265e-05,
+      "loss": 1.137,
+      "step": 2853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644906961287463e-05,
+      "loss": 1.0677,
+      "step": 2854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638941584290318e-05,
+      "loss": 1.1631,
+      "step": 2855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632976335943367e-05,
+      "loss": 1.1207,
+      "step": 2856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627011218372117e-05,
+      "loss": 0.9687,
+      "step": 2857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621046233702026e-05,
+      "loss": 0.9849,
+      "step": 2858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615081384058499e-05,
+      "loss": 1.2067,
+      "step": 2859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609116671566898e-05,
+      "loss": 1.0645,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603152098352537e-05,
+      "loss": 1.0706,
+      "step": 2861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597187666540668e-05,
+      "loss": 1.224,
+      "step": 2862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59122337825651e-05,
+      "loss": 1.2859,
+      "step": 2863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58525923562522e-05,
+      "loss": 1.2456,
+      "step": 2864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579295240771906e-05,
+      "loss": 1.1496,
+      "step": 2865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57333139582163e-05,
+      "loss": 1.1208,
+      "step": 2866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567367702899382e-05,
+      "loss": 1.2035,
+      "step": 2867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561404164130116e-05,
+      "loss": 1.1492,
+      "step": 2868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555440781638727e-05,
+      "loss": 1.2671,
+      "step": 2869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549477557550047e-05,
+      "loss": 1.2542,
+      "step": 2870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543514493988862e-05,
+      "loss": 1.1468,
+      "step": 2871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537551593079897e-05,
+      "loss": 1.1828,
+      "step": 2872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531588856947808e-05,
+      "loss": 1.2418,
+      "step": 2873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525626287717208e-05,
+      "loss": 1.1855,
+      "step": 2874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.519663887512647e-05,
+      "loss": 1.1575,
+      "step": 2875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513701658458606e-05,
+      "loss": 1.1831,
+      "step": 2876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507739602679519e-05,
+      "loss": 1.0928,
+      "step": 2877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501777722299744e-05,
+      "loss": 1.2053,
+      "step": 2878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495816019443584e-05,
+      "loss": 1.2392,
+      "step": 2879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489854496235278e-05,
+      "loss": 1.1332,
+      "step": 2880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483893154799001e-05,
+      "loss": 1.0718,
+      "step": 2881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477931997258864e-05,
+      "loss": 1.0826,
+      "step": 2882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.471971025738914e-05,
+      "loss": 1.2234,
+      "step": 2883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466010242363123e-05,
+      "loss": 1.0699,
+      "step": 2884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460049649255403e-05,
+      "loss": 1.2242,
+      "step": 2885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4540892485396e-05,
+      "loss": 1.089,
+      "step": 2886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448129042339488e-05,
+      "loss": 1.0964,
+      "step": 2887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442169032778777e-05,
+      "loss": 1.1245,
+      "step": 2888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436209221981093e-05,
+      "loss": 1.15,
+      "step": 2889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430249612070006e-05,
+      "loss": 1.1172,
+      "step": 2890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424290205169007e-05,
+      "loss": 1.2088,
+      "step": 2891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418331003401519e-05,
+      "loss": 1.1243,
+      "step": 2892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.412372008890886e-05,
+      "loss": 1.1578,
+      "step": 2893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.406413223760391e-05,
+      "loss": 1.105,
+      "step": 2894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40045465013322e-05,
+      "loss": 1.1036,
+      "step": 2895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.394496290132503e-05,
+      "loss": 1.2113,
+      "step": 2896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388538145881289e-05,
+      "loss": 1.1959,
+      "step": 2897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382580219502547e-05,
+      "loss": 1.119,
+      "step": 2898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.376622513119173e-05,
+      "loss": 1.1749,
+      "step": 2899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370665028853976e-05,
+      "loss": 1.0931,
+      "step": 2900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364707768829694e-05,
+      "loss": 1.0954,
+      "step": 2901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358750735168983e-05,
+      "loss": 1.2064,
+      "step": 2902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35279392999442e-05,
+      "loss": 1.2186,
+      "step": 2903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.346837355428494e-05,
+      "loss": 1.0256,
+      "step": 2904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340881013593625e-05,
+      "loss": 1.1713,
+      "step": 2905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334924906612131e-05,
+      "loss": 1.0995,
+      "step": 2906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.328969036606262e-05,
+      "loss": 1.1875,
+      "step": 2907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323013405698178e-05,
+      "loss": 1.1381,
+      "step": 2908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317058016009954e-05,
+      "loss": 1.0689,
+      "step": 2909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311102869663584e-05,
+      "loss": 1.2151,
+      "step": 2910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.305147968780962e-05,
+      "loss": 1.166,
+      "step": 2911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299193315483906e-05,
+      "loss": 1.2019,
+      "step": 2912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293238911894144e-05,
+      "loss": 1.2719,
+      "step": 2913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287284760133313e-05,
+      "loss": 1.0417,
+      "step": 2914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281330862322963e-05,
+      "loss": 1.1438,
+      "step": 2915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.275377220584554e-05,
+      "loss": 1.2057,
+      "step": 2916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.269423837039446e-05,
+      "loss": 1.2664,
+      "step": 2917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263470713808915e-05,
+      "loss": 1.1658,
+      "step": 2918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257517853014145e-05,
+      "loss": 1.2042,
+      "step": 2919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.251565256776223e-05,
+      "loss": 1.2215,
+      "step": 2920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.245612927216145e-05,
+      "loss": 1.1456,
+      "step": 2921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239660866454809e-05,
+      "loss": 1.0487,
+      "step": 2922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233709076613013e-05,
+      "loss": 1.2356,
+      "step": 2923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227757559811469e-05,
+      "loss": 1.1757,
+      "step": 2924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221806318170783e-05,
+      "loss": 1.1628,
+      "step": 2925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21585535381147e-05,
+      "loss": 1.1013,
+      "step": 2926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209904668853943e-05,
+      "loss": 1.076,
+      "step": 2927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20395426541851e-05,
+      "loss": 1.1132,
+      "step": 2928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198004145625386e-05,
+      "loss": 1.1724,
+      "step": 2929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192054311594682e-05,
+      "loss": 1.1824,
+      "step": 2930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.186104765446408e-05,
+      "loss": 1.1273,
+      "step": 2931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.180155509300473e-05,
+      "loss": 1.2039,
+      "step": 2932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.174206545276677e-05,
+      "loss": 1.3616,
+      "step": 2933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168257875494724e-05,
+      "loss": 1.107,
+      "step": 2934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162309502074206e-05,
+      "loss": 1.1277,
+      "step": 2935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156361427134611e-05,
+      "loss": 1.1397,
+      "step": 2936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150413652795325e-05,
+      "loss": 1.1044,
+      "step": 2937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144466181175622e-05,
+      "loss": 0.8782,
+      "step": 2938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.138519014394671e-05,
+      "loss": 1.1193,
+      "step": 2939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132572154571532e-05,
+      "loss": 1.0753,
+      "step": 2940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126625603825152e-05,
+      "loss": 1.158,
+      "step": 2941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120679364274372e-05,
+      "loss": 1.1534,
+      "step": 2942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114733438037922e-05,
+      "loss": 1.0979,
+      "step": 2943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.108787827234418e-05,
+      "loss": 1.1033,
+      "step": 2944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102842533982365e-05,
+      "loss": 1.1068,
+      "step": 2945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096897560400158e-05,
+      "loss": 1.1755,
+      "step": 2946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09095290860607e-05,
+      "loss": 1.1487,
+      "step": 2947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085008580718266e-05,
+      "loss": 1.1299,
+      "step": 2948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079064578854795e-05,
+      "loss": 1.0345,
+      "step": 2949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.073120905133588e-05,
+      "loss": 1.1673,
+      "step": 2950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067177561672456e-05,
+      "loss": 1.1317,
+      "step": 2951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061234550589103e-05,
+      "loss": 1.3414,
+      "step": 2952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055291874001103e-05,
+      "loss": 1.126,
+      "step": 2953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049349534025917e-05,
+      "loss": 1.0681,
+      "step": 2954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043407532780882e-05,
+      "loss": 1.065,
+      "step": 2955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037465872383218e-05,
+      "loss": 1.1425,
+      "step": 2956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.031524554950023e-05,
+      "loss": 1.2085,
+      "step": 2957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025583582598273e-05,
+      "loss": 0.9626,
+      "step": 2958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.019642957444819e-05,
+      "loss": 0.9804,
+      "step": 2959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013702681606389e-05,
+      "loss": 1.0576,
+      "step": 2960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007762757199587e-05,
+      "loss": 1.1011,
+      "step": 2961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.001823186340892e-05,
+      "loss": 1.0638,
+      "step": 2962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995883971146656e-05,
+      "loss": 1.1404,
+      "step": 2963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989945113733108e-05,
+      "loss": 1.1637,
+      "step": 2964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984006616216346e-05,
+      "loss": 1.0596,
+      "step": 2965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978068480712335e-05,
+      "loss": 1.2259,
+      "step": 2966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97213070933692e-05,
+      "loss": 1.1142,
+      "step": 2967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966193304205813e-05,
+      "loss": 1.1162,
+      "step": 2968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960256267434596e-05,
+      "loss": 0.9594,
+      "step": 2969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954319601138718e-05,
+      "loss": 1.0133,
+      "step": 2970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948383307433498e-05,
+      "loss": 1.3156,
+      "step": 2971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.942447388434115e-05,
+      "loss": 1.0209,
+      "step": 2972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936511846255627e-05,
+      "loss": 1.0959,
+      "step": 2973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930576683012949e-05,
+      "loss": 1.1188,
+      "step": 2974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924641900820864e-05,
+      "loss": 1.1334,
+      "step": 2975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918707501794021e-05,
+      "loss": 0.9522,
+      "step": 2976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912773488046925e-05,
+      "loss": 1.1984,
+      "step": 2977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.906839861693949e-05,
+      "loss": 1.18,
+      "step": 2978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90090662484933e-05,
+      "loss": 1.2392,
+      "step": 2979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.894973779627163e-05,
+      "loss": 1.0964,
+      "step": 2980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889041328141405e-05,
+      "loss": 1.1851,
+      "step": 2981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883109272505877e-05,
+      "loss": 1.2173,
+      "step": 2982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877177614834242e-05,
+      "loss": 1.1561,
+      "step": 2983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871246357240041e-05,
+      "loss": 1.1446,
+      "step": 2984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865315501836664e-05,
+      "loss": 1.1767,
+      "step": 2985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859385050737356e-05,
+      "loss": 1.1216,
+      "step": 2986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853455006055229e-05,
+      "loss": 1.1508,
+      "step": 2987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847525369903229e-05,
+      "loss": 1.206,
+      "step": 2988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841596144394174e-05,
+      "loss": 1.1434,
+      "step": 2989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.835667331640731e-05,
+      "loss": 1.0858,
+      "step": 2990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82973893375542e-05,
+      "loss": 0.9895,
+      "step": 2991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.823810952850612e-05,
+      "loss": 1.1926,
+      "step": 2992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817883391038533e-05,
+      "loss": 1.0319,
+      "step": 2993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811956250431253e-05,
+      "loss": 1.1582,
+      "step": 2994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806029533140693e-05,
+      "loss": 1.1574,
+      "step": 2995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800103241278634e-05,
+      "loss": 1.1944,
+      "step": 2996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79417737695669e-05,
+      "loss": 1.0363,
+      "step": 2997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788251942286337e-05,
+      "loss": 1.1398,
+      "step": 2998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.782326939378883e-05,
+      "loss": 1.053,
+      "step": 2999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776402370345494e-05,
+      "loss": 1.0548,
+      "step": 3000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770478237297176e-05,
+      "loss": 1.3369,
+      "step": 3001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.764554542344783e-05,
+      "loss": 1.1963,
+      "step": 3002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75863128759901e-05,
+      "loss": 1.2434,
+      "step": 3003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752708475170399e-05,
+      "loss": 1.0964,
+      "step": 3004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746786107169327e-05,
+      "loss": 1.255,
+      "step": 3005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740864185706016e-05,
+      "loss": 1.106,
+      "step": 3006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734942712890534e-05,
+      "loss": 1.087,
+      "step": 3007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.729021690832785e-05,
+      "loss": 1.128,
+      "step": 3008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723101121642516e-05,
+      "loss": 0.9239,
+      "step": 3009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717181007429299e-05,
+      "loss": 1.091,
+      "step": 3010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71126135030256e-05,
+      "loss": 1.1511,
+      "step": 3011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705342152371557e-05,
+      "loss": 1.1565,
+      "step": 3012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699423415745383e-05,
+      "loss": 1.1547,
+      "step": 3013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693505142532966e-05,
+      "loss": 1.1623,
+      "step": 3014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687587334843078e-05,
+      "loss": 1.2677,
+      "step": 3015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681669994784303e-05,
+      "loss": 1.141,
+      "step": 3016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67575312446508e-05,
+      "loss": 1.0403,
+      "step": 3017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669836725993674e-05,
+      "loss": 1.1266,
+      "step": 3018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66392080147818e-05,
+      "loss": 1.1954,
+      "step": 3019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65800535302653e-05,
+      "loss": 1.1671,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652090382746474e-05,
+      "loss": 1.1481,
+      "step": 3021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6461758927456e-05,
+      "loss": 1.0806,
+      "step": 3022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640261885131328e-05,
+      "loss": 1.3917,
+      "step": 3023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634348362010902e-05,
+      "loss": 1.1322,
+      "step": 3024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628435325491394e-05,
+      "loss": 1.3397,
+      "step": 3025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.622522777679699e-05,
+      "loss": 1.2579,
+      "step": 3026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61661072068254e-05,
+      "loss": 1.1412,
+      "step": 3027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61069915660647e-05,
+      "loss": 1.2435,
+      "step": 3028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60478808755786e-05,
+      "loss": 1.1183,
+      "step": 3029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598877515642911e-05,
+      "loss": 1.268,
+      "step": 3030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592967442967644e-05,
+      "loss": 1.1197,
+      "step": 3031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587057871637891e-05,
+      "loss": 1.1447,
+      "step": 3032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.581148803759324e-05,
+      "loss": 1.1857,
+      "step": 3033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575240241437427e-05,
+      "loss": 1.1563,
+      "step": 3034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5693321867775e-05,
+      "loss": 1.1515,
+      "step": 3035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563424641884674e-05,
+      "loss": 1.0991,
+      "step": 3036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55751760886388e-05,
+      "loss": 1.094,
+      "step": 3037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551611089819883e-05,
+      "loss": 1.0166,
+      "step": 3038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545705086857256e-05,
+      "loss": 1.0674,
+      "step": 3039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539799602080394e-05,
+      "loss": 1.1136,
+      "step": 3040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533894637593503e-05,
+      "loss": 1.0936,
+      "step": 3041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527990195500609e-05,
+      "loss": 1.1186,
+      "step": 3042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522086277905541e-05,
+      "loss": 1.09,
+      "step": 3043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516182886911952e-05,
+      "loss": 1.1297,
+      "step": 3044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510280024623301e-05,
+      "loss": 0.9226,
+      "step": 3045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504377693142865e-05,
+      "loss": 1.2127,
+      "step": 3046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498475894573728e-05,
+      "loss": 1.2971,
+      "step": 3047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.492574631018777e-05,
+      "loss": 1.1164,
+      "step": 3048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48667390458072e-05,
+      "loss": 1.2226,
+      "step": 3049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480773717362069e-05,
+      "loss": 1.0534,
+      "step": 3050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474874071465144e-05,
+      "loss": 1.0608,
+      "step": 3051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46897496899207e-05,
+      "loss": 1.2981,
+      "step": 3052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463076412044789e-05,
+      "loss": 1.0084,
+      "step": 3053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.457178402725025e-05,
+      "loss": 1.2863,
+      "step": 3054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451280943134332e-05,
+      "loss": 1.125,
+      "step": 3055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445384035374055e-05,
+      "loss": 1.0367,
+      "step": 3056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439487681545345e-05,
+      "loss": 1.2817,
+      "step": 3057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433591883749162e-05,
+      "loss": 1.0949,
+      "step": 3058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.427696644086251e-05,
+      "loss": 1.1196,
+      "step": 3059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421801964657175e-05,
+      "loss": 1.114,
+      "step": 3060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415907847562292e-05,
+      "loss": 1.1472,
+      "step": 3061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410014294901757e-05,
+      "loss": 1.1202,
+      "step": 3062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404121308775525e-05,
+      "loss": 1.1515,
+      "step": 3063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398228891283358e-05,
+      "loss": 1.1511,
+      "step": 3064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.392337044524797e-05,
+      "loss": 1.0768,
+      "step": 3065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.386445770599192e-05,
+      "loss": 1.0511,
+      "step": 3066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380555071605689e-05,
+      "loss": 1.2184,
+      "step": 3067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374664949643228e-05,
+      "loss": 1.1507,
+      "step": 3068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.368775406810543e-05,
+      "loss": 1.0192,
+      "step": 3069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.362886445206159e-05,
+      "loss": 1.0429,
+      "step": 3070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356998066928394e-05,
+      "loss": 1.2672,
+      "step": 3071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351110274075363e-05,
+      "loss": 1.0522,
+      "step": 3072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345223068744968e-05,
+      "loss": 1.0327,
+      "step": 3073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339336453034905e-05,
+      "loss": 1.1057,
+      "step": 3074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33345042904266e-05,
+      "loss": 1.0988,
+      "step": 3075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327564998865503e-05,
+      "loss": 1.1436,
+      "step": 3076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321680164600493e-05,
+      "loss": 1.0987,
+      "step": 3077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.315795928344482e-05,
+      "loss": 1.1273,
+      "step": 3078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30991229219411e-05,
+      "loss": 1.1621,
+      "step": 3079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304029258245795e-05,
+      "loss": 1.2133,
+      "step": 3080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298146828595748e-05,
+      "loss": 1.0005,
+      "step": 3081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.292265005339958e-05,
+      "loss": 1.1018,
+      "step": 3082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286383790574202e-05,
+      "loss": 1.158,
+      "step": 3083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280503186394042e-05,
+      "loss": 1.2132,
+      "step": 3084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274623194894818e-05,
+      "loss": 1.1722,
+      "step": 3085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268743818171657e-05,
+      "loss": 1.1002,
+      "step": 3086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26286505831946e-05,
+      "loss": 1.1078,
+      "step": 3087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256986917432914e-05,
+      "loss": 1.0197,
+      "step": 3088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251109397606483e-05,
+      "loss": 1.2089,
+      "step": 3089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245232500934408e-05,
+      "loss": 1.2076,
+      "step": 3090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239356229510715e-05,
+      "loss": 0.9508,
+      "step": 3091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.233480585429196e-05,
+      "loss": 1.1723,
+      "step": 3092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227605570783431e-05,
+      "loss": 0.9848,
+      "step": 3093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22173118766677e-05,
+      "loss": 1.0922,
+      "step": 3094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215857438172334e-05,
+      "loss": 1.2258,
+      "step": 3095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209984324393026e-05,
+      "loss": 1.1184,
+      "step": 3096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20411184842152e-05,
+      "loss": 1.1338,
+      "step": 3097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198240012350261e-05,
+      "loss": 1.187,
+      "step": 3098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.192368818271465e-05,
+      "loss": 1.1214,
+      "step": 3099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186498268277126e-05,
+      "loss": 1.1113,
+      "step": 3100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180628364458998e-05,
+      "loss": 1.1226,
+      "step": 3101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174759108908615e-05,
+      "loss": 1.0049,
+      "step": 3102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168890503717271e-05,
+      "loss": 1.1606,
+      "step": 3103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163022550976035e-05,
+      "loss": 1.0784,
+      "step": 3104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157155252775742e-05,
+      "loss": 1.1108,
+      "step": 3105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.151288611206992e-05,
+      "loss": 1.0967,
+      "step": 3106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145422628360153e-05,
+      "loss": 1.0757,
+      "step": 3107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139557306325358e-05,
+      "loss": 1.107,
+      "step": 3108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1336926471925e-05,
+      "loss": 1.2708,
+      "step": 3109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127828653051243e-05,
+      "loss": 1.1873,
+      "step": 3110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121965325991008e-05,
+      "loss": 1.1988,
+      "step": 3111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116102668100986e-05,
+      "loss": 1.2832,
+      "step": 3112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.110240681470123e-05,
+      "loss": 1.1698,
+      "step": 3113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104379368187122e-05,
+      "loss": 1.0821,
+      "step": 3114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098518730340456e-05,
+      "loss": 1.1157,
+      "step": 3115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092658770018351e-05,
+      "loss": 1.1829,
+      "step": 3116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086799489308797e-05,
+      "loss": 1.1656,
+      "step": 3117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080940890299536e-05,
+      "loss": 1.0372,
+      "step": 3118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07508297507807e-05,
+      "loss": 1.2365,
+      "step": 3119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.069225745731654e-05,
+      "loss": 1.2585,
+      "step": 3120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063369204347302e-05,
+      "loss": 1.1302,
+      "step": 3121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057513353011785e-05,
+      "loss": 1.1863,
+      "step": 3122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051658193811623e-05,
+      "loss": 1.1274,
+      "step": 3123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045803728833097e-05,
+      "loss": 1.1456,
+      "step": 3124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039949960162227e-05,
+      "loss": 1.0766,
+      "step": 3125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034096889884797e-05,
+      "loss": 1.1282,
+      "step": 3126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028244520086337e-05,
+      "loss": 1.1553,
+      "step": 3127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022392852852132e-05,
+      "loss": 1.1938,
+      "step": 3128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01654189026721e-05,
+      "loss": 1.1029,
+      "step": 3129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01069163441636e-05,
+      "loss": 1.0612,
+      "step": 3130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.004842087384096e-05,
+      "loss": 1.1839,
+      "step": 3131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998993251254705e-05,
+      "loss": 1.184,
+      "step": 3132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993145128112205e-05,
+      "loss": 1.148,
+      "step": 3133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987297720040365e-05,
+      "loss": 1.1483,
+      "step": 3134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981451029122706e-05,
+      "loss": 1.1227,
+      "step": 3135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975605057442476e-05,
+      "loss": 1.201,
+      "step": 3136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96975980708268e-05,
+      "loss": 1.1106,
+      "step": 3137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963915280126066e-05,
+      "loss": 1.1145,
+      "step": 3138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958071478655118e-05,
+      "loss": 1.1158,
+      "step": 3139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952228404752067e-05,
+      "loss": 1.1921,
+      "step": 3140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946386060498886e-05,
+      "loss": 1.0459,
+      "step": 3141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940544447977276e-05,
+      "loss": 0.9492,
+      "step": 3142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.934703569268691e-05,
+      "loss": 1.2154,
+      "step": 3143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928863426454318e-05,
+      "loss": 0.9868,
+      "step": 3144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923024021615082e-05,
+      "loss": 1.0932,
+      "step": 3145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917185356831648e-05,
+      "loss": 1.1503,
+      "step": 3146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91134743418441e-05,
+      "loss": 1.0997,
+      "step": 3147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905510255753501e-05,
+      "loss": 1.0977,
+      "step": 3148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899673823618793e-05,
+      "loss": 1.2557,
+      "step": 3149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893838139859891e-05,
+      "loss": 1.2678,
+      "step": 3150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.888003206556126e-05,
+      "loss": 1.0974,
+      "step": 3151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882169025786575e-05,
+      "loss": 1.2378,
+      "step": 3152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87633559963003e-05,
+      "loss": 1.0408,
+      "step": 3153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870502930165026e-05,
+      "loss": 0.9884,
+      "step": 3154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864671019469826e-05,
+      "loss": 1.0159,
+      "step": 3155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858839869622422e-05,
+      "loss": 1.1755,
+      "step": 3156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853009482700539e-05,
+      "loss": 1.0817,
+      "step": 3157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847179860781616e-05,
+      "loss": 1.1309,
+      "step": 3158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841351005942836e-05,
+      "loss": 1.0647,
+      "step": 3159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8355229202611e-05,
+      "loss": 1.2224,
+      "step": 3160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829695605813039e-05,
+      "loss": 1.0081,
+      "step": 3161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823869064675007e-05,
+      "loss": 1.1445,
+      "step": 3162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818043298923086e-05,
+      "loss": 1.1952,
+      "step": 3163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81221831063307e-05,
+      "loss": 1.0846,
+      "step": 3164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806394101880489e-05,
+      "loss": 1.1801,
+      "step": 3165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80057067474059e-05,
+      "loss": 1.0914,
+      "step": 3166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794748031288342e-05,
+      "loss": 1.0859,
+      "step": 3167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.788926173598441e-05,
+      "loss": 1.0192,
+      "step": 3168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783105103745286e-05,
+      "loss": 1.1546,
+      "step": 3169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777284823803012e-05,
+      "loss": 1.1766,
+      "step": 3170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771465335845467e-05,
+      "loss": 1.0625,
+      "step": 3171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.765646641946215e-05,
+      "loss": 1.0765,
+      "step": 3172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759828744178539e-05,
+      "loss": 1.1468,
+      "step": 3173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.754011644615444e-05,
+      "loss": 1.0343,
+      "step": 3174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748195345329634e-05,
+      "loss": 1.1511,
+      "step": 3175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742379848393543e-05,
+      "loss": 1.0864,
+      "step": 3176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736565155879314e-05,
+      "loss": 1.0561,
+      "step": 3177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730751269858806e-05,
+      "loss": 0.9745,
+      "step": 3178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72493819240359e-05,
+      "loss": 1.1102,
+      "step": 3179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719125925584941e-05,
+      "loss": 1.0523,
+      "step": 3180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713314471473855e-05,
+      "loss": 1.1171,
+      "step": 3181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707503832141034e-05,
+      "loss": 1.1898,
+      "step": 3182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701694009656892e-05,
+      "loss": 1.2348,
+      "step": 3183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695885006091552e-05,
+      "loss": 1.1847,
+      "step": 3184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690076823514844e-05,
+      "loss": 1.0157,
+      "step": 3185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6842694639963e-05,
+      "loss": 1.2321,
+      "step": 3186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678462929605167e-05,
+      "loss": 1.1772,
+      "step": 3187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672657222410395e-05,
+      "loss": 1.0534,
+      "step": 3188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666852344480641e-05,
+      "loss": 1.1113,
+      "step": 3189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.661048297884266e-05,
+      "loss": 1.1037,
+      "step": 3190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655245084689326e-05,
+      "loss": 1.1486,
+      "step": 3191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649442706963594e-05,
+      "loss": 1.0509,
+      "step": 3192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643641166774538e-05,
+      "loss": 1.1276,
+      "step": 3193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637840466189326e-05,
+      "loss": 1.2689,
+      "step": 3194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632040607274832e-05,
+      "loss": 1.157,
+      "step": 3195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626241592097631e-05,
+      "loss": 1.1377,
+      "step": 3196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.620443422723985e-05,
+      "loss": 1.1501,
+      "step": 3197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614646101219868e-05,
+      "loss": 1.199,
+      "step": 3198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608849629650947e-05,
+      "loss": 1.1213,
+      "step": 3199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603054010082588e-05,
+      "loss": 1.2034,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597259244579853e-05,
+      "loss": 1.1787,
+      "step": 3201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591465335207492e-05,
+      "loss": 1.1466,
+      "step": 3202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585672284029962e-05,
+      "loss": 1.1366,
+      "step": 3203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579880093111407e-05,
+      "loss": 1.057,
+      "step": 3204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574088764515665e-05,
+      "loss": 1.1525,
+      "step": 3205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568298300306271e-05,
+      "loss": 1.1291,
+      "step": 3206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562508702546452e-05,
+      "loss": 1.0518,
+      "step": 3207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556719973299115e-05,
+      "loss": 1.113,
+      "step": 3208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550932114626869e-05,
+      "loss": 1.2159,
+      "step": 3209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54514512859201e-05,
+      "loss": 1.1312,
+      "step": 3210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539359017256522e-05,
+      "loss": 1.0772,
+      "step": 3211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533573782682084e-05,
+      "loss": 1.2186,
+      "step": 3212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.527789426930046e-05,
+      "loss": 1.104,
+      "step": 3213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522005952061462e-05,
+      "loss": 1.0655,
+      "step": 3214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516223360137065e-05,
+      "loss": 1.0864,
+      "step": 3215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510441653217272e-05,
+      "loss": 1.215,
+      "step": 3216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504660833362187e-05,
+      "loss": 1.0473,
+      "step": 3217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498880902631604e-05,
+      "loss": 1.0355,
+      "step": 3218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493101863084985e-05,
+      "loss": 1.1933,
+      "step": 3219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487323716781485e-05,
+      "loss": 1.1756,
+      "step": 3220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48154646577994e-05,
+      "loss": 1.1431,
+      "step": 3221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475770112138866e-05,
+      "loss": 1.0798,
+      "step": 3222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469994657916463e-05,
+      "loss": 1.0109,
+      "step": 3223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464220105170603e-05,
+      "loss": 1.0241,
+      "step": 3224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458446455958839e-05,
+      "loss": 1.154,
+      "step": 3225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.452673712338404e-05,
+      "loss": 1.0658,
+      "step": 3226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44690187636621e-05,
+      "loss": 1.1563,
+      "step": 3227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441130950098841e-05,
+      "loss": 1.1024,
+      "step": 3228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435360935592563e-05,
+      "loss": 0.9866,
+      "step": 3229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429591834903314e-05,
+      "loss": 1.0766,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423823650086699e-05,
+      "loss": 1.0853,
+      "step": 3231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.418056383198007e-05,
+      "loss": 1.1761,
+      "step": 3232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.412290036292197e-05,
+      "loss": 1.0165,
+      "step": 3233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406524611423903e-05,
+      "loss": 1.2967,
+      "step": 3234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40076011064742e-05,
+      "loss": 1.1091,
+      "step": 3235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394996536016729e-05,
+      "loss": 1.1849,
+      "step": 3236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389233889585465e-05,
+      "loss": 1.1426,
+      "step": 3237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383472173406944e-05,
+      "loss": 1.1224,
+      "step": 3238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377711389534145e-05,
+      "loss": 1.0081,
+      "step": 3239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.371951540019721e-05,
+      "loss": 1.046,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366192626915981e-05,
+      "loss": 1.1591,
+      "step": 3241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360434652274913e-05,
+      "loss": 1.0869,
+      "step": 3242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35467761814816e-05,
+      "loss": 1.2311,
+      "step": 3243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.348921526587034e-05,
+      "loss": 1.1525,
+      "step": 3244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.343166379642517e-05,
+      "loss": 1.1103,
+      "step": 3245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337412179365243e-05,
+      "loss": 1.0426,
+      "step": 3246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.331658927805516e-05,
+      "loss": 1.2594,
+      "step": 3247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325906627013304e-05,
+      "loss": 1.0406,
+      "step": 3248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32015527903823e-05,
+      "loss": 0.9441,
+      "step": 3249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314404885929578e-05,
+      "loss": 1.1859,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3086554497363e-05,
+      "loss": 0.9717,
+      "step": 3251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302906972506995e-05,
+      "loss": 1.1589,
+      "step": 3252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29715945628993e-05,
+      "loss": 1.1238,
+      "step": 3253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291412903133026e-05,
+      "loss": 1.1324,
+      "step": 3254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.28566731508386e-05,
+      "loss": 1.1963,
+      "step": 3255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279922694189666e-05,
+      "loss": 1.1863,
+      "step": 3256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27417904249733e-05,
+      "loss": 1.1373,
+      "step": 3257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.268436362053403e-05,
+      "loss": 1.1364,
+      "step": 3258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.262694654904077e-05,
+      "loss": 1.0194,
+      "step": 3259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256953923095209e-05,
+      "loss": 1.1818,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251214168672298e-05,
+      "loss": 1.0542,
+      "step": 3261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245475393680499e-05,
+      "loss": 1.1721,
+      "step": 3262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239737600164618e-05,
+      "loss": 1.0847,
+      "step": 3263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234000790169114e-05,
+      "loss": 1.0345,
+      "step": 3264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228264965738093e-05,
+      "loss": 1.0901,
+      "step": 3265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222530128915313e-05,
+      "loss": 1.1637,
+      "step": 3266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216796281744172e-05,
+      "loss": 1.1334,
+      "step": 3267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211063426267721e-05,
+      "loss": 1.1499,
+      "step": 3268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.205331564528658e-05,
+      "loss": 1.0228,
+      "step": 3269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199600698569327e-05,
+      "loss": 1.1231,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193870830431715e-05,
+      "loss": 1.1008,
+      "step": 3271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188141962157461e-05,
+      "loss": 1.1347,
+      "step": 3272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.18241409578783e-05,
+      "loss": 1.1925,
+      "step": 3273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17668723336375e-05,
+      "loss": 1.026,
+      "step": 3274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170961376925779e-05,
+      "loss": 1.0538,
+      "step": 3275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.16523652851412e-05,
+      "loss": 1.1736,
+      "step": 3276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.159512690168622e-05,
+      "loss": 1.2162,
+      "step": 3277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153789863928769e-05,
+      "loss": 1.1371,
+      "step": 3278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148068051833676e-05,
+      "loss": 1.2695,
+      "step": 3279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142347255922112e-05,
+      "loss": 1.2569,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.136627478232476e-05,
+      "loss": 1.2343,
+      "step": 3281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130908720802805e-05,
+      "loss": 1.1145,
+      "step": 3282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.125190985670777e-05,
+      "loss": 1.086,
+      "step": 3283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119474274873693e-05,
+      "loss": 1.1666,
+      "step": 3284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113758590448502e-05,
+      "loss": 1.1027,
+      "step": 3285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.108043934431785e-05,
+      "loss": 1.215,
+      "step": 3286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102330308859753e-05,
+      "loss": 1.0828,
+      "step": 3287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.09661771576825e-05,
+      "loss": 0.9659,
+      "step": 3288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090906157192758e-05,
+      "loss": 1.1638,
+      "step": 3289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085195635168377e-05,
+      "loss": 1.2037,
+      "step": 3290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079486151729854e-05,
+      "loss": 1.0959,
+      "step": 3291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.073777708911556e-05,
+      "loss": 1.2897,
+      "step": 3292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.068070308747479e-05,
+      "loss": 1.292,
+      "step": 3293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.062363953271259e-05,
+      "loss": 1.1087,
+      "step": 3294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056658644516138e-05,
+      "loss": 1.1715,
+      "step": 3295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.050954384515005e-05,
+      "loss": 0.9097,
+      "step": 3296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.04525117530037e-05,
+      "loss": 1.2046,
+      "step": 3297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03954901890436e-05,
+      "loss": 1.1817,
+      "step": 3298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033847917358741e-05,
+      "loss": 1.1052,
+      "step": 3299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028147872694897e-05,
+      "loss": 1.1743,
+      "step": 3300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022448886943824e-05,
+      "loss": 1.1383,
+      "step": 3301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016750962136158e-05,
+      "loss": 1.1679,
+      "step": 3302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011054100302148e-05,
+      "loss": 1.2427,
+      "step": 3303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005358303471667e-05,
+      "loss": 1.127,
+      "step": 3304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999663573674211e-05,
+      "loss": 1.1007,
+      "step": 3305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993969912938887e-05,
+      "loss": 1.1633,
+      "step": 3306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988277323294426e-05,
+      "loss": 1.1544,
+      "step": 3307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982585806769181e-05,
+      "loss": 1.132,
+      "step": 3308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97689536539112e-05,
+      "loss": 0.9402,
+      "step": 3309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971206001187824e-05,
+      "loss": 1.0496,
+      "step": 3310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965517716186502e-05,
+      "loss": 1.0992,
+      "step": 3311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.95983051241396e-05,
+      "loss": 1.125,
+      "step": 3312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954144391896631e-05,
+      "loss": 1.1088,
+      "step": 3313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.948459356660563e-05,
+      "loss": 1.1083,
+      "step": 3314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942775408731413e-05,
+      "loss": 1.0246,
+      "step": 3315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.937092550134454e-05,
+      "loss": 1.1468,
+      "step": 3316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931410782894562e-05,
+      "loss": 1.1172,
+      "step": 3317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.925730109036236e-05,
+      "loss": 1.1054,
+      "step": 3318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920050530583578e-05,
+      "loss": 1.273,
+      "step": 3319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914372049560304e-05,
+      "loss": 1.2008,
+      "step": 3320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908694667989735e-05,
+      "loss": 1.0215,
+      "step": 3321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903018387894805e-05,
+      "loss": 1.1444,
+      "step": 3322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897343211298047e-05,
+      "loss": 1.1119,
+      "step": 3323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89166914022161e-05,
+      "loss": 1.1033,
+      "step": 3324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885996176687244e-05,
+      "loss": 1.1069,
+      "step": 3325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.880324322716306e-05,
+      "loss": 1.1489,
+      "step": 3326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874653580329764e-05,
+      "loss": 1.065,
+      "step": 3327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868983951548171e-05,
+      "loss": 1.1035,
+      "step": 3328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.863315438391705e-05,
+      "loss": 1.0294,
+      "step": 3329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857648042880133e-05,
+      "loss": 1.154,
+      "step": 3330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85198176703283e-05,
+      "loss": 1.1528,
+      "step": 3331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846316612868765e-05,
+      "loss": 1.1558,
+      "step": 3332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.840652582406525e-05,
+      "loss": 1.0721,
+      "step": 3333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83498967766427e-05,
+      "loss": 1.0806,
+      "step": 3334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829327900659776e-05,
+      "loss": 0.9791,
+      "step": 3335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823667253410417e-05,
+      "loss": 1.0732,
+      "step": 3336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.81800773793316e-05,
+      "loss": 1.1795,
+      "step": 3337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812349356244574e-05,
+      "loss": 0.9927,
+      "step": 3338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.806692110360812e-05,
+      "loss": 1.1296,
+      "step": 3339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801036002297634e-05,
+      "loss": 1.0663,
+      "step": 3340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.795381034070394e-05,
+      "loss": 1.1162,
+      "step": 3341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789727207694033e-05,
+      "loss": 1.0895,
+      "step": 3342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78407452518309e-05,
+      "loss": 1.0273,
+      "step": 3343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.778422988551701e-05,
+      "loss": 1.0596,
+      "step": 3344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772772599813577e-05,
+      "loss": 1.0937,
+      "step": 3345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767123360982038e-05,
+      "loss": 1.0752,
+      "step": 3346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761475274069986e-05,
+      "loss": 1.0884,
+      "step": 3347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755828341089917e-05,
+      "loss": 1.1279,
+      "step": 3348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750182564053911e-05,
+      "loss": 1.0794,
+      "step": 3349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744537944973635e-05,
+      "loss": 1.2396,
+      "step": 3350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738894485860348e-05,
+      "loss": 1.0755,
+      "step": 3351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733252188724892e-05,
+      "loss": 1.0125,
+      "step": 3352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.727611055577703e-05,
+      "loss": 1.123,
+      "step": 3353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.721971088428789e-05,
+      "loss": 1.0441,
+      "step": 3354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716332289287759e-05,
+      "loss": 1.1861,
+      "step": 3355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710694660163787e-05,
+      "loss": 1.1581,
+      "step": 3356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705058203065644e-05,
+      "loss": 1.1241,
+      "step": 3357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699422920001677e-05,
+      "loss": 1.1545,
+      "step": 3358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693788812979819e-05,
+      "loss": 1.1591,
+      "step": 3359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688155884007586e-05,
+      "loss": 1.055,
+      "step": 3360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682524135092058e-05,
+      "loss": 1.221,
+      "step": 3361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676893568239915e-05,
+      "loss": 1.0342,
+      "step": 3362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.671264185457407e-05,
+      "loss": 1.1931,
+      "step": 3363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.665635988750358e-05,
+      "loss": 0.9978,
+      "step": 3364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66000898012418e-05,
+      "loss": 1.0563,
+      "step": 3365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654383161583855e-05,
+      "loss": 1.0092,
+      "step": 3366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648758535133934e-05,
+      "loss": 1.1265,
+      "step": 3367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643135102778555e-05,
+      "loss": 1.1786,
+      "step": 3368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637512866521427e-05,
+      "loss": 1.1273,
+      "step": 3369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631891828365833e-05,
+      "loss": 1.1136,
+      "step": 3370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626271990314626e-05,
+      "loss": 1.0124,
+      "step": 3371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620653354370235e-05,
+      "loss": 1.2785,
+      "step": 3372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615035922534657e-05,
+      "loss": 1.1804,
+      "step": 3373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609419696809462e-05,
+      "loss": 1.1054,
+      "step": 3374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603804679195792e-05,
+      "loss": 1.1064,
+      "step": 3375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598190871694356e-05,
+      "loss": 1.2409,
+      "step": 3376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.592578276305436e-05,
+      "loss": 0.9732,
+      "step": 3377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586966895028876e-05,
+      "loss": 1.0801,
+      "step": 3378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581356729864087e-05,
+      "loss": 1.1631,
+      "step": 3379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575747782810053e-05,
+      "loss": 1.2809,
+      "step": 3380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570140055865321e-05,
+      "loss": 1.1101,
+      "step": 3381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564533551028005e-05,
+      "loss": 1.2128,
+      "step": 3382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558928270295779e-05,
+      "loss": 1.1091,
+      "step": 3383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553324215665883e-05,
+      "loss": 1.2042,
+      "step": 3384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.547721389135121e-05,
+      "loss": 1.1536,
+      "step": 3385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.542119792699861e-05,
+      "loss": 1.185,
+      "step": 3386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.536519428356029e-05,
+      "loss": 1.2076,
+      "step": 3387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.530920298099115e-05,
+      "loss": 1.1469,
+      "step": 3388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.525322403924166e-05,
+      "loss": 1.0598,
+      "step": 3389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519725747825795e-05,
+      "loss": 1.0364,
+      "step": 3390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514130331798162e-05,
+      "loss": 1.0272,
+      "step": 3391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508536157834996e-05,
+      "loss": 1.0343,
+      "step": 3392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502943227929586e-05,
+      "loss": 1.0356,
+      "step": 3393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497351544074761e-05,
+      "loss": 1.1444,
+      "step": 3394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.491761108262923e-05,
+      "loss": 1.1718,
+      "step": 3395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486171922486024e-05,
+      "loss": 1.0802,
+      "step": 3396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480583988735565e-05,
+      "loss": 1.0561,
+      "step": 3397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47499730900261e-05,
+      "loss": 0.9797,
+      "step": 3398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46941188527777e-05,
+      "loss": 1.1883,
+      "step": 3399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46382771955121e-05,
+      "loss": 1.0437,
+      "step": 3400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458244813812646e-05,
+      "loss": 1.0546,
+      "step": 3401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45266317005135e-05,
+      "loss": 1.1714,
+      "step": 3402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447082790256134e-05,
+      "loss": 1.0804,
+      "step": 3403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44150367641537e-05,
+      "loss": 1.0601,
+      "step": 3404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.435925830516973e-05,
+      "loss": 1.184,
+      "step": 3405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43034925454841e-05,
+      "loss": 1.0652,
+      "step": 3406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424773950496692e-05,
+      "loss": 1.2654,
+      "step": 3407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419199920348381e-05,
+      "loss": 1.165,
+      "step": 3408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413627166089579e-05,
+      "loss": 1.0988,
+      "step": 3409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40805568970594e-05,
+      "loss": 1.1947,
+      "step": 3410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402485493182655e-05,
+      "loss": 1.0915,
+      "step": 3411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396916578504467e-05,
+      "loss": 0.9958,
+      "step": 3412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391348947655657e-05,
+      "loss": 1.1958,
+      "step": 3413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385782602620056e-05,
+      "loss": 0.9541,
+      "step": 3414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380217545381024e-05,
+      "loss": 1.1626,
+      "step": 3415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374653777921471e-05,
+      "loss": 1.1146,
+      "step": 3416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.369091302223847e-05,
+      "loss": 1.0974,
+      "step": 3417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363530120270141e-05,
+      "loss": 1.0686,
+      "step": 3418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357970234041877e-05,
+      "loss": 1.0973,
+      "step": 3419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352411645520126e-05,
+      "loss": 1.1156,
+      "step": 3420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.346854356685488e-05,
+      "loss": 1.106,
+      "step": 3421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.341298369518103e-05,
+      "loss": 1.0691,
+      "step": 3422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335743685997648e-05,
+      "loss": 1.0921,
+      "step": 3423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330190308103336e-05,
+      "loss": 1.136,
+      "step": 3424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324638237813912e-05,
+      "loss": 1.112,
+      "step": 3425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.319087477107663e-05,
+      "loss": 1.0709,
+      "step": 3426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313538027962394e-05,
+      "loss": 1.1227,
+      "step": 3427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.307989892355455e-05,
+      "loss": 1.0836,
+      "step": 3428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302443072263728e-05,
+      "loss": 1.1906,
+      "step": 3429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296897569663621e-05,
+      "loss": 1.1079,
+      "step": 3430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291353386531074e-05,
+      "loss": 1.0141,
+      "step": 3431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285810524841563e-05,
+      "loss": 1.1028,
+      "step": 3432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280268986570079e-05,
+      "loss": 1.0406,
+      "step": 3433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274728773691154e-05,
+      "loss": 1.079,
+      "step": 3434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269189888178843e-05,
+      "loss": 1.2429,
+      "step": 3435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263652332006734e-05,
+      "loss": 1.0168,
+      "step": 3436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258116107147933e-05,
+      "loss": 0.958,
+      "step": 3437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252581215575071e-05,
+      "loss": 1.1684,
+      "step": 3438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247047659260311e-05,
+      "loss": 1.059,
+      "step": 3439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241515440175338e-05,
+      "loss": 1.1514,
+      "step": 3440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235984560291357e-05,
+      "loss": 1.0361,
+      "step": 3441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230455021579102e-05,
+      "loss": 1.0805,
+      "step": 3442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224926826008828e-05,
+      "loss": 1.0911,
+      "step": 3443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2193999755503e-05,
+      "loss": 1.1332,
+      "step": 3444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213874472172815e-05,
+      "loss": 1.0844,
+      "step": 3445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.208350317845193e-05,
+      "loss": 1.0353,
+      "step": 3446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202827514535765e-05,
+      "loss": 1.0725,
+      "step": 3447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197306064212387e-05,
+      "loss": 1.1013,
+      "step": 3448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191785968842422e-05,
+      "loss": 1.1153,
+      "step": 3449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186267230392762e-05,
+      "loss": 1.118,
+      "step": 3450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.180749850829812e-05,
+      "loss": 1.1262,
+      "step": 3451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175233832119489e-05,
+      "loss": 1.1,
+      "step": 3452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.169719176227234e-05,
+      "loss": 1.1734,
+      "step": 3453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164205885117993e-05,
+      "loss": 1.2041,
+      "step": 3454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158693960756228e-05,
+      "loss": 1.109,
+      "step": 3455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153183405105915e-05,
+      "loss": 0.9999,
+      "step": 3456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.147674220130547e-05,
+      "loss": 1.1923,
+      "step": 3457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142166407793119e-05,
+      "loss": 1.1535,
+      "step": 3458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136659970056149e-05,
+      "loss": 1.0617,
+      "step": 3459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131154908881648e-05,
+      "loss": 1.1146,
+      "step": 3460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125651226231155e-05,
+      "loss": 1.0492,
+      "step": 3461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120148924065707e-05,
+      "loss": 1.0338,
+      "step": 3462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114648004345853e-05,
+      "loss": 1.2382,
+      "step": 3463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109148469031646e-05,
+      "loss": 1.0657,
+      "step": 3464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.103650320082655e-05,
+      "loss": 1.0971,
+      "step": 3465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098153559457935e-05,
+      "loss": 1.045,
+      "step": 3466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092658189116068e-05,
+      "loss": 0.9964,
+      "step": 3467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0871642110151305e-05,
+      "loss": 1.0464,
+      "step": 3468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081671627112704e-05,
+      "loss": 1.1524,
+      "step": 3469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0761804393658775e-05,
+      "loss": 1.1306,
+      "step": 3470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.070690649731231e-05,
+      "loss": 1.197,
+      "step": 3471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0652022601648575e-05,
+      "loss": 1.1043,
+      "step": 3472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059715272622346e-05,
+      "loss": 1.0815,
+      "step": 3473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0542296890587904e-05,
+      "loss": 0.9544,
+      "step": 3474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0487455114287794e-05,
+      "loss": 1.1646,
+      "step": 3475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043262741686408e-05,
+      "loss": 1.1465,
+      "step": 3476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037781381785256e-05,
+      "loss": 1.1574,
+      "step": 3477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032301433678411e-05,
+      "loss": 1.1747,
+      "step": 3478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026822899318458e-05,
+      "loss": 1.2477,
+      "step": 3479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021345780657477e-05,
+      "loss": 1.1319,
+      "step": 3480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015870079647044e-05,
+      "loss": 1.13,
+      "step": 3481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01039579823822e-05,
+      "loss": 1.0228,
+      "step": 3482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0049229383815755e-05,
+      "loss": 1.0275,
+      "step": 3483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9994515020271644e-05,
+      "loss": 1.0349,
+      "step": 3484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993981491124541e-05,
+      "loss": 1.2311,
+      "step": 3485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9885129076227456e-05,
+      "loss": 1.0984,
+      "step": 3486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983045753470308e-05,
+      "loss": 1.109,
+      "step": 3487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977580030615254e-05,
+      "loss": 1.1557,
+      "step": 3488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9721157410050976e-05,
+      "loss": 1.0801,
+      "step": 3489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966652886586843e-05,
+      "loss": 1.0554,
+      "step": 3490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961191469306984e-05,
+      "loss": 1.1005,
+      "step": 3491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9557314911115006e-05,
+      "loss": 1.0159,
+      "step": 3492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9502729539458535e-05,
+      "loss": 1.118,
+      "step": 3493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944815859755002e-05,
+      "loss": 1.1158,
+      "step": 3494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9393602104833824e-05,
+      "loss": 1.1619,
+      "step": 3495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933906008074923e-05,
+      "loss": 1.0688,
+      "step": 3496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.928453254473034e-05,
+      "loss": 0.9525,
+      "step": 3497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9230019516206034e-05,
+      "loss": 1.1962,
+      "step": 3498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.917552101460008e-05,
+      "loss": 1.135,
+      "step": 3499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.912103705933107e-05,
+      "loss": 1.0233,
+      "step": 3500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.906656766981242e-05,
+      "loss": 1.1583,
+      "step": 3501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901211286545234e-05,
+      "loss": 1.1482,
+      "step": 3502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.895767266565386e-05,
+      "loss": 0.8926,
+      "step": 3503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8903247089814736e-05,
+      "loss": 1.2208,
+      "step": 3504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884883615732759e-05,
+      "loss": 1.1015,
+      "step": 3505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.879443988757982e-05,
+      "loss": 1.105,
+      "step": 3506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874005829995357e-05,
+      "loss": 1.1509,
+      "step": 3507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868569141382581e-05,
+      "loss": 1.0446,
+      "step": 3508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863133924856814e-05,
+      "loss": 1.0195,
+      "step": 3509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857700182354704e-05,
+      "loss": 1.1436,
+      "step": 3510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.852267915812373e-05,
+      "loss": 1.0786,
+      "step": 3511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8468371271654096e-05,
+      "loss": 1.1681,
+      "step": 3512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841407818348885e-05,
+      "loss": 1.1124,
+      "step": 3513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8359799912973365e-05,
+      "loss": 1.0574,
+      "step": 3514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8305536479447765e-05,
+      "loss": 1.195,
+      "step": 3515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.825128790224681e-05,
+      "loss": 1.1133,
+      "step": 3516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.819705420070012e-05,
+      "loss": 1.011,
+      "step": 3517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.814283539413185e-05,
+      "loss": 1.0027,
+      "step": 3518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8088631501861034e-05,
+      "loss": 1.1713,
+      "step": 3519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803444254320115e-05,
+      "loss": 1.1182,
+      "step": 3520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798026853746059e-05,
+      "loss": 1.1442,
+      "step": 3521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792610950394222e-05,
+      "loss": 1.1895,
+      "step": 3522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7871965461943765e-05,
+      "loss": 1.1761,
+      "step": 3523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.781783643075743e-05,
+      "loss": 1.174,
+      "step": 3524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7763722429670274e-05,
+      "loss": 1.0728,
+      "step": 3525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7709623477963694e-05,
+      "loss": 0.9746,
+      "step": 3526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765553959491406e-05,
+      "loss": 1.1236,
+      "step": 3527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760147079979212e-05,
+      "loss": 1.1028,
+      "step": 3528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7547417111863444e-05,
+      "loss": 1.1351,
+      "step": 3529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7493378550388064e-05,
+      "loss": 1.0124,
+      "step": 3530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7439355134620696e-05,
+      "loss": 1.128,
+      "step": 3531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7385346883810596e-05,
+      "loss": 1.1857,
+      "step": 3532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733135381720174e-05,
+      "loss": 1.1611,
+      "step": 3533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7277375954032555e-05,
+      "loss": 1.1514,
+      "step": 3534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722341331353618e-05,
+      "loss": 1.2045,
+      "step": 3535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716946591494022e-05,
+      "loss": 1.2385,
+      "step": 3536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711553377746691e-05,
+      "loss": 0.9946,
+      "step": 3537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.706161692033298e-05,
+      "loss": 1.0875,
+      "step": 3538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700771536274985e-05,
+      "loss": 1.1313,
+      "step": 3539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695382912392331e-05,
+      "loss": 1.0484,
+      "step": 3540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6899958223053915e-05,
+      "loss": 1.0825,
+      "step": 3541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684610267933648e-05,
+      "loss": 1.1641,
+      "step": 3542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67922625119606e-05,
+      "loss": 1.1711,
+      "step": 3543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673843774011021e-05,
+      "loss": 1.2011,
+      "step": 3544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66846283829639e-05,
+      "loss": 1.2088,
+      "step": 3545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663083445969464e-05,
+      "loss": 1.1835,
+      "step": 3546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6577055989470076e-05,
+      "loss": 0.9808,
+      "step": 3547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.652329299145207e-05,
+      "loss": 0.9946,
+      "step": 3548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646954548479726e-05,
+      "loss": 1.1848,
+      "step": 3549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6415813488656565e-05,
+      "loss": 1.0989,
+      "step": 3550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636209702217552e-05,
+      "loss": 1.1398,
+      "step": 3551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630839610449403e-05,
+      "loss": 1.0816,
+      "step": 3552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625471075474649e-05,
+      "loss": 1.1275,
+      "step": 3553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6201040992061694e-05,
+      "loss": 1.1384,
+      "step": 3554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614738683556302e-05,
+      "loss": 1.1594,
+      "step": 3555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609374830436812e-05,
+      "loss": 1.1849,
+      "step": 3556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604012541758924e-05,
+      "loss": 1.1124,
+      "step": 3557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598651819433294e-05,
+      "loss": 1.1594,
+      "step": 3558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5932926653700203e-05,
+      "loss": 1.212,
+      "step": 3559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5879350814786426e-05,
+      "loss": 1.0658,
+      "step": 3560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.582579069668151e-05,
+      "loss": 1.0939,
+      "step": 3561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577224631846961e-05,
+      "loss": 1.1059,
+      "step": 3562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571871769922945e-05,
+      "loss": 1.2605,
+      "step": 3563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566520485803388e-05,
+      "loss": 1.0479,
+      "step": 3564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5611707813950396e-05,
+      "loss": 1.1187,
+      "step": 3565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5558226586040674e-05,
+      "loss": 1.2146,
+      "step": 3566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55047611933609e-05,
+      "loss": 1.0977,
+      "step": 3567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5451311654961456e-05,
+      "loss": 1.0772,
+      "step": 3568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5397877989887314e-05,
+      "loss": 0.9909,
+      "step": 3569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5344460217177454e-05,
+      "loss": 1.0689,
+      "step": 3570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529105835586552e-05,
+      "loss": 0.9891,
+      "step": 3571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523767242497927e-05,
+      "loss": 1.1737,
+      "step": 3572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518430244354093e-05,
+      "loss": 1.1421,
+      "step": 3573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5130948430566934e-05,
+      "loss": 1.0646,
+      "step": 3574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.507761040506807e-05,
+      "loss": 1.2255,
+      "step": 3575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.50242883860494e-05,
+      "loss": 1.118,
+      "step": 3576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497098239251038e-05,
+      "loss": 1.2044,
+      "step": 3577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4917692443444614e-05,
+      "loss": 1.1062,
+      "step": 3578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4864418557840136e-05,
+      "loss": 1.2171,
+      "step": 3579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4811160754679134e-05,
+      "loss": 1.0664,
+      "step": 3580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4757919052938134e-05,
+      "loss": 1.2069,
+      "step": 3581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470469347158784e-05,
+      "loss": 0.9919,
+      "step": 3582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465148402959339e-05,
+      "loss": 1.0257,
+      "step": 3583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459829074591394e-05,
+      "loss": 1.0929,
+      "step": 3584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454511363950314e-05,
+      "loss": 1.0922,
+      "step": 3585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44919527293086e-05,
+      "loss": 1.1473,
+      "step": 3586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443880803427239e-05,
+      "loss": 1.0737,
+      "step": 3587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4385679573330675e-05,
+      "loss": 0.9704,
+      "step": 3588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43325673654139e-05,
+      "loss": 1.1391,
+      "step": 3589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427947142944666e-05,
+      "loss": 1.1599,
+      "step": 3590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422639178434788e-05,
+      "loss": 1.0874,
+      "step": 3591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4173328449030436e-05,
+      "loss": 1.0805,
+      "step": 3592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4120281442401655e-05,
+      "loss": 1.0812,
+      "step": 3593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4067250783362856e-05,
+      "loss": 1.1754,
+      "step": 3594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401423649080969e-05,
+      "loss": 1.1085,
+      "step": 3595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3961238583631855e-05,
+      "loss": 1.045,
+      "step": 3596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390825708071325e-05,
+      "loss": 1.1862,
+      "step": 3597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38552920009319e-05,
+      "loss": 1.1159,
+      "step": 3598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380234336316008e-05,
+      "loss": 1.1864,
+      "step": 3599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.374941118626406e-05,
+      "loss": 1.1238,
+      "step": 3600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369649548910438e-05,
+      "loss": 1.1769,
+      "step": 3601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.364359629053566e-05,
+      "loss": 1.209,
+      "step": 3602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359071360940657e-05,
+      "loss": 1.2212,
+      "step": 3603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3537847464559946e-05,
+      "loss": 1.172,
+      "step": 3604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348499787483281e-05,
+      "loss": 0.9784,
+      "step": 3605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3432164859056154e-05,
+      "loss": 1.0827,
+      "step": 3606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337934843605519e-05,
+      "loss": 1.1459,
+      "step": 3607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.33265486246491e-05,
+      "loss": 1.0646,
+      "step": 3608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3273765443651235e-05,
+      "loss": 1.0034,
+      "step": 3609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3220998911868913e-05,
+      "loss": 1.274,
+      "step": 3610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316824904810369e-05,
+      "loss": 1.2058,
+      "step": 3611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3115515871151e-05,
+      "loss": 1.1416,
+      "step": 3612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30627993998005e-05,
+      "loss": 1.1827,
+      "step": 3613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.301009965283576e-05,
+      "loss": 1.0412,
+      "step": 3614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2957416649034475e-05,
+      "loss": 0.9322,
+      "step": 3615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290475040716827e-05,
+      "loss": 0.9815,
+      "step": 3616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285210094600295e-05,
+      "loss": 1.1357,
+      "step": 3617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279946828429823e-05,
+      "loss": 1.0297,
+      "step": 3618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274685244080783e-05,
+      "loss": 1.2611,
+      "step": 3619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2694253434279585e-05,
+      "loss": 1.3117,
+      "step": 3620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264167128345523e-05,
+      "loss": 1.1544,
+      "step": 3621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2589106007070486e-05,
+      "loss": 1.064,
+      "step": 3622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2536557623855185e-05,
+      "loss": 1.2438,
+      "step": 3623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248402615253301e-05,
+      "loss": 1.166,
+      "step": 3624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2431511611821626e-05,
+      "loss": 1.1175,
+      "step": 3625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237901402043278e-05,
+      "loss": 1.182,
+      "step": 3626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2326533397072076e-05,
+      "loss": 1.133,
+      "step": 3627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227406976043905e-05,
+      "loss": 1.1564,
+      "step": 3628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22216231292273e-05,
+      "loss": 0.9762,
+      "step": 3629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2169193522124284e-05,
+      "loss": 1.0497,
+      "step": 3630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211678095781136e-05,
+      "loss": 1.139,
+      "step": 3631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206438545496395e-05,
+      "loss": 1.1853,
+      "step": 3632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2012007032251256e-05,
+      "loss": 1.1171,
+      "step": 3633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195964570833641e-05,
+      "loss": 1.1865,
+      "step": 3634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190730150187658e-05,
+      "loss": 1.0779,
+      "step": 3635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1854974431522696e-05,
+      "loss": 1.0759,
+      "step": 3636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1802664515919595e-05,
+      "loss": 1.1321,
+      "step": 3637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175037177370612e-05,
+      "loss": 1.1127,
+      "step": 3638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169809622351486e-05,
+      "loss": 1.0204,
+      "step": 3639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164583788397234e-05,
+      "loss": 1.1019,
+      "step": 3640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159359677369889e-05,
+      "loss": 1.0556,
+      "step": 3641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154137291130885e-05,
+      "loss": 1.1694,
+      "step": 3642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148916631541023e-05,
+      "loss": 1.1383,
+      "step": 3643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1436977004605046e-05,
+      "loss": 1.0735,
+      "step": 3644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138480499748905e-05,
+      "loss": 1.2048,
+      "step": 3645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133265031265189e-05,
+      "loss": 1.062,
+      "step": 3646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.128051296867693e-05,
+      "loss": 1.0966,
+      "step": 3647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122839298414156e-05,
+      "loss": 1.1293,
+      "step": 3648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1176290377616754e-05,
+      "loss": 1.24,
+      "step": 3649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11242051676675e-05,
+      "loss": 1.0429,
+      "step": 3650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1072137372852455e-05,
+      "loss": 1.1123,
+      "step": 3651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1020087011724095e-05,
+      "loss": 1.0326,
+      "step": 3652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096805410282868e-05,
+      "loss": 1.1792,
+      "step": 3653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091603866470631e-05,
+      "loss": 1.182,
+      "step": 3654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086404071589077e-05,
+      "loss": 1.2008,
+      "step": 3655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0812060274909745e-05,
+      "loss": 1.0656,
+      "step": 3656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076009736028453e-05,
+      "loss": 1.082,
+      "step": 3657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.070815199053026e-05,
+      "loss": 1.1013,
+      "step": 3658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065622418415577e-05,
+      "loss": 1.0777,
+      "step": 3659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0604313959663726e-05,
+      "loss": 1.1422,
+      "step": 3660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0552421335550426e-05,
+      "loss": 1.075,
+      "step": 3661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050054633030603e-05,
+      "loss": 1.1669,
+      "step": 3662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0448688962414194e-05,
+      "loss": 1.0267,
+      "step": 3663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0396849250352526e-05,
+      "loss": 1.0334,
+      "step": 3664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03450272125922e-05,
+      "loss": 0.9526,
+      "step": 3665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029322286759819e-05,
+      "loss": 1.1315,
+      "step": 3666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0241436233829044e-05,
+      "loss": 1.0728,
+      "step": 3667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018966732973719e-05,
+      "loss": 1.0266,
+      "step": 3668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.013791617376846e-05,
+      "loss": 1.1222,
+      "step": 3669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0086182784362625e-05,
+      "loss": 1.1009,
+      "step": 3670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003446717995296e-05,
+      "loss": 1.1066,
+      "step": 3671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998276937896653e-05,
+      "loss": 1.0769,
+      "step": 3672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993108939982395e-05,
+      "loss": 1.2136,
+      "step": 3673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987942726093954e-05,
+      "loss": 1.0422,
+      "step": 3674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982778298072118e-05,
+      "loss": 1.2031,
+      "step": 3675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977615657757056e-05,
+      "loss": 1.0289,
+      "step": 3676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972454806988281e-05,
+      "loss": 1.0338,
+      "step": 3677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967295747604685e-05,
+      "loss": 1.2799,
+      "step": 3678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9621384814445084e-05,
+      "loss": 1.2522,
+      "step": 3679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956983010345357e-05,
+      "loss": 1.1116,
+      "step": 3680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9518293361441966e-05,
+      "loss": 1.004,
+      "step": 3681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9466774606773605e-05,
+      "loss": 1.1639,
+      "step": 3682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9415273857805255e-05,
+      "loss": 1.1331,
+      "step": 3683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9363791132887494e-05,
+      "loss": 1.1534,
+      "step": 3684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9312326450364176e-05,
+      "loss": 1.0097,
+      "step": 3685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9260879828573006e-05,
+      "loss": 1.0105,
+      "step": 3686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.920945128584507e-05,
+      "loss": 1.2717,
+      "step": 3687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.915804084050516e-05,
+      "loss": 1.1253,
+      "step": 3688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9106648510871456e-05,
+      "loss": 1.2089,
+      "step": 3689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905527431525589e-05,
+      "loss": 1.1116,
+      "step": 3690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900391827196367e-05,
+      "loss": 1.0693,
+      "step": 3691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8952580399293766e-05,
+      "loss": 1.0486,
+      "step": 3692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890126071553853e-05,
+      "loss": 1.1518,
+      "step": 3693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884995923898396e-05,
+      "loss": 1.1351,
+      "step": 3694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879867598790946e-05,
+      "loss": 0.9381,
+      "step": 3695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874741098058796e-05,
+      "loss": 1.0508,
+      "step": 3696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869616423528588e-05,
+      "loss": 1.0777,
+      "step": 3697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864493577026322e-05,
+      "loss": 1.0017,
+      "step": 3698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859372560377334e-05,
+      "loss": 1.0737,
+      "step": 3699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854253375406322e-05,
+      "loss": 1.2536,
+      "step": 3700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849136023937321e-05,
+      "loss": 1.2136,
+      "step": 3701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844020507793712e-05,
+      "loss": 1.0095,
+      "step": 3702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8389068287982244e-05,
+      "loss": 1.1265,
+      "step": 3703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833794988772941e-05,
+      "loss": 1.1204,
+      "step": 3704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828684989539275e-05,
+      "loss": 1.0967,
+      "step": 3705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.823576832918001e-05,
+      "loss": 0.9641,
+      "step": 3706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8184705207292144e-05,
+      "loss": 1.118,
+      "step": 3707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8133660547923754e-05,
+      "loss": 1.1874,
+      "step": 3708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808263436926271e-05,
+      "loss": 1.0879,
+      "step": 3709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803162668949044e-05,
+      "loss": 1.0692,
+      "step": 3710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7980637526781635e-05,
+      "loss": 1.0759,
+      "step": 3711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792966689930447e-05,
+      "loss": 1.2337,
+      "step": 3712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7878714825220464e-05,
+      "loss": 1.101,
+      "step": 3713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782778132268462e-05,
+      "loss": 1.2124,
+      "step": 3714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7776866409845186e-05,
+      "loss": 1.1078,
+      "step": 3715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772597010484395e-05,
+      "loss": 1.1205,
+      "step": 3716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.767509242581595e-05,
+      "loss": 1.1333,
+      "step": 3717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7624233390889606e-05,
+      "loss": 1.0879,
+      "step": 3718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757339301818667e-05,
+      "loss": 1.036,
+      "step": 3719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7522571325822376e-05,
+      "loss": 1.2444,
+      "step": 3720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747176833190511e-05,
+      "loss": 1.107,
+      "step": 3721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742098405453682e-05,
+      "loss": 1.1024,
+      "step": 3722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7370218511812515e-05,
+      "loss": 1.2115,
+      "step": 3723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731947172182078e-05,
+      "loss": 1.2413,
+      "step": 3724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726874370264333e-05,
+      "loss": 1.0447,
+      "step": 3725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7218034472355344e-05,
+      "loss": 1.0397,
+      "step": 3726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7167344049025175e-05,
+      "loss": 1.0428,
+      "step": 3727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711667245071464e-05,
+      "loss": 1.2467,
+      "step": 3728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706601969547859e-05,
+      "loss": 1.0534,
+      "step": 3729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701538580136543e-05,
+      "loss": 1.2835,
+      "step": 3730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696477078641667e-05,
+      "loss": 1.1105,
+      "step": 3731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6914174668667197e-05,
+      "loss": 1.1457,
+      "step": 3732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6863597466145116e-05,
+      "loss": 1.1288,
+      "step": 3733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6813039196871776e-05,
+      "loss": 1.1897,
+      "step": 3734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6762499878861764e-05,
+      "loss": 1.0902,
+      "step": 3735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671197953012303e-05,
+      "loss": 1.1987,
+      "step": 3736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666147816865661e-05,
+      "loss": 1.1423,
+      "step": 3737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661099581245694e-05,
+      "loss": 1.1178,
+      "step": 3738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6560532479511535e-05,
+      "loss": 1.1645,
+      "step": 3739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.651008818780119e-05,
+      "loss": 1.086,
+      "step": 3740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64596629552999e-05,
+      "loss": 1.1762,
+      "step": 3741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6409256799974944e-05,
+      "loss": 1.2224,
+      "step": 3742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.635886973978668e-05,
+      "loss": 1.0288,
+      "step": 3743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63085017926888e-05,
+      "loss": 1.2216,
+      "step": 3744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625815297662808e-05,
+      "loss": 1.1146,
+      "step": 3745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6207823309544495e-05,
+      "loss": 1.0916,
+      "step": 3746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6157512809371185e-05,
+      "loss": 1.025,
+      "step": 3747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610722149403458e-05,
+      "loss": 1.1459,
+      "step": 3748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605694938145408e-05,
+      "loss": 1.2603,
+      "step": 3749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600669648954246e-05,
+      "loss": 1.1042,
+      "step": 3750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595646283620547e-05,
+      "loss": 1.2045,
+      "step": 3751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590624843934207e-05,
+      "loss": 1.0919,
+      "step": 3752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5856053316844317e-05,
+      "loss": 1.099,
+      "step": 3753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580587748659752e-05,
+      "loss": 1.0782,
+      "step": 3754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575572096648001e-05,
+      "loss": 1.0914,
+      "step": 3755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5705583774363204e-05,
+      "loss": 1.1245,
+      "step": 3756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5655465928111776e-05,
+      "loss": 0.9898,
+      "step": 3757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560536744558338e-05,
+      "loss": 1.1313,
+      "step": 3758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5555288344628766e-05,
+      "loss": 1.0153,
+      "step": 3759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.550522864309191e-05,
+      "loss": 1.2169,
+      "step": 3760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5455188358809755e-05,
+      "loss": 1.1477,
+      "step": 3761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5405167509612304e-05,
+      "loss": 1.0719,
+      "step": 3762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535516611332278e-05,
+      "loss": 1.162,
+      "step": 3763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530518418775733e-05,
+      "loss": 0.9118,
+      "step": 3764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525522175072521e-05,
+      "loss": 1.1812,
+      "step": 3765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5205278820028784e-05,
+      "loss": 1.0954,
+      "step": 3766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515535541346341e-05,
+      "loss": 1.0538,
+      "step": 3767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.510545154881744e-05,
+      "loss": 1.1568,
+      "step": 3768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505556724387241e-05,
+      "loss": 1.1273,
+      "step": 3769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500570251640277e-05,
+      "loss": 1.1155,
+      "step": 3770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4955857384175984e-05,
+      "loss": 1.0461,
+      "step": 3771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.490603186495264e-05,
+      "loss": 1.0759,
+      "step": 3772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485622597648624e-05,
+      "loss": 1.0465,
+      "step": 3773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4806439736523295e-05,
+      "loss": 1.1971,
+      "step": 3774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475667316280341e-05,
+      "loss": 1.1035,
+      "step": 3775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470692627305908e-05,
+      "loss": 1.1341,
+      "step": 3776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465719908501581e-05,
+      "loss": 1.2121,
+      "step": 3777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4607491616392094e-05,
+      "loss": 1.1331,
+      "step": 3778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4557803884899444e-05,
+      "loss": 1.09,
+      "step": 3779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450813590824224e-05,
+      "loss": 1.1486,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445848770411795e-05,
+      "loss": 1.031,
+      "step": 3781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4408859290216895e-05,
+      "loss": 1.2219,
+      "step": 3782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4359250684222383e-05,
+      "loss": 1.1754,
+      "step": 3783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4309661903810616e-05,
+      "loss": 1.0919,
+      "step": 3784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426009296665084e-05,
+      "loss": 1.0651,
+      "step": 3785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421054389040511e-05,
+      "loss": 1.0824,
+      "step": 3786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416101469272853e-05,
+      "loss": 1.14,
+      "step": 3787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4111505391269e-05,
+      "loss": 1.1776,
+      "step": 3788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406201600366739e-05,
+      "loss": 1.1027,
+      "step": 3789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4012546547557434e-05,
+      "loss": 1.1752,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.396309704056587e-05,
+      "loss": 1.018,
+      "step": 3791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391366750031217e-05,
+      "loss": 1.0545,
+      "step": 3792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3864257944408846e-05,
+      "loss": 0.9805,
+      "step": 3793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381486839046121e-05,
+      "loss": 1.1791,
+      "step": 3794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376549885606742e-05,
+      "loss": 0.9248,
+      "step": 3795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3716149358818526e-05,
+      "loss": 1.1379,
+      "step": 3796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366681991629851e-05,
+      "loss": 1.0514,
+      "step": 3797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3617510546084073e-05,
+      "loss": 1.1314,
+      "step": 3798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356822126574496e-05,
+      "loss": 1.234,
+      "step": 3799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351895209284347e-05,
+      "loss": 1.0009,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3469703044935026e-05,
+      "loss": 1.204,
+      "step": 3801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342047413956769e-05,
+      "loss": 1.0751,
+      "step": 3802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337126539428248e-05,
+      "loss": 1.215,
+      "step": 3803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332207682661309e-05,
+      "loss": 1.2157,
+      "step": 3804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327290845408622e-05,
+      "loss": 1.079,
+      "step": 3805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3223760294221084e-05,
+      "loss": 1.1196,
+      "step": 3806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317463236453e-05,
+      "loss": 1.1248,
+      "step": 3807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312552468251785e-05,
+      "loss": 0.9879,
+      "step": 3808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3076437265682476e-05,
+      "loss": 1.1624,
+      "step": 3809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302737013151438e-05,
+      "loss": 1.0846,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297832329749687e-05,
+      "loss": 1.1436,
+      "step": 3811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292929678110598e-05,
+      "loss": 1.2243,
+      "step": 3812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2880290599810636e-05,
+      "loss": 1.1428,
+      "step": 3813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283130477107234e-05,
+      "loss": 1.1023,
+      "step": 3814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278233931234552e-05,
+      "loss": 1.1027,
+      "step": 3815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2733394241077215e-05,
+      "loss": 1.2648,
+      "step": 3816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268446957470725e-05,
+      "loss": 1.0445,
+      "step": 3817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263556533066815e-05,
+      "loss": 1.1149,
+      "step": 3818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258668152638522e-05,
+      "loss": 1.0795,
+      "step": 3819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.25378181792764e-05,
+      "loss": 1.1137,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24889753067525e-05,
+      "loss": 1.1363,
+      "step": 3821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244015292621677e-05,
+      "loss": 1.1948,
+      "step": 3822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.239135105506543e-05,
+      "loss": 1.0981,
+      "step": 3823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2342569710687184e-05,
+      "loss": 1.0887,
+      "step": 3824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2293808910463606e-05,
+      "loss": 1.2024,
+      "step": 3825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224506867176877e-05,
+      "loss": 1.0967,
+      "step": 3826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219634901196963e-05,
+      "loss": 1.0635,
+      "step": 3827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2147649948425525e-05,
+      "loss": 1.0936,
+      "step": 3828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2098971498488734e-05,
+      "loss": 1.0513,
+      "step": 3829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2050313679504015e-05,
+      "loss": 1.1384,
+      "step": 3830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2001676508808886e-05,
+      "loss": 1.0778,
+      "step": 3831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.195306000373344e-05,
+      "loss": 1.1438,
+      "step": 3832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19044641816004e-05,
+      "loss": 1.1358,
+      "step": 3833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1855889059725117e-05,
+      "loss": 1.2715,
+      "step": 3834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180733465541568e-05,
+      "loss": 1.1102,
+      "step": 3835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17588009859726e-05,
+      "loss": 1.0803,
+      "step": 3836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171028806868921e-05,
+      "loss": 1.149,
+      "step": 3837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166179592085129e-05,
+      "loss": 1.0204,
+      "step": 3838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.161332455973729e-05,
+      "loss": 1.1283,
+      "step": 3839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1564874002618194e-05,
+      "loss": 1.1551,
+      "step": 3840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1516444266757694e-05,
+      "loss": 1.1275,
+      "step": 3841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1468035369411904e-05,
+      "loss": 1.0128,
+      "step": 3842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1419647327829724e-05,
+      "loss": 1.0717,
+      "step": 3843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137128015925233e-05,
+      "loss": 0.954,
+      "step": 3844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132293388091375e-05,
+      "loss": 1.1154,
+      "step": 3845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127460851004037e-05,
+      "loss": 1.0278,
+      "step": 3846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122630406385126e-05,
+      "loss": 1.0079,
+      "step": 3847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117802055955792e-05,
+      "loss": 0.988,
+      "step": 3848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112975801436454e-05,
+      "loss": 1.1164,
+      "step": 3849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10815164454676e-05,
+      "loss": 1.0555,
+      "step": 3850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103329587005635e-05,
+      "loss": 1.1075,
+      "step": 3851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0985096305312407e-05,
+      "loss": 1.0592,
+      "step": 3852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093691776841002e-05,
+      "loss": 1.2827,
+      "step": 3853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088876027651585e-05,
+      "loss": 1.1398,
+      "step": 3854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084062384678907e-05,
+      "loss": 1.1062,
+      "step": 3855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0792508496381366e-05,
+      "loss": 0.9963,
+      "step": 3856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0744414242436944e-05,
+      "loss": 1.2199,
+      "step": 3857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0696341102092425e-05,
+      "loss": 0.9629,
+      "step": 3858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064828909247701e-05,
+      "loss": 1.022,
+      "step": 3859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060025823071226e-05,
+      "loss": 1.0639,
+      "step": 3860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055224853391226e-05,
+      "loss": 1.1327,
+      "step": 3861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05042600191835e-05,
+      "loss": 1.1247,
+      "step": 3862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0456292703625044e-05,
+      "loss": 1.2655,
+      "step": 3863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040834660432824e-05,
+      "loss": 1.2365,
+      "step": 3864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036042173837706e-05,
+      "loss": 1.0528,
+      "step": 3865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031251812284769e-05,
+      "loss": 1.0379,
+      "step": 3866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0264635774808945e-05,
+      "loss": 1.097,
+      "step": 3867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0216774711321925e-05,
+      "loss": 1.0776,
+      "step": 3868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016893494944026e-05,
+      "loss": 1.1543,
+      "step": 3869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0121116506209876e-05,
+      "loss": 1.1606,
+      "step": 3870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007331939866925e-05,
+      "loss": 1.0545,
+      "step": 3871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.002554364384903e-05,
+      "loss": 1.1188,
+      "step": 3872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99777892587725e-05,
+      "loss": 1.0402,
+      "step": 3873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993005626045514e-05,
+      "loss": 0.9382,
+      "step": 3874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9882344665904956e-05,
+      "loss": 1.1758,
+      "step": 3875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983465449212225e-05,
+      "loss": 1.0588,
+      "step": 3876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9786985756099674e-05,
+      "loss": 1.0028,
+      "step": 3877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9739338474822253e-05,
+      "loss": 1.2163,
+      "step": 3878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969171266526745e-05,
+      "loss": 0.9874,
+      "step": 3879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.964410834440494e-05,
+      "loss": 1.0988,
+      "step": 3880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959652552919688e-05,
+      "loss": 1.0397,
+      "step": 3881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954896423659767e-05,
+      "loss": 1.0689,
+      "step": 3882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9501424483554074e-05,
+      "loss": 1.0401,
+      "step": 3883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945390628700513e-05,
+      "loss": 1.248,
+      "step": 3884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940640966388231e-05,
+      "loss": 1.0897,
+      "step": 3885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935893463110926e-05,
+      "loss": 1.0001,
+      "step": 3886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931148120560211e-05,
+      "loss": 1.0901,
+      "step": 3887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.926404940426904e-05,
+      "loss": 1.0199,
+      "step": 3888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921663924401079e-05,
+      "loss": 1.0199,
+      "step": 3889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916925074172019e-05,
+      "loss": 1.1137,
+      "step": 3890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912188391428249e-05,
+      "loss": 1.1406,
+      "step": 3891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90745387785751e-05,
+      "loss": 1.0378,
+      "step": 3892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9027215351467864e-05,
+      "loss": 1.1226,
+      "step": 3893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897991364982263e-05,
+      "loss": 1.1269,
+      "step": 3894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893263369049378e-05,
+      "loss": 1.0733,
+      "step": 3895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888537549032776e-05,
+      "loss": 1.1071,
+      "step": 3896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8838139066163394e-05,
+      "loss": 1.12,
+      "step": 3897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8790924434831655e-05,
+      "loss": 0.9865,
+      "step": 3898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8743731613155785e-05,
+      "loss": 1.208,
+      "step": 3899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869656061795121e-05,
+      "loss": 1.0725,
+      "step": 3900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8649411466025696e-05,
+      "loss": 1.0006,
+      "step": 3901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860228417417907e-05,
+      "loss": 0.9458,
+      "step": 3902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855517875920354e-05,
+      "loss": 1.0974,
+      "step": 3903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850809523788337e-05,
+      "loss": 1.0718,
+      "step": 3904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846103362699509e-05,
+      "loss": 1.0345,
+      "step": 3905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841399394330739e-05,
+      "loss": 1.0296,
+      "step": 3906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8366976203581226e-05,
+      "loss": 1.1689,
+      "step": 3907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8319980424569625e-05,
+      "loss": 1.0831,
+      "step": 3908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827300662301791e-05,
+      "loss": 1.1152,
+      "step": 3909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822605481566347e-05,
+      "loss": 1.2104,
+      "step": 3910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81791250192359e-05,
+      "loss": 1.1702,
+      "step": 3911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81322172504569e-05,
+      "loss": 1.0461,
+      "step": 3912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808533152604047e-05,
+      "loss": 1.0794,
+      "step": 3913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803846786269255e-05,
+      "loss": 1.0761,
+      "step": 3914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7991626277111404e-05,
+      "loss": 1.2169,
+      "step": 3915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7944806785987316e-05,
+      "loss": 1.0682,
+      "step": 3916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7898009406002746e-05,
+      "loss": 1.1774,
+      "step": 3917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7851234153832195e-05,
+      "loss": 1.03,
+      "step": 3918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.780448104614244e-05,
+      "loss": 1.1805,
+      "step": 3919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.775775009959221e-05,
+      "loss": 1.041,
+      "step": 3920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.77110413308324e-05,
+      "loss": 1.0076,
+      "step": 3921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7664354756506036e-05,
+      "loss": 1.09,
+      "step": 3922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.761769039324818e-05,
+      "loss": 1.1011,
+      "step": 3923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.757104825768599e-05,
+      "loss": 1.0387,
+      "step": 3924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.752442836643876e-05,
+      "loss": 1.098,
+      "step": 3925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.747783073611777e-05,
+      "loss": 1.1473,
+      "step": 3926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743125538332641e-05,
+      "loss": 1.0322,
+      "step": 3927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738470232466018e-05,
+      "loss": 1.103,
+      "step": 3928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733817157670656e-05,
+      "loss": 1.1478,
+      "step": 3929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7291663156045085e-05,
+      "loss": 1.1204,
+      "step": 3930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724517707924742e-05,
+      "loss": 1.0659,
+      "step": 3931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719871336287718e-05,
+      "loss": 1.1612,
+      "step": 3932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7152272023489996e-05,
+      "loss": 1.0439,
+      "step": 3933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.710585307763366e-05,
+      "loss": 1.1846,
+      "step": 3934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.705945654184785e-05,
+      "loss": 1.031,
+      "step": 3935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701308243266427e-05,
+      "loss": 1.0967,
+      "step": 3936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.696673076660674e-05,
+      "loss": 1.1056,
+      "step": 3937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6920401560190975e-05,
+      "loss": 1.1599,
+      "step": 3938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.687409482992469e-05,
+      "loss": 1.0687,
+      "step": 3939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.68278105923077e-05,
+      "loss": 1.1833,
+      "step": 3940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.678154886383168e-05,
+      "loss": 1.2226,
+      "step": 3941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.673530966098036e-05,
+      "loss": 1.0901,
+      "step": 3942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.668909300022937e-05,
+      "loss": 1.0379,
+      "step": 3943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.664289889804643e-05,
+      "loss": 1.0862,
+      "step": 3944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.659672737089107e-05,
+      "loss": 1.0437,
+      "step": 3945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6550578435214936e-05,
+      "loss": 1.0089,
+      "step": 3946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650445210746151e-05,
+      "loss": 1.2325,
+      "step": 3947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645834840406625e-05,
+      "loss": 1.1324,
+      "step": 3948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6412267341456505e-05,
+      "loss": 1.0579,
+      "step": 3949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6366208936051705e-05,
+      "loss": 1.0277,
+      "step": 3950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6320173204263016e-05,
+      "loss": 1.1842,
+      "step": 3951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.62741601624937e-05,
+      "loss": 1.1998,
+      "step": 3952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6228169827138815e-05,
+      "loss": 1.0118,
+      "step": 3953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618220221458536e-05,
+      "loss": 1.2026,
+      "step": 3954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613625734121221e-05,
+      "loss": 1.11,
+      "step": 3955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609033522339025e-05,
+      "loss": 1.028,
+      "step": 3956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.604443587748209e-05,
+      "loss": 1.1469,
+      "step": 3957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.599855931984243e-05,
+      "loss": 1.3163,
+      "step": 3958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595270556681759e-05,
+      "loss": 1.0422,
+      "step": 3959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5906874634746015e-05,
+      "loss": 1.0087,
+      "step": 3960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5861066539957844e-05,
+      "loss": 1.0731,
+      "step": 3961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58152812987752e-05,
+      "loss": 0.9968,
+      "step": 3962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5769518927511966e-05,
+      "loss": 1.0977,
+      "step": 3963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5723779442474014e-05,
+      "loss": 1.1538,
+      "step": 3964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.567806285995883e-05,
+      "loss": 1.1945,
+      "step": 3965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.563236919625595e-05,
+      "loss": 1.1948,
+      "step": 3966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558669846764665e-05,
+      "loss": 1.2501,
+      "step": 3967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.55410506904041e-05,
+      "loss": 1.0492,
+      "step": 3968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.54954258807932e-05,
+      "loss": 1.1874,
+      "step": 3969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.544982405507072e-05,
+      "loss": 1.0948,
+      "step": 3970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540424522948521e-05,
+      "loss": 1.0558,
+      "step": 3971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5358689420277115e-05,
+      "loss": 1.1318,
+      "step": 3972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531315664367852e-05,
+      "loss": 1.0285,
+      "step": 3973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.526764691591349e-05,
+      "loss": 1.1824,
+      "step": 3974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.522216025319773e-05,
+      "loss": 1.2099,
+      "step": 3975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5176696671738795e-05,
+      "loss": 1.197,
+      "step": 3976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.513125618773595e-05,
+      "loss": 1.2338,
+      "step": 3977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5085838817380356e-05,
+      "loss": 0.9394,
+      "step": 3978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.504044457685479e-05,
+      "loss": 1.162,
+      "step": 3979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.499507348233396e-05,
+      "loss": 1.1827,
+      "step": 3980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49497255499841e-05,
+      "loss": 1.1167,
+      "step": 3981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.490440079596341e-05,
+      "loss": 1.0978,
+      "step": 3982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.485909923642168e-05,
+      "loss": 1.1518,
+      "step": 3983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481382088750054e-05,
+      "loss": 1.1388,
+      "step": 3984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.476856576533326e-05,
+      "loss": 1.0785,
+      "step": 3985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4723333886044964e-05,
+      "loss": 1.0734,
+      "step": 3986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467812526575228e-05,
+      "loss": 1.1485,
+      "step": 3987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.463293992056378e-05,
+      "loss": 1.083,
+      "step": 3988
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4587777866579566e-05,
+      "loss": 1.0833,
+      "step": 3989
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4542639119891575e-05,
+      "loss": 1.0623,
+      "step": 3990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4497523696583344e-05,
+      "loss": 1.2452,
+      "step": 3991
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.445243161273016e-05,
+      "loss": 1.1422,
+      "step": 3992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4407362884398906e-05,
+      "loss": 1.104,
+      "step": 3993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.436231752764829e-05,
+      "loss": 1.1466,
+      "step": 3994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.431729555852853e-05,
+      "loss": 1.1481,
+      "step": 3995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4272296993081644e-05,
+      "loss": 1.2325,
+      "step": 3996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422732184734124e-05,
+      "loss": 1.1282,
+      "step": 3997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.41823701373326e-05,
+      "loss": 1.3084,
+      "step": 3998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413744187907261e-05,
+      "loss": 1.2073,
+      "step": 3999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4092537088569885e-05,
+      "loss": 1.1476,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.40476557818246e-05,
+      "loss": 1.1374,
+      "step": 4001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4002797974828695e-05,
+      "loss": 1.0588,
+      "step": 4002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.395796368356548e-05,
+      "loss": 1.2085,
+      "step": 4003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.391315292401017e-05,
+      "loss": 1.021,
+      "step": 4004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3868365712129405e-05,
+      "loss": 1.1059,
+      "step": 4005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.382360206388153e-05,
+      "loss": 1.0669,
+      "step": 4006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.377886199521644e-05,
+      "loss": 1.1203,
+      "step": 4007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.373414552207572e-05,
+      "loss": 1.0487,
+      "step": 4008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.368945266039237e-05,
+      "loss": 1.1325,
+      "step": 4009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3644783426091176e-05,
+      "loss": 1.1753,
+      "step": 4010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3600137835088344e-05,
+      "loss": 1.0965,
+      "step": 4011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.35555159032918e-05,
+      "loss": 1.2198,
+      "step": 4012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.351091764660093e-05,
+      "loss": 1.1164,
+      "step": 4013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3466343080906716e-05,
+      "loss": 1.108,
+      "step": 4014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3421792222091684e-05,
+      "loss": 1.0333,
+      "step": 4015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.337726508602999e-05,
+      "loss": 1.1409,
+      "step": 4016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.333276168858722e-05,
+      "loss": 1.1209,
+      "step": 4017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328828204562062e-05,
+      "loss": 1.1063,
+      "step": 4018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324382617297887e-05,
+      "loss": 1.1368,
+      "step": 4019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.319939408650225e-05,
+      "loss": 1.1074,
+      "step": 4020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.31549858020225e-05,
+      "loss": 1.1695,
+      "step": 4021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311060133536297e-05,
+      "loss": 1.0984,
+      "step": 4022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.306624070233842e-05,
+      "loss": 1.1461,
+      "step": 4023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3021903918755236e-05,
+      "loss": 1.031,
+      "step": 4024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.297759100041113e-05,
+      "loss": 1.0834,
+      "step": 4025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.293330196309553e-05,
+      "loss": 1.1385,
+      "step": 4026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288903682258917e-05,
+      "loss": 1.2092,
+      "step": 4027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.284479559466439e-05,
+      "loss": 1.1014,
+      "step": 4028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280057829508492e-05,
+      "loss": 1.1597,
+      "step": 4029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.275638493960608e-05,
+      "loss": 1.0684,
+      "step": 4030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2712215543974475e-05,
+      "loss": 1.1991,
+      "step": 4031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2668070123928385e-05,
+      "loss": 1.1358,
+      "step": 4032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262394869519736e-05,
+      "loss": 1.1505,
+      "step": 4033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.257985127350257e-05,
+      "loss": 1.1023,
+      "step": 4034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253577787455651e-05,
+      "loss": 1.0287,
+      "step": 4035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.249172851406313e-05,
+      "loss": 1.0859,
+      "step": 4036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244770320771784e-05,
+      "loss": 1.0562,
+      "step": 4037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2403701971207536e-05,
+      "loss": 1.0959,
+      "step": 4038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235972482021039e-05,
+      "loss": 1.124,
+      "step": 4039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2315771770396165e-05,
+      "loss": 1.0795,
+      "step": 4040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.227184283742591e-05,
+      "loss": 1.1218,
+      "step": 4041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.222793803695213e-05,
+      "loss": 1.2218,
+      "step": 4042
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.218405738461868e-05,
+      "loss": 1.161,
+      "step": 4043
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214020089606094e-05,
+      "loss": 1.0874,
+      "step": 4044
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.209636858690551e-05,
+      "loss": 1.0379,
+      "step": 4045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.205256047277054e-05,
+      "loss": 0.9527,
+      "step": 4046
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200877656926543e-05,
+      "loss": 1.0966,
+      "step": 4047
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1965016891991e-05,
+      "loss": 1.1112,
+      "step": 4048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.192128145653943e-05,
+      "loss": 1.1744,
+      "step": 4049
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1877570278494315e-05,
+      "loss": 1.1894,
+      "step": 4050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1833883373430515e-05,
+      "loss": 1.1662,
+      "step": 4051
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.179022075691435e-05,
+      "loss": 1.1394,
+      "step": 4052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174658244450337e-05,
+      "loss": 1.1085,
+      "step": 4053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1702968451746554e-05,
+      "loss": 1.214,
+      "step": 4054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1659378794184126e-05,
+      "loss": 1.0739,
+      "step": 4055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.161581348734777e-05,
+      "loss": 1.0185,
+      "step": 4056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1572272546760383e-05,
+      "loss": 1.0019,
+      "step": 4057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1528755987936186e-05,
+      "loss": 1.1427,
+      "step": 4058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148526382638081e-05,
+      "loss": 0.9478,
+      "step": 4059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.144179607759108e-05,
+      "loss": 1.1791,
+      "step": 4060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.139835275705514e-05,
+      "loss": 1.1714,
+      "step": 4061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135493388025253e-05,
+      "loss": 1.0059,
+      "step": 4062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131153946265397e-05,
+      "loss": 1.1601,
+      "step": 4063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.126816951972147e-05,
+      "loss": 1.0811,
+      "step": 4064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.122482406690842e-05,
+      "loss": 1.0183,
+      "step": 4065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.118150311965937e-05,
+      "loss": 1.1174,
+      "step": 4066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.113820669341017e-05,
+      "loss": 1.2645,
+      "step": 4067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1094934803587996e-05,
+      "loss": 1.0673,
+      "step": 4068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10516874656112e-05,
+      "loss": 1.1835,
+      "step": 4069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.100846469488939e-05,
+      "loss": 0.9629,
+      "step": 4070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096526650682352e-05,
+      "loss": 1.1195,
+      "step": 4071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0922092916805665e-05,
+      "loss": 1.1401,
+      "step": 4072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087894394021917e-05,
+      "loss": 1.017,
+      "step": 4073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0835819592438674e-05,
+      "loss": 1.2009,
+      "step": 4074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.079271988882997e-05,
+      "loss": 1.1013,
+      "step": 4075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.074964484475004e-05,
+      "loss": 1.0393,
+      "step": 4076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.070659447554719e-05,
+      "loss": 1.0915,
+      "step": 4077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.066356879656087e-05,
+      "loss": 1.172,
+      "step": 4078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.06205678231217e-05,
+      "loss": 0.9634,
+      "step": 4079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.057759157055153e-05,
+      "loss": 0.9433,
+      "step": 4080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.053464005416347e-05,
+      "loss": 1.1977,
+      "step": 4081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0491713289261657e-05,
+      "loss": 0.9426,
+      "step": 4082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0448811291141577e-05,
+      "loss": 1.0548,
+      "step": 4083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0405934075089802e-05,
+      "loss": 1.1745,
+      "step": 4084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0363081656384075e-05,
+      "loss": 1.0448,
+      "step": 4085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032025405029327e-05,
+      "loss": 1.0903,
+      "step": 4086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0277451272077562e-05,
+      "loss": 1.1961,
+      "step": 4087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0234673336988094e-05,
+      "loss": 1.2333,
+      "step": 4088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0191920260267325e-05,
+      "loss": 1.1165,
+      "step": 4089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0149192057148734e-05,
+      "loss": 1.1327,
+      "step": 4090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0106488742856996e-05,
+      "loss": 1.0277,
+      "step": 4091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0063810332607865e-05,
+      "loss": 1.1368,
+      "step": 4092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.002115684160832e-05,
+      "loss": 0.9614,
+      "step": 4093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997852828505634e-05,
+      "loss": 1.049,
+      "step": 4094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993592467814115e-05,
+      "loss": 1.2416,
+      "step": 4095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989334603604297e-05,
+      "loss": 1.135,
+      "step": 4096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9850792373933178e-05,
+      "loss": 1.1255,
+      "step": 4097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9808263706974216e-05,
+      "loss": 1.0261,
+      "step": 4098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9765760050319703e-05,
+      "loss": 1.2663,
+      "step": 4099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9723281419114225e-05,
+      "loss": 1.2087,
+      "step": 4100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9680827828493618e-05,
+      "loss": 1.1019,
+      "step": 4101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9638399293584572e-05,
+      "loss": 1.0079,
+      "step": 4102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9595995829505053e-05,
+      "loss": 1.0628,
+      "step": 4103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.955361745136397e-05,
+      "loss": 0.9799,
+      "step": 4104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9511264174261377e-05,
+      "loss": 1.143,
+      "step": 4105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9468936013288284e-05,
+      "loss": 1.1997,
+      "step": 4106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9426632983526924e-05,
+      "loss": 1.0896,
+      "step": 4107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9384355100050322e-05,
+      "loss": 1.0908,
+      "step": 4108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9342102377922774e-05,
+      "loss": 1.1977,
+      "step": 4109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9299874832199458e-05,
+      "loss": 1.2401,
+      "step": 4110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9257672477926722e-05,
+      "loss": 1.1556,
+      "step": 4111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9215495330141806e-05,
+      "loss": 1.0213,
+      "step": 4112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9173343403873043e-05,
+      "loss": 0.9492,
+      "step": 4113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.913121671413971e-05,
+      "loss": 1.1066,
+      "step": 4114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9089115275952218e-05,
+      "loss": 1.0366,
+      "step": 4115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9047039104311824e-05,
+      "loss": 1.0275,
+      "step": 4116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9004988214210927e-05,
+      "loss": 1.0602,
+      "step": 4117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8962962620632827e-05,
+      "loss": 1.1819,
+      "step": 4118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8920962338551838e-05,
+      "loss": 1.0267,
+      "step": 4119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8878987382933198e-05,
+      "loss": 1.1702,
+      "step": 4120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8837037768733265e-05,
+      "loss": 1.1104,
+      "step": 4121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.879511351089921e-05,
+      "loss": 1.0961,
+      "step": 4122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8753214624369316e-05,
+      "loss": 1.0863,
+      "step": 4123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8711341124072633e-05,
+      "loss": 1.1373,
+      "step": 4124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.866949302492936e-05,
+      "loss": 1.1661,
+      "step": 4125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8627670341850522e-05,
+      "loss": 1.1561,
+      "step": 4126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.858587308973818e-05,
+      "loss": 1.1237,
+      "step": 4127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8544101283485223e-05,
+      "loss": 1.1094,
+      "step": 4128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850235493797563e-05,
+      "loss": 1.1171,
+      "step": 4129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.846063406808408e-05,
+      "loss": 1.2339,
+      "step": 4130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841893868867641e-05,
+      "loss": 1.1503,
+      "step": 4131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8377268814609203e-05,
+      "loss": 1.1379,
+      "step": 4132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8335624460730083e-05,
+      "loss": 1.0068,
+      "step": 4133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8294005641877486e-05,
+      "loss": 1.009,
+      "step": 4134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.82524123728808e-05,
+      "loss": 1.063,
+      "step": 4135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8210844668560244e-05,
+      "loss": 1.0135,
+      "step": 4136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816930254372705e-05,
+      "loss": 1.0616,
+      "step": 4137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8127786013183187e-05,
+      "loss": 1.0898,
+      "step": 4138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.808629509172165e-05,
+      "loss": 1.233,
+      "step": 4139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8044829794126215e-05,
+      "loss": 1.1632,
+      "step": 4140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8003390135171537e-05,
+      "loss": 1.3135,
+      "step": 4141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7961976129623134e-05,
+      "loss": 1.0983,
+      "step": 4142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.792058779223744e-05,
+      "loss": 1.0675,
+      "step": 4143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7879225137761666e-05,
+      "loss": 1.0494,
+      "step": 4144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783788818093399e-05,
+      "loss": 1.0686,
+      "step": 4145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7796576936483233e-05,
+      "loss": 1.1424,
+      "step": 4146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7755291419129247e-05,
+      "loss": 1.1846,
+      "step": 4147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7714031643582607e-05,
+      "loss": 1.1444,
+      "step": 4148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76727976245448e-05,
+      "loss": 1.0493,
+      "step": 4149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7631589376708035e-05,
+      "loss": 0.9427,
+      "step": 4150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7590406914755463e-05,
+      "loss": 1.1571,
+      "step": 4151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.754925025336088e-05,
+      "loss": 1.0743,
+      "step": 4152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.750811940718906e-05,
+      "loss": 1.1467,
+      "step": 4153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746701439089544e-05,
+      "loss": 1.1467,
+      "step": 4154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7425935219126387e-05,
+      "loss": 1.0198,
+      "step": 4155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7384881906518957e-05,
+      "loss": 1.0293,
+      "step": 4156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7343854467701014e-05,
+      "loss": 0.9827,
+      "step": 4157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7302852917291187e-05,
+      "loss": 1.145,
+      "step": 4158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7261877269898972e-05,
+      "loss": 1.1868,
+      "step": 4159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7220927540124507e-05,
+      "loss": 1.1033,
+      "step": 4160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.718000374255881e-05,
+      "loss": 1.083,
+      "step": 4161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7139105891783588e-05,
+      "loss": 1.0114,
+      "step": 4162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7098234002371313e-05,
+      "loss": 1.0034,
+      "step": 4163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705738808888518e-05,
+      "loss": 1.1612,
+      "step": 4164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701656816587924e-05,
+      "loss": 1.2265,
+      "step": 4165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6975774247898133e-05,
+      "loss": 1.2942,
+      "step": 4166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.693500634947741e-05,
+      "loss": 1.2067,
+      "step": 4167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6894264485143107e-05,
+      "loss": 1.1523,
+      "step": 4168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6853548669412233e-05,
+      "loss": 1.1097,
+      "step": 4169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6812858916792328e-05,
+      "loss": 1.1583,
+      "step": 4170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6772195241781805e-05,
+      "loss": 1.088,
+      "step": 4171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6731557658869666e-05,
+      "loss": 1.1223,
+      "step": 4172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669094618253565e-05,
+      "loss": 1.2614,
+      "step": 4173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6650360827250166e-05,
+      "loss": 1.067,
+      "step": 4174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6609801607474415e-05,
+      "loss": 1.0934,
+      "step": 4175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6569268537660163e-05,
+      "loss": 1.034,
+      "step": 4176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6528761632249965e-05,
+      "loss": 1.043,
+      "step": 4177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6488280905676965e-05,
+      "loss": 1.013,
+      "step": 4178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6447826372365037e-05,
+      "loss": 1.1152,
+      "step": 4179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6407398046728648e-05,
+      "loss": 1.1616,
+      "step": 4180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6366995943173057e-05,
+      "loss": 1.1169,
+      "step": 4181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6326620076094033e-05,
+      "loss": 1.0443,
+      "step": 4182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.628627045987817e-05,
+      "loss": 1.0118,
+      "step": 4183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6245947108902468e-05,
+      "loss": 1.0001,
+      "step": 4184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.62056500375348e-05,
+      "loss": 1.1085,
+      "step": 4185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6165379260133516e-05,
+      "loss": 1.1914,
+      "step": 4186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6125134791047733e-05,
+      "loss": 0.9617,
+      "step": 4187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6084916644617062e-05,
+      "loss": 1.1158,
+      "step": 4188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6044724835171874e-05,
+      "loss": 1.0361,
+      "step": 4189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6004559377032955e-05,
+      "loss": 1.0057,
+      "step": 4190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5964420284511936e-05,
+      "loss": 1.0281,
+      "step": 4191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5924307571910857e-05,
+      "loss": 1.1368,
+      "step": 4192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588422125352251e-05,
+      "loss": 1.1279,
+      "step": 4193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584416134363019e-05,
+      "loss": 1.1715,
+      "step": 4194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5804127856507796e-05,
+      "loss": 1.1862,
+      "step": 4195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5764120806419812e-05,
+      "loss": 1.1994,
+      "step": 4196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572414020762136e-05,
+      "loss": 1.1222,
+      "step": 4197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.568418607435803e-05,
+      "loss": 1.1184,
+      "step": 4198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5644258420866107e-05,
+      "loss": 1.1373,
+      "step": 4199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5604357261372335e-05,
+      "loss": 0.9658,
+      "step": 4200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5564482610094088e-05,
+      "loss": 1.1367,
+      "step": 4201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5524634481239197e-05,
+      "loss": 1.0883,
+      "step": 4202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5484812889006205e-05,
+      "loss": 0.8986,
+      "step": 4203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5445017847584028e-05,
+      "loss": 1.1434,
+      "step": 4204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5405249371152273e-05,
+      "loss": 1.1032,
+      "step": 4205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5365507473880957e-05,
+      "loss": 1.2246,
+      "step": 4206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5325792169930705e-05,
+      "loss": 0.9815,
+      "step": 4207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5286103473452583e-05,
+      "loss": 1.1005,
+      "step": 4208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5246441398588316e-05,
+      "loss": 1.176,
+      "step": 4209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5206805959469982e-05,
+      "loss": 1.0249,
+      "step": 4210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5167197170220314e-05,
+      "loss": 1.0971,
+      "step": 4211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5127615044952446e-05,
+      "loss": 1.0398,
+      "step": 4212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5088059597770062e-05,
+      "loss": 1.1619,
+      "step": 4213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.504853084276728e-05,
+      "loss": 1.2046,
+      "step": 4214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.500902879402881e-05,
+      "loss": 1.1076,
+      "step": 4215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496955346562978e-05,
+      "loss": 1.0954,
+      "step": 4216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4930104871635752e-05,
+      "loss": 1.0814,
+      "step": 4217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4890683026102878e-05,
+      "loss": 1.1664,
+      "step": 4218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4851287943077706e-05,
+      "loss": 1.0219,
+      "step": 4219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4811919636597214e-05,
+      "loss": 1.2661,
+      "step": 4220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4772578120688938e-05,
+      "loss": 1.1392,
+      "step": 4221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4733263409370798e-05,
+      "loss": 1.0788,
+      "step": 4222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4693975516651147e-05,
+      "loss": 1.277,
+      "step": 4223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4654714456528873e-05,
+      "loss": 1.0426,
+      "step": 4224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4615480242993217e-05,
+      "loss": 1.138,
+      "step": 4225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4576272890023844e-05,
+      "loss": 1.132,
+      "step": 4226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4537092411590945e-05,
+      "loss": 1.0454,
+      "step": 4227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4497938821655063e-05,
+      "loss": 0.9764,
+      "step": 4228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4458812134167132e-05,
+      "loss": 1.0024,
+      "step": 4229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4419712363068593e-05,
+      "loss": 1.1076,
+      "step": 4230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4380639522291237e-05,
+      "loss": 1.1181,
+      "step": 4231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434159362575722e-05,
+      "loss": 1.141,
+      "step": 4232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.430257468737921e-05,
+      "loss": 1.2729,
+      "step": 4233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4263582721060174e-05,
+      "loss": 1.11,
+      "step": 4234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4224617740693478e-05,
+      "loss": 1.1655,
+      "step": 4235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.418567976016294e-05,
+      "loss": 1.0366,
+      "step": 4236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.414676879334269e-05,
+      "loss": 1.0359,
+      "step": 4237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.410788485409724e-05,
+      "loss": 1.0018,
+      "step": 4238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4069027956281475e-05,
+      "loss": 1.0979,
+      "step": 4239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.40301981137407e-05,
+      "loss": 1.0648,
+      "step": 4240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3991395340310484e-05,
+      "loss": 1.1131,
+      "step": 4241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3952619649816864e-05,
+      "loss": 1.1267,
+      "step": 4242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.391387105607612e-05,
+      "loss": 1.0809,
+      "step": 4243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3875149572894927e-05,
+      "loss": 1.1709,
+      "step": 4244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3836455214070276e-05,
+      "loss": 1.2498,
+      "step": 4245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3797787993389563e-05,
+      "loss": 1.1342,
+      "step": 4246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375914792463041e-05,
+      "loss": 1.1245,
+      "step": 4247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3720535021560865e-05,
+      "loss": 1.2619,
+      "step": 4248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3681949297939233e-05,
+      "loss": 1.2297,
+      "step": 4249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364339076751414e-05,
+      "loss": 1.1778,
+      "step": 4250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360485944402452e-05,
+      "loss": 1.0304,
+      "step": 4251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3566355341199663e-05,
+      "loss": 1.0399,
+      "step": 4252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3527878472759078e-05,
+      "loss": 1.0736,
+      "step": 4253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3489428852412676e-05,
+      "loss": 1.1242,
+      "step": 4254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.345100649386056e-05,
+      "loss": 1.1274,
+      "step": 4255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341261141079316e-05,
+      "loss": 1.2116,
+      "step": 4256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3374243616891166e-05,
+      "loss": 1.0445,
+      "step": 4257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333590312582561e-05,
+      "loss": 1.1256,
+      "step": 4258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3297589951257694e-05,
+      "loss": 1.0186,
+      "step": 4259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.325930410683903e-05,
+      "loss": 1.121,
+      "step": 4260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322104560621129e-05,
+      "loss": 1.0446,
+      "step": 4261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3182814463006596e-05,
+      "loss": 1.052,
+      "step": 4262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3144610690847203e-05,
+      "loss": 1.1521,
+      "step": 4263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3106434303345683e-05,
+      "loss": 1.1353,
+      "step": 4264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3068285314104787e-05,
+      "loss": 1.1106,
+      "step": 4265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3030163736717613e-05,
+      "loss": 1.2656,
+      "step": 4266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2992069584767306e-05,
+      "loss": 1.2125,
+      "step": 4267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2954002871827417e-05,
+      "loss": 1.0285,
+      "step": 4268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.291596361146162e-05,
+      "loss": 1.0393,
+      "step": 4269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2877951817223896e-05,
+      "loss": 1.1407,
+      "step": 4270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2839967502658334e-05,
+      "loss": 1.0589,
+      "step": 4271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280201068129929e-05,
+      "loss": 1.1144,
+      "step": 4272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2764081366671285e-05,
+      "loss": 1.0612,
+      "step": 4273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272617957228913e-05,
+      "loss": 0.9605,
+      "step": 4274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268830531165771e-05,
+      "loss": 1.1535,
+      "step": 4275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2650458598272206e-05,
+      "loss": 1.1314,
+      "step": 4276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2612639445617912e-05,
+      "loss": 1.1242,
+      "step": 4277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2574847867170334e-05,
+      "loss": 1.004,
+      "step": 4278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2537083876395105e-05,
+      "loss": 1.1339,
+      "step": 4279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2499347486748112e-05,
+      "loss": 1.1065,
+      "step": 4280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2461638711675337e-05,
+      "loss": 1.1869,
+      "step": 4281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2423957564613006e-05,
+      "loss": 1.0223,
+      "step": 4282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.238630405898734e-05,
+      "loss": 1.0554,
+      "step": 4283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2348678208214903e-05,
+      "loss": 1.1672,
+      "step": 4284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2311080025702235e-05,
+      "loss": 1.2218,
+      "step": 4285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2273509524846192e-05,
+      "loss": 1.0779,
+      "step": 4286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2235966719033586e-05,
+      "loss": 1.0472,
+      "step": 4287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2198451621641546e-05,
+      "loss": 1.0488,
+      "step": 4288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.216096424603711e-05,
+      "loss": 1.1172,
+      "step": 4289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212350460557765e-05,
+      "loss": 1.074,
+      "step": 4290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2086072713610505e-05,
+      "loss": 1.1081,
+      "step": 4291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2048668583473232e-05,
+      "loss": 1.0054,
+      "step": 4292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.201129222849344e-05,
+      "loss": 1.0691,
+      "step": 4293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.197394366198884e-05,
+      "loss": 1.0983,
+      "step": 4294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193662289726721e-05,
+      "loss": 1.1393,
+      "step": 4295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1899329947626544e-05,
+      "loss": 1.0063,
+      "step": 4296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186206482635479e-05,
+      "loss": 0.9719,
+      "step": 4297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1824827546730066e-05,
+      "loss": 1.1626,
+      "step": 4298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1787618122020538e-05,
+      "loss": 1.0711,
+      "step": 4299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1750436565484433e-05,
+      "loss": 1.0344,
+      "step": 4300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1713282890370034e-05,
+      "loss": 1.0874,
+      "step": 4301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1676157109915786e-05,
+      "loss": 1.1651,
+      "step": 4302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163905923735007e-05,
+      "loss": 1.1628,
+      "step": 4303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1601989285891456e-05,
+      "loss": 0.9838,
+      "step": 4304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156494726874838e-05,
+      "loss": 1.0984,
+      "step": 4305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152793319911952e-05,
+      "loss": 1.1336,
+      "step": 4306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1490947090193457e-05,
+      "loss": 1.0897,
+      "step": 4307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.145398895514892e-05,
+      "loss": 1.0782,
+      "step": 4308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1417058807154555e-05,
+      "loss": 1.1334,
+      "step": 4309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1380156659369178e-05,
+      "loss": 1.1153,
+      "step": 4310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1343282524941422e-05,
+      "loss": 1.1319,
+      "step": 4311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1306436417010168e-05,
+      "loss": 1.1638,
+      "step": 4312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1269618348704125e-05,
+      "loss": 1.0933,
+      "step": 4313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1232828333142152e-05,
+      "loss": 1.1332,
+      "step": 4314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1196066383433024e-05,
+      "loss": 1.0257,
+      "step": 4315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1159332512675534e-05,
+      "loss": 1.0487,
+      "step": 4316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112262673395847e-05,
+      "loss": 1.0834,
+      "step": 4317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1085949060360654e-05,
+      "loss": 1.0034,
+      "step": 4318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1049299504950803e-05,
+      "loss": 1.1796,
+      "step": 4319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.101267808078774e-05,
+      "loss": 1.1904,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.097608480092016e-05,
+      "loss": 1.1033,
+      "step": 4321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0939519678386753e-05,
+      "loss": 1.2439,
+      "step": 4322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090298272621617e-05,
+      "loss": 0.9343,
+      "step": 4323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.086647395742709e-05,
+      "loss": 1.1613,
+      "step": 4324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082999338502806e-05,
+      "loss": 1.0808,
+      "step": 4325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0793541022017692e-05,
+      "loss": 1.1723,
+      "step": 4326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0757116881384374e-05,
+      "loss": 1.1608,
+      "step": 4327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.072072097610661e-05,
+      "loss": 1.1708,
+      "step": 4328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0684353319152737e-05,
+      "loss": 0.9553,
+      "step": 4329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0648013923481115e-05,
+      "loss": 1.0479,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0611702802039912e-05,
+      "loss": 1.2434,
+      "step": 4331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0575419967767385e-05,
+      "loss": 1.1878,
+      "step": 4332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0539165433591513e-05,
+      "loss": 1.0348,
+      "step": 4333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.050293921243036e-05,
+      "loss": 1.1361,
+      "step": 4334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0466741317191816e-05,
+      "loss": 1.279,
+      "step": 4335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0430571760773742e-05,
+      "loss": 1.0074,
+      "step": 4336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0394430556063816e-05,
+      "loss": 1.0559,
+      "step": 4337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035831771593968e-05,
+      "loss": 1.1377,
+      "step": 4338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0322233253268818e-05,
+      "loss": 0.9873,
+      "step": 4339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0286177180908673e-05,
+      "loss": 1.0825,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0250149511706483e-05,
+      "loss": 1.1851,
+      "step": 4341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0214150258499487e-05,
+      "loss": 1.0325,
+      "step": 4342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0178179434114674e-05,
+      "loss": 1.08,
+      "step": 4343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0142237051368963e-05,
+      "loss": 1.1093,
+      "step": 4344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0106323123069104e-05,
+      "loss": 1.0713,
+      "step": 4345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0070437662011798e-05,
+      "loss": 1.1025,
+      "step": 4346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0034580680983473e-05,
+      "loss": 1.0426,
+      "step": 4347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999875219276054e-05,
+      "loss": 0.8875,
+      "step": 4348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9962952210109166e-05,
+      "loss": 1.1241,
+      "step": 4349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992718074578539e-05,
+      "loss": 1.2039,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9891437812535052e-05,
+      "loss": 1.0333,
+      "step": 4351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.985572342309393e-05,
+      "loss": 1.1065,
+      "step": 4352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98200375901875e-05,
+      "loss": 0.9877,
+      "step": 4353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9784380326531183e-05,
+      "loss": 1.0509,
+      "step": 4354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9748751644830142e-05,
+      "loss": 1.091,
+      "step": 4355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9713151557779374e-05,
+      "loss": 1.018,
+      "step": 4356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9677580078063662e-05,
+      "loss": 1.0932,
+      "step": 4357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9642037218357688e-05,
+      "loss": 1.0799,
+      "step": 4358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960652299132585e-05,
+      "loss": 0.9539,
+      "step": 4359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.957103740962232e-05,
+      "loss": 1.2344,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9535580485891182e-05,
+      "loss": 1.0193,
+      "step": 4361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95001522327662e-05,
+      "loss": 1.1512,
+      "step": 4362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9464752662870945e-05,
+      "loss": 1.0773,
+      "step": 4363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9429381788818824e-05,
+      "loss": 1.1246,
+      "step": 4364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9394039623212945e-05,
+      "loss": 1.1288,
+      "step": 4365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9358726178646224e-05,
+      "loss": 1.1292,
+      "step": 4366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9323441467701352e-05,
+      "loss": 1.2633,
+      "step": 4367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9288185502950775e-05,
+      "loss": 1.0899,
+      "step": 4368
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9252958296956648e-05,
+      "loss": 1.032,
+      "step": 4369
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9217759862270977e-05,
+      "loss": 1.0682,
+      "step": 4370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9182590211435423e-05,
+      "loss": 1.0267,
+      "step": 4371
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914744935698143e-05,
+      "loss": 1.09,
+      "step": 4372
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112337311430194e-05,
+      "loss": 1.1402,
+      "step": 4373
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.907725408729263e-05,
+      "loss": 1.1302,
+      "step": 4374
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.904219969706935e-05,
+      "loss": 1.077,
+      "step": 4375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9007174153250797e-05,
+      "loss": 1.0526,
+      "step": 4376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.897217746831701e-05,
+      "loss": 1.1469,
+      "step": 4377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8937209654737796e-05,
+      "loss": 1.0512,
+      "step": 4378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8902270724972726e-05,
+      "loss": 0.9796,
+      "step": 4379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8867360691471002e-05,
+      "loss": 1.0097,
+      "step": 4380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.883247956667157e-05,
+      "loss": 1.1034,
+      "step": 4381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8797627363003022e-05,
+      "loss": 1.1056,
+      "step": 4382
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8762804092883766e-05,
+      "loss": 1.0854,
+      "step": 4383
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8728009768721765e-05,
+      "loss": 1.0703,
+      "step": 4384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.869324440291477e-05,
+      "loss": 1.0318,
+      "step": 4385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8658508007850138e-05,
+      "loss": 0.9658,
+      "step": 4386
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.862380059590495e-05,
+      "loss": 0.9022,
+      "step": 4387
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8589122179445917e-05,
+      "loss": 1.2719,
+      "step": 4388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.85544727708295e-05,
+      "loss": 1.0378,
+      "step": 4389
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8519852382401715e-05,
+      "loss": 1.1178,
+      "step": 4390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8485261026498356e-05,
+      "loss": 1.2466,
+      "step": 4391
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845069871544477e-05,
+      "loss": 0.9731,
+      "step": 4392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.841616546155601e-05,
+      "loss": 1.0814,
+      "step": 4393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838166127713672e-05,
+      "loss": 1.1395,
+      "step": 4394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.834718617448129e-05,
+      "loss": 1.097,
+      "step": 4395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.831274016587362e-05,
+      "loss": 1.1164,
+      "step": 4396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8278323263587404e-05,
+      "loss": 1.0972,
+      "step": 4397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8243935479885753e-05,
+      "loss": 1.2387,
+      "step": 4398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82095768270216e-05,
+      "loss": 1.2001,
+      "step": 4399
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8175247317237366e-05,
+      "loss": 1.1527,
+      "step": 4400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8140946962765194e-05,
+      "loss": 1.0786,
+      "step": 4401
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810667577582672e-05,
+      "loss": 1.1728,
+      "step": 4402
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8072433768633333e-05,
+      "loss": 1.026,
+      "step": 4403
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8038220953385853e-05,
+      "loss": 1.2197,
+      "step": 4404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.800403734227485e-05,
+      "loss": 1.1577,
+      "step": 4405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7969882947480375e-05,
+      "loss": 1.1287,
+      "step": 4406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.793575778117218e-05,
+      "loss": 1.1311,
+      "step": 4407
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.790166185550951e-05,
+      "loss": 1.0482,
+      "step": 4408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7867595182641226e-05,
+      "loss": 1.0689,
+      "step": 4409
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7833557774705733e-05,
+      "loss": 0.997,
+      "step": 4410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7799549643831104e-05,
+      "loss": 1.0839,
+      "step": 4411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7765570802134844e-05,
+      "loss": 1.0842,
+      "step": 4412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7731621261724164e-05,
+      "loss": 1.092,
+      "step": 4413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7697701034695724e-05,
+      "loss": 1.0757,
+      "step": 4414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7663810133135784e-05,
+      "loss": 1.161,
+      "step": 4415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7629948569120126e-05,
+      "loss": 1.022,
+      "step": 4416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7596116354714155e-05,
+      "loss": 1.1229,
+      "step": 4417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7562313501972692e-05,
+      "loss": 0.9922,
+      "step": 4418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7528540022940288e-05,
+      "loss": 1.0911,
+      "step": 4419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7494795929650766e-05,
+      "loss": 0.9982,
+      "step": 4420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.746108123412773e-05,
+      "loss": 1.1743,
+      "step": 4421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7427395948384117e-05,
+      "loss": 0.998,
+      "step": 4422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.739374008442256e-05,
+      "loss": 1.0741,
+      "step": 4423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7360113654235034e-05,
+      "loss": 1.0517,
+      "step": 4424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7326516669803193e-05,
+      "loss": 1.1076,
+      "step": 4425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7292949143098026e-05,
+      "loss": 1.0897,
+      "step": 4426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.725941108608019e-05,
+      "loss": 1.0976,
+      "step": 4427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7225902510699697e-05,
+      "loss": 1.1106,
+      "step": 4428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7192423428896198e-05,
+      "loss": 1.1403,
+      "step": 4429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7158973852598725e-05,
+      "loss": 1.1808,
+      "step": 4430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7125553793725836e-05,
+      "loss": 1.1345,
+      "step": 4431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7092163264185545e-05,
+      "loss": 1.0442,
+      "step": 4432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7058802275875408e-05,
+      "loss": 1.0614,
+      "step": 4433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025470840682378e-05,
+      "loss": 1.0956,
+      "step": 4434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699216897048297e-05,
+      "loss": 1.0534,
+      "step": 4435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6958896677143065e-05,
+      "loss": 1.1523,
+      "step": 4436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.692565397251805e-05,
+      "loss": 1.0553,
+      "step": 4437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6892440868452764e-05,
+      "loss": 0.9781,
+      "step": 4438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6859257376781545e-05,
+      "loss": 0.9909,
+      "step": 4439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6826103509328083e-05,
+      "loss": 0.9957,
+      "step": 4440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.679297927790565e-05,
+      "loss": 1.0365,
+      "step": 4441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6759884694316774e-05,
+      "loss": 1.0476,
+      "step": 4442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6726819770353585e-05,
+      "loss": 1.0473,
+      "step": 4443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.669378451779755e-05,
+      "loss": 1.0801,
+      "step": 4444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.666077894841964e-05,
+      "loss": 1.1323,
+      "step": 4445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.662780307398014e-05,
+      "loss": 1.0239,
+      "step": 4446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6594856906228918e-05,
+      "loss": 1.0349,
+      "step": 4447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6561940456905033e-05,
+      "loss": 1.1051,
+      "step": 4448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6529053737737164e-05,
+      "loss": 1.0683,
+      "step": 4449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.649619676044327e-05,
+      "loss": 0.957,
+      "step": 4450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.64633695367308e-05,
+      "loss": 1.1026,
+      "step": 4451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6430572078296525e-05,
+      "loss": 0.9712,
+      "step": 4452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6397804396826643e-05,
+      "loss": 1.0689,
+      "step": 4453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636506650399673e-05,
+      "loss": 1.2022,
+      "step": 4454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6332358411471792e-05,
+      "loss": 0.916,
+      "step": 4455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6299680130906138e-05,
+      "loss": 1.0738,
+      "step": 4456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6267031673943543e-05,
+      "loss": 1.0925,
+      "step": 4457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.62344130522171e-05,
+      "loss": 1.1663,
+      "step": 4458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6201824277349277e-05,
+      "loss": 1.0289,
+      "step": 4459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.616926536095189e-05,
+      "loss": 1.0351,
+      "step": 4460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.613673631462619e-05,
+      "loss": 1.2434,
+      "step": 4461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6104237149962686e-05,
+      "loss": 1.1046,
+      "step": 4462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6071767878541354e-05,
+      "loss": 1.1878,
+      "step": 4463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6039328511931362e-05,
+      "loss": 1.1485,
+      "step": 4464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6006919061691384e-05,
+      "loss": 1.099,
+      "step": 4465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5974539539369328e-05,
+      "loss": 1.0728,
+      "step": 4466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5942189956502497e-05,
+      "loss": 1.1709,
+      "step": 4467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5909870324617472e-05,
+      "loss": 0.96,
+      "step": 4468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587758065523025e-05,
+      "loss": 1.0229,
+      "step": 4469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5845320959846023e-05,
+      "loss": 1.0554,
+      "step": 4470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5813091249959434e-05,
+      "loss": 1.0018,
+      "step": 4471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.578089153705433e-05,
+      "loss": 1.1365,
+      "step": 4472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5748721832603973e-05,
+      "loss": 1.1171,
+      "step": 4473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.571658214807087e-05,
+      "loss": 1.0844,
+      "step": 4474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.568447249490682e-05,
+      "loss": 1.1338,
+      "step": 4475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5652392884552947e-05,
+      "loss": 1.0958,
+      "step": 4476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5620343328439703e-05,
+      "loss": 1.1146,
+      "step": 4477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.558832383798674e-05,
+      "loss": 1.0593,
+      "step": 4478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5556334424603114e-05,
+      "loss": 1.1555,
+      "step": 4479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5524375099687072e-05,
+      "loss": 0.8507,
+      "step": 4480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.549244587462618e-05,
+      "loss": 1.1032,
+      "step": 4481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5460546760797236e-05,
+      "loss": 0.9478,
+      "step": 4482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.54286777695664e-05,
+      "loss": 1.1436,
+      "step": 4483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5396838912289e-05,
+      "loss": 1.2092,
+      "step": 4484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5365030200309727e-05,
+      "loss": 1.0472,
+      "step": 4485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5333251644962376e-05,
+      "loss": 1.0801,
+      "step": 4486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5301503257570184e-05,
+      "loss": 1.1871,
+      "step": 4487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5269785049445484e-05,
+      "loss": 1.0638,
+      "step": 4488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5238097031889987e-05,
+      "loss": 1.0299,
+      "step": 4489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5206439216194513e-05,
+      "loss": 1.0426,
+      "step": 4490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5174811613639262e-05,
+      "loss": 1.0361,
+      "step": 4491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5143214235493497e-05,
+      "loss": 1.1369,
+      "step": 4492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5111647093015901e-05,
+      "loss": 1.157,
+      "step": 4493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5080110197454223e-05,
+      "loss": 1.1365,
+      "step": 4494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5048603560045549e-05,
+      "loss": 1.0744,
+      "step": 4495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5017127192016134e-05,
+      "loss": 1.0166,
+      "step": 4496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4985681104581428e-05,
+      "loss": 1.1754,
+      "step": 4497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4954265308946114e-05,
+      "loss": 1.0248,
+      "step": 4498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4922879816304113e-05,
+      "loss": 1.0108,
+      "step": 4499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4891524637838473e-05,
+      "loss": 1.1446,
+      "step": 4500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.486019978472154e-05,
+      "loss": 1.0096,
+      "step": 4501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.482890526811478e-05,
+      "loss": 1.2132,
+      "step": 4502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4797641099168858e-05,
+      "loss": 1.0349,
+      "step": 4503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4766407289023609e-05,
+      "loss": 1.015,
+      "step": 4504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4735203848808155e-05,
+      "loss": 1.1787,
+      "step": 4505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4704030789640643e-05,
+      "loss": 1.1605,
+      "step": 4506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4672888122628536e-05,
+      "loss": 1.1635,
+      "step": 4507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4641775858868379e-05,
+      "loss": 1.1243,
+      "step": 4508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4610694009445914e-05,
+      "loss": 1.0431,
+      "step": 4509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4579642585436015e-05,
+      "loss": 1.2482,
+      "step": 4510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4548621597902779e-05,
+      "loss": 1.1795,
+      "step": 4511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4517631057899394e-05,
+      "loss": 1.1418,
+      "step": 4512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4486670976468253e-05,
+      "loss": 1.1386,
+      "step": 4513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4455741364640862e-05,
+      "loss": 1.0527,
+      "step": 4514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4424842233437863e-05,
+      "loss": 1.2785,
+      "step": 4515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4393973593869037e-05,
+      "loss": 0.8626,
+      "step": 4516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4363135456933364e-05,
+      "loss": 1.1076,
+      "step": 4517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4332327833618875e-05,
+      "loss": 1.109,
+      "step": 4518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4301550734902736e-05,
+      "loss": 1.1147,
+      "step": 4519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4270804171751317e-05,
+      "loss": 0.9804,
+      "step": 4520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4240088155120023e-05,
+      "loss": 1.1061,
+      "step": 4521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4209402695953388e-05,
+      "loss": 1.1665,
+      "step": 4522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.417874780518511e-05,
+      "loss": 1.1233,
+      "step": 4523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.414812349373793e-05,
+      "loss": 1.1522,
+      "step": 4524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4117529772523707e-05,
+      "loss": 1.0337,
+      "step": 4525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4086966652443468e-05,
+      "loss": 0.9545,
+      "step": 4526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4056434144387253e-05,
+      "loss": 1.1805,
+      "step": 4527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4025932259234198e-05,
+      "loss": 1.0216,
+      "step": 4528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3995461007852605e-05,
+      "loss": 1.0598,
+      "step": 4529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3965020401099781e-05,
+      "loss": 1.0633,
+      "step": 4530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3934610449822128e-05,
+      "loss": 1.0314,
+      "step": 4531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3904231164855186e-05,
+      "loss": 1.1477,
+      "step": 4532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3873882557023488e-05,
+      "loss": 1.0451,
+      "step": 4533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.384356463714066e-05,
+      "loss": 1.2491,
+      "step": 4534
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3813277416009452e-05,
+      "loss": 1.2526,
+      "step": 4535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3783020904421584e-05,
+      "loss": 1.0263,
+      "step": 4536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3752795113157879e-05,
+      "loss": 0.973,
+      "step": 4537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3722600052988243e-05,
+      "loss": 1.0481,
+      "step": 4538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3692435734671583e-05,
+      "loss": 1.1587,
+      "step": 4539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3662302168955877e-05,
+      "loss": 1.1702,
+      "step": 4540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3632199366578103e-05,
+      "loss": 1.2457,
+      "step": 4541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.360212733826436e-05,
+      "loss": 1.106,
+      "step": 4542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3572086094729709e-05,
+      "loss": 1.1156,
+      "step": 4543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3542075646678286e-05,
+      "loss": 1.1323,
+      "step": 4544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3512096004803232e-05,
+      "loss": 1.1451,
+      "step": 4545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.348214717978672e-05,
+      "loss": 0.9775,
+      "step": 4546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3452229182299892e-05,
+      "loss": 1.1475,
+      "step": 4547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3422342023003021e-05,
+      "loss": 1.0424,
+      "step": 4548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3392485712545266e-05,
+      "loss": 0.9339,
+      "step": 4549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3362660261564908e-05,
+      "loss": 1.1266,
+      "step": 4550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3332865680689132e-05,
+      "loss": 1.023,
+      "step": 4551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3303101980534184e-05,
+      "loss": 1.2101,
+      "step": 4552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3273369171705263e-05,
+      "loss": 1.0938,
+      "step": 4553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3243667264796633e-05,
+      "loss": 1.1656,
+      "step": 4554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3213996270391448e-05,
+      "loss": 1.0657,
+      "step": 4555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3184356199061965e-05,
+      "loss": 1.1322,
+      "step": 4556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.315474706136931e-05,
+      "loss": 0.9879,
+      "step": 4557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3125168867863668e-05,
+      "loss": 1.0406,
+      "step": 4558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3095621629084109e-05,
+      "loss": 0.9611,
+      "step": 4559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306610535555881e-05,
+      "loss": 1.008,
+      "step": 4560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303662005780476e-05,
+      "loss": 1.2059,
+      "step": 4561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3007165746328064e-05,
+      "loss": 0.9282,
+      "step": 4562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297774243162363e-05,
+      "loss": 1.1274,
+      "step": 4563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2948350124175456e-05,
+      "loss": 1.0291,
+      "step": 4564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2918988834456392e-05,
+      "loss": 1.0925,
+      "step": 4565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.288965857292832e-05,
+      "loss": 1.0822,
+      "step": 4566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2860359350041984e-05,
+      "loss": 0.9658,
+      "step": 4567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2831091176237175e-05,
+      "loss": 1.1759,
+      "step": 4568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2801854061942475e-05,
+      "loss": 1.1405,
+      "step": 4569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2772648017575539e-05,
+      "loss": 1.1029,
+      "step": 4570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2743473053542842e-05,
+      "loss": 1.2343,
+      "step": 4571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2714329180239892e-05,
+      "loss": 1.1688,
+      "step": 4572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2685216408051026e-05,
+      "loss": 0.9445,
+      "step": 4573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2656134747349557e-05,
+      "loss": 1.0808,
+      "step": 4574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2627084208497642e-05,
+      "loss": 1.1188,
+      "step": 4575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2598064801846454e-05,
+      "loss": 1.0032,
+      "step": 4576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2569076537735969e-05,
+      "loss": 1.0894,
+      "step": 4577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2540119426495156e-05,
+      "loss": 1.1801,
+      "step": 4578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.251119347844183e-05,
+      "loss": 1.2062,
+      "step": 4579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.24822987038827e-05,
+      "loss": 1.2462,
+      "step": 4580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2453435113113377e-05,
+      "loss": 1.1142,
+      "step": 4581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2424602716418398e-05,
+      "loss": 1.0634,
+      "step": 4582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2395801524071104e-05,
+      "loss": 0.9893,
+      "step": 4583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2367031546333829e-05,
+      "loss": 1.0788,
+      "step": 4584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.233829279345765e-05,
+      "loss": 1.1759,
+      "step": 4585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2309585275682645e-05,
+      "loss": 1.1992,
+      "step": 4586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2280909003237662e-05,
+      "loss": 1.1148,
+      "step": 4587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2252263986340495e-05,
+      "loss": 1.0865,
+      "step": 4588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2223650235197747e-05,
+      "loss": 1.204,
+      "step": 4589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2195067760004953e-05,
+      "loss": 1.0609,
+      "step": 4590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2166516570946363e-05,
+      "loss": 0.9064,
+      "step": 4591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213799667819524e-05,
+      "loss": 1.1236,
+      "step": 4592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2109508091913568e-05,
+      "loss": 1.1401,
+      "step": 4593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2081050822252294e-05,
+      "loss": 1.2443,
+      "step": 4594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2052624879351104e-05,
+      "loss": 1.1383,
+      "step": 4595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2024230273338567e-05,
+      "loss": 1.0192,
+      "step": 4596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1995867014332063e-05,
+      "loss": 1.0027,
+      "step": 4597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1967535112437855e-05,
+      "loss": 1.0695,
+      "step": 4598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1939234577750979e-05,
+      "loss": 1.0195,
+      "step": 4599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1910965420355324e-05,
+      "loss": 1.0874,
+      "step": 4600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.18827276503236e-05,
+      "loss": 1.1043,
+      "step": 4601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185452127771729e-05,
+      "loss": 1.0545,
+      "step": 4602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1826346312586733e-05,
+      "loss": 1.0288,
+      "step": 4603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.179820276497109e-05,
+      "loss": 1.0004,
+      "step": 4604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1770090644898257e-05,
+      "loss": 1.1198,
+      "step": 4605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1742009962385048e-05,
+      "loss": 1.0906,
+      "step": 4606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1713960727436924e-05,
+      "loss": 1.0811,
+      "step": 4607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.168594295004829e-05,
+      "loss": 1.2069,
+      "step": 4608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1657956640202217e-05,
+      "loss": 1.0471,
+      "step": 4609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1630001807870684e-05,
+      "loss": 0.9602,
+      "step": 4610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1602078463014332e-05,
+      "loss": 1.079,
+      "step": 4611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1574186615582706e-05,
+      "loss": 1.0714,
+      "step": 4612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1546326275513996e-05,
+      "loss": 1.2122,
+      "step": 4613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1518497452735277e-05,
+      "loss": 1.1043,
+      "step": 4614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1490700157162327e-05,
+      "loss": 1.122,
+      "step": 4615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1462934398699743e-05,
+      "loss": 0.9868,
+      "step": 4616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1435200187240836e-05,
+      "loss": 1.1501,
+      "step": 4617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.140749753266771e-05,
+      "loss": 0.9767,
+      "step": 4618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1379826444851172e-05,
+      "loss": 1.1661,
+      "step": 4619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1352186933650866e-05,
+      "loss": 1.0787,
+      "step": 4620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1324579008915104e-05,
+      "loss": 1.0201,
+      "step": 4621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1297002680481017e-05,
+      "loss": 1.1781,
+      "step": 4622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1269457958174412e-05,
+      "loss": 1.0644,
+      "step": 4623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1241944851809883e-05,
+      "loss": 0.9305,
+      "step": 4624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1214463371190686e-05,
+      "loss": 1.0575,
+      "step": 4625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1187013526108925e-05,
+      "loss": 1.0814,
+      "step": 4626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1159595326345318e-05,
+      "loss": 1.0631,
+      "step": 4627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1132208781669417e-05,
+      "loss": 1.1386,
+      "step": 4628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.110485390183934e-05,
+      "loss": 1.1486,
+      "step": 4629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10775306966021e-05,
+      "loss": 1.0878,
+      "step": 4630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1050239175693288e-05,
+      "loss": 1.0271,
+      "step": 4631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1022979348837292e-05,
+      "loss": 1.0401,
+      "step": 4632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0995751225747164e-05,
+      "loss": 1.1617,
+      "step": 4633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0968554816124653e-05,
+      "loss": 1.1252,
+      "step": 4634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0941390129660211e-05,
+      "loss": 0.9315,
+      "step": 4635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0914257176033049e-05,
+      "loss": 1.1999,
+      "step": 4636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0887155964910967e-05,
+      "loss": 1.0759,
+      "step": 4637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0860086505950551e-05,
+      "loss": 1.2932,
+      "step": 4638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0833048808797008e-05,
+      "loss": 1.0347,
+      "step": 4639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0806042883084255e-05,
+      "loss": 1.1041,
+      "step": 4640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.077906873843486e-05,
+      "loss": 1.0715,
+      "step": 4641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0752126384460138e-05,
+      "loss": 1.0909,
+      "step": 4642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0725215830759982e-05,
+      "loss": 1.1472,
+      "step": 4643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0698337086923038e-05,
+      "loss": 1.1227,
+      "step": 4644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0671490162526577e-05,
+      "loss": 1.0824,
+      "step": 4645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0644675067136522e-05,
+      "loss": 1.0921,
+      "step": 4646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0617891810307456e-05,
+      "loss": 0.9835,
+      "step": 4647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0591140401582666e-05,
+      "loss": 1.1538,
+      "step": 4648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0564420850494017e-05,
+      "loss": 1.0945,
+      "step": 4649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0537733166562103e-05,
+      "loss": 1.1268,
+      "step": 4650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0511077359296096e-05,
+      "loss": 1.1565,
+      "step": 4651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0484453438193843e-05,
+      "loss": 1.0317,
+      "step": 4652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045786141274181e-05,
+      "loss": 1.085,
+      "step": 4653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.043130129241513e-05,
+      "loss": 1.1702,
+      "step": 4654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0404773086677532e-05,
+      "loss": 1.1101,
+      "step": 4655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0378276804981391e-05,
+      "loss": 1.0383,
+      "step": 4656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0351812456767718e-05,
+      "loss": 1.2931,
+      "step": 4657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0325380051466126e-05,
+      "loss": 1.0755,
+      "step": 4658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.029897959849484e-05,
+      "loss": 1.1735,
+      "step": 4659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0272611107260743e-05,
+      "loss": 1.0943,
+      "step": 4660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0246274587159278e-05,
+      "loss": 1.0228,
+      "step": 4661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0219970047574512e-05,
+      "loss": 1.1152,
+      "step": 4662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.019369749787914e-05,
+      "loss": 1.1612,
+      "step": 4663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.016745694743444e-05,
+      "loss": 1.071,
+      "step": 4664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0141248405590275e-05,
+      "loss": 1.0229,
+      "step": 4665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0115071881685134e-05,
+      "loss": 1.0829,
+      "step": 4666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0088927385046077e-05,
+      "loss": 1.1291,
+      "step": 4667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006281492498874e-05,
+      "loss": 1.1198,
+      "step": 4668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0036734510817391e-05,
+      "loss": 0.9113,
+      "step": 4669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0010686151824832e-05,
+      "loss": 1.0496,
+      "step": 4670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.98466985729245e-06,
+      "loss": 1.1001,
+      "step": 4671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.958685636490239e-06,
+      "loss": 1.0645,
+      "step": 4672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.932733498676727e-06,
+      "loss": 1.0529,
+      "step": 4673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.906813453099018e-06,
+      "loss": 0.9616,
+      "step": 4674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.880925508992822e-06,
+      "loss": 0.981,
+      "step": 4675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.855069675582362e-06,
+      "loss": 1.0335,
+      "step": 4676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.829245962080425e-06,
+      "loss": 1.0724,
+      "step": 4677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.803454377688359e-06,
+      "loss": 1.1379,
+      "step": 4678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.777694931596093e-06,
+      "loss": 1.0464,
+      "step": 4679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.751967632982062e-06,
+      "loss": 1.1679,
+      "step": 4680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.726272491013277e-06,
+      "loss": 1.024,
+      "step": 4681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.700609514845282e-06,
+      "loss": 1.1703,
+      "step": 4682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.67497871362214e-06,
+      "loss": 1.1177,
+      "step": 4683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.649380096476468e-06,
+      "loss": 1.1347,
+      "step": 4684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.623813672529435e-06,
+      "loss": 1.1728,
+      "step": 4685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59827945089068e-06,
+      "loss": 1.2297,
+      "step": 4686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.572777440658453e-06,
+      "loss": 1.1337,
+      "step": 4687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.547307650919456e-06,
+      "loss": 1.0346,
+      "step": 4688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.52187009074893e-06,
+      "loss": 1.1072,
+      "step": 4689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.496464769210622e-06,
+      "loss": 1.1352,
+      "step": 4690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471091695356848e-06,
+      "loss": 1.0404,
+      "step": 4691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.445750878228354e-06,
+      "loss": 1.0812,
+      "step": 4692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.420442326854451e-06,
+      "loss": 1.0816,
+      "step": 4693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.395166050252935e-06,
+      "loss": 1.0467,
+      "step": 4694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.36992205743008e-06,
+      "loss": 1.1059,
+      "step": 4695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.344710357380671e-06,
+      "loss": 1.1982,
+      "step": 4696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.31953095908803e-06,
+      "loss": 0.9536,
+      "step": 4697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.294383871523882e-06,
+      "loss": 1.1094,
+      "step": 4698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.269269103648558e-06,
+      "loss": 1.1066,
+      "step": 4699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.244186664410715e-06,
+      "loss": 0.982,
+      "step": 4700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.219136562747655e-06,
+      "loss": 1.278,
+      "step": 4701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.194118807585029e-06,
+      "loss": 1.0392,
+      "step": 4702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.169133407837049e-06,
+      "loss": 1.2486,
+      "step": 4703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.144180372406341e-06,
+      "loss": 0.9807,
+      "step": 4704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.11925971018408e-06,
+      "loss": 1.0851,
+      "step": 4705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.094371430049764e-06,
+      "loss": 0.9396,
+      "step": 4706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.069515540871499e-06,
+      "loss": 1.0928,
+      "step": 4707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.044692051505744e-06,
+      "loss": 1.0755,
+      "step": 4708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.019900970797502e-06,
+      "loss": 1.1545,
+      "step": 4709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.995142307580162e-06,
+      "loss": 1.0463,
+      "step": 4710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.970416070675603e-06,
+      "loss": 1.2609,
+      "step": 4711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.945722268894085e-06,
+      "loss": 1.1596,
+      "step": 4712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.921060911034407e-06,
+      "loss": 1.0385,
+      "step": 4713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.896432005883726e-06,
+      "loss": 1.1232,
+      "step": 4714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.87183556221769e-06,
+      "loss": 1.0386,
+      "step": 4715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.847271588800354e-06,
+      "loss": 1.1724,
+      "step": 4716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.822740094384208e-06,
+      "loss": 1.0903,
+      "step": 4717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.798241087710147e-06,
+      "loss": 1.1044,
+      "step": 4718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773774577507542e-06,
+      "loss": 1.1522,
+      "step": 4719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.749340572494125e-06,
+      "loss": 1.0552,
+      "step": 4720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.724939081376127e-06,
+      "loss": 1.0534,
+      "step": 4721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.700570112848083e-06,
+      "loss": 1.1445,
+      "step": 4722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.676233675593037e-06,
+      "loss": 1.1305,
+      "step": 4723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.651929778282387e-06,
+      "loss": 1.0612,
+      "step": 4724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.627658429575968e-06,
+      "loss": 1.1627,
+      "step": 4725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.60341963812199e-06,
+      "loss": 0.9738,
+      "step": 4726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.579213412557107e-06,
+      "loss": 1.0835,
+      "step": 4727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.555039761506277e-06,
+      "loss": 1.1519,
+      "step": 4728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.530898693582979e-06,
+      "loss": 1.0958,
+      "step": 4729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.50679021738896e-06,
+      "loss": 1.0079,
+      "step": 4730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.482714341514453e-06,
+      "loss": 1.0504,
+      "step": 4731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.458671074538017e-06,
+      "loss": 1.0999,
+      "step": 4732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.434660425026597e-06,
+      "loss": 1.0899,
+      "step": 4733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.41068240153552e-06,
+      "loss": 1.13,
+      "step": 4734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.386737012608526e-06,
+      "loss": 1.1519,
+      "step": 4735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.362824266777658e-06,
+      "loss": 1.0573,
+      "step": 4736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.33894417256339e-06,
+      "loss": 1.0904,
+      "step": 4737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.315096738474537e-06,
+      "loss": 1.1048,
+      "step": 4738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.291281973008259e-06,
+      "loss": 1.0745,
+      "step": 4739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.267499884650076e-06,
+      "loss": 1.1991,
+      "step": 4740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.243750481873924e-06,
+      "loss": 1.0751,
+      "step": 4741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.220033773142023e-06,
+      "loss": 1.1404,
+      "step": 4742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.196349766904998e-06,
+      "loss": 1.1171,
+      "step": 4743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.172698471601748e-06,
+      "loss": 1.1011,
+      "step": 4744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.149079895659595e-06,
+      "loss": 1.1715,
+      "step": 4745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.125494047494153e-06,
+      "loss": 0.982,
+      "step": 4746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.101940935509412e-06,
+      "loss": 1.0897,
+      "step": 4747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.078420568097656e-06,
+      "loss": 1.0506,
+      "step": 4748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.054932953639571e-06,
+      "loss": 1.1725,
+      "step": 4749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.031478100504053e-06,
+      "loss": 1.0432,
+      "step": 4750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.008056017048449e-06,
+      "loss": 1.009,
+      "step": 4751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.984666711618337e-06,
+      "loss": 1.0975,
+      "step": 4752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.961310192547711e-06,
+      "loss": 1.1056,
+      "step": 4753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.937986468158787e-06,
+      "loss": 1.1234,
+      "step": 4754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.914695546762141e-06,
+      "loss": 0.9648,
+      "step": 4755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.891437436656646e-06,
+      "loss": 1.0199,
+      "step": 4756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.86821214612954e-06,
+      "loss": 1.1264,
+      "step": 4757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.845019683456256e-06,
+      "loss": 1.0433,
+      "step": 4758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.821860056900654e-06,
+      "loss": 1.1119,
+      "step": 4759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.79873327471482e-06,
+      "loss": 1.0488,
+      "step": 4760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.77563934513913e-06,
+      "loss": 1.2036,
+      "step": 4761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.752578276402278e-06,
+      "loss": 1.028,
+      "step": 4762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.72955007672127e-06,
+      "loss": 1.1706,
+      "step": 4763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.706554754301365e-06,
+      "loss": 1.0833,
+      "step": 4764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.683592317336142e-06,
+      "loss": 1.095,
+      "step": 4765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.660662774007398e-06,
+      "loss": 1.0097,
+      "step": 4766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.63776613248528e-06,
+      "loss": 1.0721,
+      "step": 4767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.614902400928181e-06,
+      "loss": 1.0151,
+      "step": 4768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.592071587482774e-06,
+      "loss": 1.0896,
+      "step": 4769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.569273700283974e-06,
+      "loss": 1.0892,
+      "step": 4770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.54650874745505e-06,
+      "loss": 1.104,
+      "step": 4771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.523776737107402e-06,
+      "loss": 1.1255,
+      "step": 4772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5010776773408176e-06,
+      "loss": 1.1362,
+      "step": 4773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.478411576243249e-06,
+      "loss": 0.9929,
+      "step": 4774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.455778441890993e-06,
+      "loss": 1.2026,
+      "step": 4775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4331782823485295e-06,
+      "loss": 1.1838,
+      "step": 4776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.410611105668619e-06,
+      "loss": 1.0532,
+      "step": 4777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.38807691989225e-06,
+      "loss": 1.1568,
+      "step": 4778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.365575733048691e-06,
+      "loss": 1.1277,
+      "step": 4779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.343107553155404e-06,
+      "loss": 1.0359,
+      "step": 4780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.320672388218164e-06,
+      "loss": 0.9714,
+      "step": 4781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2982702462308986e-06,
+      "loss": 1.042,
+      "step": 4782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2759011351758225e-06,
+      "loss": 1.1202,
+      "step": 4783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.253565063023349e-06,
+      "loss": 1.0842,
+      "step": 4784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.231262037732156e-06,
+      "loss": 1.1149,
+      "step": 4785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.208992067249099e-06,
+      "loss": 1.0811,
+      "step": 4786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1867551595093396e-06,
+      "loss": 1.2003,
+      "step": 4787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.16455132243612e-06,
+      "loss": 1.1016,
+      "step": 4788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.142380563941042e-06,
+      "loss": 1.1024,
+      "step": 4789
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.120242891923823e-06,
+      "loss": 0.9457,
+      "step": 4790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.098138314272451e-06,
+      "loss": 1.1389,
+      "step": 4791
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.076066838863083e-06,
+      "loss": 1.053,
+      "step": 4792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.054028473560126e-06,
+      "loss": 0.9613,
+      "step": 4793
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.032023226216111e-06,
+      "loss": 1.1645,
+      "step": 4794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.010051104671866e-06,
+      "loss": 1.1669,
+      "step": 4795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.988112116756318e-06,
+      "loss": 1.1529,
+      "step": 4796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9662062702866906e-06,
+      "loss": 1.1762,
+      "step": 4797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.944333573068307e-06,
+      "loss": 0.9973,
+      "step": 4798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.922494032894744e-06,
+      "loss": 1.061,
+      "step": 4799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.900687657547699e-06,
+      "loss": 1.0916,
+      "step": 4800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.878914454797136e-06,
+      "loss": 1.2637,
+      "step": 4801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.857174432401115e-06,
+      "loss": 1.145,
+      "step": 4802
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.835467598105949e-06,
+      "loss": 1.0736,
+      "step": 4803
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.813793959646064e-06,
+      "loss": 1.0571,
+      "step": 4804
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.792153524744093e-06,
+      "loss": 1.1893,
+      "step": 4805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.770546301110803e-06,
+      "loss": 1.0011,
+      "step": 4806
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.748972296445199e-06,
+      "loss": 1.129,
+      "step": 4807
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.727431518434357e-06,
+      "loss": 1.039,
+      "step": 4808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.705923974753592e-06,
+      "loss": 1.227,
+      "step": 4809
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.684449673066329e-06,
+      "loss": 1.1145,
+      "step": 4810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6630086210241695e-06,
+      "loss": 1.2019,
+      "step": 4811
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.641600826266847e-06,
+      "loss": 1.2063,
+      "step": 4812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.620226296422294e-06,
+      "loss": 0.8993,
+      "step": 4813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.598885039106517e-06,
+      "loss": 1.1393,
+      "step": 4814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.577577061923734e-06,
+      "loss": 1.1965,
+      "step": 4815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.556302372466272e-06,
+      "loss": 1.042,
+      "step": 4816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.535060978314611e-06,
+      "loss": 1.0066,
+      "step": 4817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.51385288703732e-06,
+      "loss": 1.1331,
+      "step": 4818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.492678106191186e-06,
+      "loss": 1.1569,
+      "step": 4819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4715366433210634e-06,
+      "loss": 1.0841,
+      "step": 4820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.450428505959938e-06,
+      "loss": 0.9742,
+      "step": 4821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4293537016289706e-06,
+      "loss": 1.0743,
+      "step": 4822
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.408312237837399e-06,
+      "loss": 1.1183,
+      "step": 4823
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3873041220825605e-06,
+      "loss": 0.9797,
+      "step": 4824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.366329361849988e-06,
+      "loss": 1.1207,
+      "step": 4825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.345387964613259e-06,
+      "loss": 1.1538,
+      "step": 4826
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.324479937834093e-06,
+      "loss": 0.9652,
+      "step": 4827
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.303605288962322e-06,
+      "loss": 1.1432,
+      "step": 4828
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.282764025435872e-06,
+      "loss": 1.1167,
+      "step": 4829
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.261956154680759e-06,
+      "loss": 1.1158,
+      "step": 4830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2411816841111634e-06,
+      "loss": 0.9381,
+      "step": 4831
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2204406211292865e-06,
+      "loss": 0.9171,
+      "step": 4832
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.199732973125472e-06,
+      "loss": 0.9437,
+      "step": 4833
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.17905874747815e-06,
+      "loss": 1.2455,
+      "step": 4834
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.158417951553841e-06,
+      "loss": 1.0436,
+      "step": 4835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.137810592707127e-06,
+      "loss": 1.0285,
+      "step": 4836
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1172366782807354e-06,
+      "loss": 1.0889,
+      "step": 4837
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.096696215605424e-06,
+      "loss": 0.9833,
+      "step": 4838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.076189212000039e-06,
+      "loss": 1.0073,
+      "step": 4839
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.055715674771534e-06,
+      "loss": 1.0962,
+      "step": 4840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.035275611214919e-06,
+      "loss": 1.0485,
+      "step": 4841
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0148690286132685e-06,
+      "loss": 1.117,
+      "step": 4842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9944959342377205e-06,
+      "loss": 0.9238,
+      "step": 4843
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.974156335347514e-06,
+      "loss": 1.2221,
+      "step": 4844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.953850239189929e-06,
+      "loss": 1.1863,
+      "step": 4845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.933577653000321e-06,
+      "loss": 0.9824,
+      "step": 4846
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.913338584002093e-06,
+      "loss": 1.2653,
+      "step": 4847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.893133039406718e-06,
+      "loss": 1.0207,
+      "step": 4848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.872961026413693e-06,
+      "loss": 1.2159,
+      "step": 4849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.852822552210613e-06,
+      "loss": 0.9972,
+      "step": 4850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.832717623973072e-06,
+      "loss": 0.9726,
+      "step": 4851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8126462488647835e-06,
+      "loss": 1.0867,
+      "step": 4852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.792608434037439e-06,
+      "loss": 0.9762,
+      "step": 4853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.772604186630792e-06,
+      "loss": 1.2168,
+      "step": 4854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.752633513772621e-06,
+      "loss": 1.0071,
+      "step": 4855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.732696422578787e-06,
+      "loss": 1.0645,
+      "step": 4856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.712792920153132e-06,
+      "loss": 1.1659,
+      "step": 4857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.692923013587603e-06,
+      "loss": 1.1427,
+      "step": 4858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.67308670996205e-06,
+      "loss": 1.126,
+      "step": 4859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.653284016344495e-06,
+      "loss": 1.0574,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.633514939790862e-06,
+      "loss": 1.0219,
+      "step": 4861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.613779487345205e-06,
+      "loss": 1.1544,
+      "step": 4862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.59407766603951e-06,
+      "loss": 1.2507,
+      "step": 4863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.574409482893839e-06,
+      "loss": 1.036,
+      "step": 4864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.554774944916197e-06,
+      "loss": 1.2427,
+      "step": 4865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.535174059102699e-06,
+      "loss": 0.9983,
+      "step": 4866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.51560683243737e-06,
+      "loss": 1.1266,
+      "step": 4867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.496073271892332e-06,
+      "loss": 1.0747,
+      "step": 4868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.476573384427652e-06,
+      "loss": 1.2108,
+      "step": 4869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.457107176991405e-06,
+      "loss": 1.0939,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.437674656519665e-06,
+      "loss": 1.0136,
+      "step": 4871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.418275829936537e-06,
+      "loss": 1.0175,
+      "step": 4872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.398910704154081e-06,
+      "loss": 1.1995,
+      "step": 4873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.379579286072378e-06,
+      "loss": 1.0973,
+      "step": 4874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3602815825794735e-06,
+      "loss": 1.1887,
+      "step": 4875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.341017600551423e-06,
+      "loss": 1.078,
+      "step": 4876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.321787346852236e-06,
+      "loss": 1.1666,
+      "step": 4877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.302590828333931e-06,
+      "loss": 1.1517,
+      "step": 4878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283428051836503e-06,
+      "loss": 1.0541,
+      "step": 4879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.264299024187935e-06,
+      "loss": 0.9948,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24520375220412e-06,
+      "loss": 1.1788,
+      "step": 4881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.226142242689014e-06,
+      "loss": 1.0109,
+      "step": 4882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.207114502434485e-06,
+      "loss": 1.0714,
+      "step": 4883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.188120538220398e-06,
+      "loss": 1.0359,
+      "step": 4884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.16916035681454e-06,
+      "loss": 1.1743,
+      "step": 4885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.150233964972751e-06,
+      "loss": 1.0476,
+      "step": 4886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.131341369438703e-06,
+      "loss": 1.1591,
+      "step": 4887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1124825769441335e-06,
+      "loss": 1.1728,
+      "step": 4888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.093657594208679e-06,
+      "loss": 1.072,
+      "step": 4889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0748664279399615e-06,
+      "loss": 1.1053,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.056109084833527e-06,
+      "loss": 0.9939,
+      "step": 4891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.037385571572884e-06,
+      "loss": 1.0986,
+      "step": 4892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.018695894829473e-06,
+      "loss": 1.1285,
+      "step": 4893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0000400612627114e-06,
+      "loss": 1.0665,
+      "step": 4894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.981418077519906e-06,
+      "loss": 1.0922,
+      "step": 4895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.962829950236369e-06,
+      "loss": 1.0443,
+      "step": 4896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.944275686035294e-06,
+      "loss": 1.1203,
+      "step": 4897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.925755291527823e-06,
+      "loss": 1.1114,
+      "step": 4898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.907268773313023e-06,
+      "loss": 1.0453,
+      "step": 4899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888816137977925e-06,
+      "loss": 1.1058,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.870397392097437e-06,
+      "loss": 1.1254,
+      "step": 4901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.852012542234463e-06,
+      "loss": 1.1478,
+      "step": 4902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.833661594939732e-06,
+      "loss": 0.9834,
+      "step": 4903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.815344556751966e-06,
+      "loss": 1.1143,
+      "step": 4904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.79706143419778e-06,
+      "loss": 1.0477,
+      "step": 4905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.778812233791719e-06,
+      "loss": 1.171,
+      "step": 4906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.760596962036212e-06,
+      "loss": 1.001,
+      "step": 4907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.742415625421659e-06,
+      "loss": 1.0514,
+      "step": 4908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.724268230426265e-06,
+      "loss": 1.1909,
+      "step": 4909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.706154783516248e-06,
+      "loss": 1.0229,
+      "step": 4910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.688075291145655e-06,
+      "loss": 1.1239,
+      "step": 4911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6700297597565e-06,
+      "loss": 1.1357,
+      "step": 4912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.652018195778629e-06,
+      "loss": 1.1818,
+      "step": 4913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.63404060562983e-06,
+      "loss": 1.176,
+      "step": 4914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.616096995715746e-06,
+      "loss": 1.2285,
+      "step": 4915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.598187372429974e-06,
+      "loss": 1.1004,
+      "step": 4916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.580311742153942e-06,
+      "loss": 1.0686,
+      "step": 4917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5624701112570004e-06,
+      "loss": 1.278,
+      "step": 4918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.544662486096374e-06,
+      "loss": 1.1184,
+      "step": 4919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5268888730171766e-06,
+      "loss": 1.0896,
+      "step": 4920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.509149278352365e-06,
+      "loss": 1.086,
+      "step": 4921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.491443708422838e-06,
+      "loss": 1.0237,
+      "step": 4922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.473772169537316e-06,
+      "loss": 1.1065,
+      "step": 4923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.456134667992462e-06,
+      "loss": 1.1581,
+      "step": 4924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.438531210072716e-06,
+      "loss": 1.0191,
+      "step": 4925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.420961802050461e-06,
+      "loss": 1.0143,
+      "step": 4926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.40342645018591e-06,
+      "loss": 1.209,
+      "step": 4927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.385925160727178e-06,
+      "loss": 0.9196,
+      "step": 4928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.368457939910209e-06,
+      "loss": 1.1442,
+      "step": 4929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.351024793958835e-06,
+      "loss": 1.152,
+      "step": 4930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.33362572908469e-06,
+      "loss": 0.8875,
+      "step": 4931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.316260751487356e-06,
+      "loss": 1.1221,
+      "step": 4932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.298929867354174e-06,
+      "loss": 1.0307,
+      "step": 4933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2816330828604145e-06,
+      "loss": 1.1306,
+      "step": 4934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.26437040416916e-06,
+      "loss": 1.0708,
+      "step": 4935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.247141837431345e-06,
+      "loss": 1.2254,
+      "step": 4936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.229947388785738e-06,
+      "loss": 1.0179,
+      "step": 4937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.21278706435898e-06,
+      "loss": 1.0309,
+      "step": 4938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.195660870265516e-06,
+      "loss": 1.1924,
+      "step": 4939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.178568812607675e-06,
+      "loss": 1.03,
+      "step": 4940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.161510897475584e-06,
+      "loss": 1.188,
+      "step": 4941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.144487130947217e-06,
+      "loss": 1.0727,
+      "step": 4942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.127497519088375e-06,
+      "loss": 1.0481,
+      "step": 4943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1105420679527275e-06,
+      "loss": 1.2158,
+      "step": 4944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.093620783581698e-06,
+      "loss": 0.9981,
+      "step": 4945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.076733672004618e-06,
+      "loss": 1.0804,
+      "step": 4946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.059880739238575e-06,
+      "loss": 1.0426,
+      "step": 4947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.043061991288522e-06,
+      "loss": 1.0445,
+      "step": 4948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0262774341471965e-06,
+      "loss": 0.9691,
+      "step": 4949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.009527073795205e-06,
+      "loss": 1.0642,
+      "step": 4950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.992810916200895e-06,
+      "loss": 1.1034,
+      "step": 4951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9761289673205135e-06,
+      "loss": 1.1797,
+      "step": 4952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.95948123309805e-06,
+      "loss": 1.186,
+      "step": 4953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.942867719465326e-06,
+      "loss": 0.9731,
+      "step": 4954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.926288432341963e-06,
+      "loss": 0.9351,
+      "step": 4955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909743377635422e-06,
+      "loss": 1.0956,
+      "step": 4956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8932325612409115e-06,
+      "loss": 1.0054,
+      "step": 4957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.876755989041469e-06,
+      "loss": 1.092,
+      "step": 4958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.860313666907955e-06,
+      "loss": 1.1788,
+      "step": 4959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.843905600698982e-06,
+      "loss": 1.0505,
+      "step": 4960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.827531796260975e-06,
+      "loss": 1.166,
+      "step": 4961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8111922594281536e-06,
+      "loss": 0.9974,
+      "step": 4962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7948869960225287e-06,
+      "loss": 0.9994,
+      "step": 4963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.778616011853886e-06,
+      "loss": 1.1241,
+      "step": 4964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7623793127198104e-06,
+      "loss": 1.0722,
+      "step": 4965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.746176904405685e-06,
+      "loss": 1.1558,
+      "step": 4966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.730008792684614e-06,
+      "loss": 0.9818,
+      "step": 4967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7138749833175668e-06,
+      "loss": 1.0539,
+      "step": 4968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.697775482053223e-06,
+      "loss": 1.0997,
+      "step": 4969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6817102946280602e-06,
+      "loss": 1.1308,
+      "step": 4970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.665679426766355e-06,
+      "loss": 0.9806,
+      "step": 4971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6496828841801056e-06,
+      "loss": 1.0561,
+      "step": 4972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6337206725691075e-06,
+      "loss": 1.1438,
+      "step": 4973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6177927976209335e-06,
+      "loss": 1.1778,
+      "step": 4974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6018992650109106e-06,
+      "loss": 1.1048,
+      "step": 4975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5860400804021198e-06,
+      "loss": 1.1025,
+      "step": 4976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.570215249445419e-06,
+      "loss": 1.0408,
+      "step": 4977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.554424777779408e-06,
+      "loss": 1.0478,
+      "step": 4978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.538668671030465e-06,
+      "loss": 1.0621,
+      "step": 4979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.522946934812699e-06,
+      "loss": 1.0937,
+      "step": 4980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5072595747280057e-06,
+      "loss": 1.0604,
+      "step": 4981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4916065963659927e-06,
+      "loss": 1.0492,
+      "step": 4982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4759880053040538e-06,
+      "loss": 1.0799,
+      "step": 4983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.460403807107304e-06,
+      "loss": 1.1953,
+      "step": 4984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4448540073286238e-06,
+      "loss": 1.1565,
+      "step": 4985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4293386115085924e-06,
+      "loss": 1.0645,
+      "step": 4986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4138576251756094e-06,
+      "loss": 1.0587,
+      "step": 4987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.398411053845729e-06,
+      "loss": 1.1291,
+      "step": 4988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.382998903022816e-06,
+      "loss": 1.0108,
+      "step": 4989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.36762117819841e-06,
+      "loss": 0.9893,
+      "step": 4990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3522778848518287e-06,
+      "loss": 1.1613,
+      "step": 4991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3369690284500878e-06,
+      "loss": 1.0803,
+      "step": 4992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.321694614447968e-06,
+      "loss": 1.1458,
+      "step": 4993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3064546482879377e-06,
+      "loss": 1.0997,
+      "step": 4994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2912491354002205e-06,
+      "loss": 1.1329,
+      "step": 4995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2760780812027715e-06,
+      "loss": 1.0355,
+      "step": 4996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2609414911012215e-06,
+      "loss": 0.9484,
+      "step": 4997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2458393704889568e-06,
+      "loss": 0.9982,
+      "step": 4998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.230771724747106e-06,
+      "loss": 1.0008,
+      "step": 4999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.215738559244441e-06,
+      "loss": 0.9984,
+      "step": 5000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.200739879337522e-06,
+      "loss": 1.1602,
+      "step": 5001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1857756903705737e-06,
+      "loss": 0.9704,
+      "step": 5002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1708459976755533e-06,
+      "loss": 1.1174,
+      "step": 5003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1559508065721055e-06,
+      "loss": 1.2723,
+      "step": 5004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.141090122367629e-06,
+      "loss": 1.0781,
+      "step": 5005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1262639503571666e-06,
+      "loss": 1.2147,
+      "step": 5006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1114722958235144e-06,
+      "loss": 1.1479,
+      "step": 5007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.096715164037123e-06,
+      "loss": 1.0927,
+      "step": 5008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.081992560256186e-06,
+      "loss": 1.0892,
+      "step": 5009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.067304489726552e-06,
+      "loss": 1.0388,
+      "step": 5010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.052650957681813e-06,
+      "loss": 1.15,
+      "step": 5011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0380319693432247e-06,
+      "loss": 1.0364,
+      "step": 5012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0234475299197317e-06,
+      "loss": 1.1205,
+      "step": 5013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0088976446079554e-06,
+      "loss": 1.1037,
+      "step": 5014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.994382318592259e-06,
+      "loss": 1.0989,
+      "step": 5015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.97990155704464e-06,
+      "loss": 1.1885,
+      "step": 5016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.965455365124803e-06,
+      "loss": 1.1514,
+      "step": 5017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9510437479801312e-06,
+      "loss": 1.0882,
+      "step": 5018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9366667107456837e-06,
+      "loss": 1.0325,
+      "step": 5019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.922324258544196e-06,
+      "loss": 0.9147,
+      "step": 5020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9080163964861017e-06,
+      "loss": 1.1736,
+      "step": 5021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8937431296694686e-06,
+      "loss": 0.9818,
+      "step": 5022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.879504463180094e-06,
+      "loss": 1.0715,
+      "step": 5023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8653004020913753e-06,
+      "loss": 1.2085,
+      "step": 5024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8511309514644644e-06,
+      "loss": 1.0781,
+      "step": 5025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.836996116348101e-06,
+      "loss": 1.1885,
+      "step": 5026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.822895901778744e-06,
+      "loss": 1.1961,
+      "step": 5027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.808830312780486e-06,
+      "loss": 1.1805,
+      "step": 5028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7947993543651295e-06,
+      "loss": 1.0032,
+      "step": 5029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7808030315320534e-06,
+      "loss": 1.096,
+      "step": 5030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7668413492683674e-06,
+      "loss": 1.167,
+      "step": 5031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7529143125488157e-06,
+      "loss": 1.1298,
+      "step": 5032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7390219263358054e-06,
+      "loss": 1.1451,
+      "step": 5033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7251641955793773e-06,
+      "loss": 1.0965,
+      "step": 5034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.711341125217237e-06,
+      "loss": 1.0075,
+      "step": 5035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6975527201747342e-06,
+      "loss": 1.0196,
+      "step": 5036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.683798985364894e-06,
+      "loss": 1.015,
+      "step": 5037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.67007992568834e-06,
+      "loss": 1.0298,
+      "step": 5038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6563955460333856e-06,
+      "loss": 1.0345,
+      "step": 5039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.642745851275963e-06,
+      "loss": 1.0552,
+      "step": 5040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.629130846279648e-06,
+      "loss": 1.0295,
+      "step": 5041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.61555053589565e-06,
+      "loss": 1.0309,
+      "step": 5042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.602004924962842e-06,
+      "loss": 0.9245,
+      "step": 5043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5884940183076966e-06,
+      "loss": 1.0318,
+      "step": 5044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5750178207443744e-06,
+      "loss": 1.1024,
+      "step": 5045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5615763370745895e-06,
+      "loss": 1.1694,
+      "step": 5046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5481695720877663e-06,
+      "loss": 0.9674,
+      "step": 5047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.534797530560895e-06,
+      "loss": 1.0522,
+      "step": 5048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5214602172586533e-06,
+      "loss": 1.0363,
+      "step": 5049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.508157636933284e-06,
+      "loss": 1.0639,
+      "step": 5050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4948897943247284e-06,
+      "loss": 1.1392,
+      "step": 5051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.48165669416045e-06,
+      "loss": 1.2711,
+      "step": 5052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4684583411556217e-06,
+      "loss": 1.0962,
+      "step": 5053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4552947400129922e-06,
+      "loss": 1.1312,
+      "step": 5054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.442165895422954e-06,
+      "loss": 0.9876,
+      "step": 5055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.429071812063488e-06,
+      "loss": 1.1495,
+      "step": 5056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.416012494600195e-06,
+      "loss": 1.1044,
+      "step": 5057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.402987947686286e-06,
+      "loss": 1.1287,
+      "step": 5058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3899981759626155e-06,
+      "loss": 1.0715,
+      "step": 5059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3770431840576036e-06,
+      "loss": 1.1431,
+      "step": 5060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.364122976587313e-06,
+      "loss": 0.9982,
+      "step": 5061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3512375581553837e-06,
+      "loss": 1.1659,
+      "step": 5062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.338386933353065e-06,
+      "loss": 0.981,
+      "step": 5063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.325571106759228e-06,
+      "loss": 1.1387,
+      "step": 5064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3127900829403306e-06,
+      "loss": 1.3083,
+      "step": 5065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3000438664504188e-06,
+      "loss": 1.204,
+      "step": 5066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.287332461831182e-06,
+      "loss": 1.1022,
+      "step": 5067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2746558736118305e-06,
+      "loss": 1.0263,
+      "step": 5068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2620141063092405e-06,
+      "loss": 1.0285,
+      "step": 5069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24940716442783e-06,
+      "loss": 1.2277,
+      "step": 5070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2368350524596513e-06,
+      "loss": 1.0792,
+      "step": 5071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.22429777488431e-06,
+      "loss": 1.1459,
+      "step": 5072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2117953361690425e-06,
+      "loss": 1.1945,
+      "step": 5073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1993277407685974e-06,
+      "loss": 0.9891,
+      "step": 5074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186894993125388e-06,
+      "loss": 1.2296,
+      "step": 5075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.174497097669359e-06,
+      "loss": 1.1288,
+      "step": 5076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1621340588180772e-06,
+      "loss": 1.0588,
+      "step": 5077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1498058809766517e-06,
+      "loss": 1.1508,
+      "step": 5078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1375125685377915e-06,
+      "loss": 1.0477,
+      "step": 5079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1252541258817703e-06,
+      "loss": 1.0072,
+      "step": 5080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1130305573764495e-06,
+      "loss": 0.982,
+      "step": 5081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.100841867377257e-06,
+      "loss": 1.0752,
+      "step": 5082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0886880602272063e-06,
+      "loss": 1.0752,
+      "step": 5083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0765691402568454e-06,
+      "loss": 1.0647,
+      "step": 5084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.064485111784342e-06,
+      "loss": 1.0203,
+      "step": 5085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0524359791153746e-06,
+      "loss": 0.9323,
+      "step": 5086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.040421746543253e-06,
+      "loss": 1.2125,
+      "step": 5087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.028442418348797e-06,
+      "loss": 1.0169,
+      "step": 5088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.016497998800426e-06,
+      "loss": 1.1081,
+      "step": 5089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0045884921540804e-06,
+      "loss": 1.0779,
+      "step": 5090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.992713902653309e-06,
+      "loss": 1.0951,
+      "step": 5091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.980874234529173e-06,
+      "loss": 1.0798,
+      "step": 5092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9690694920003416e-06,
+      "loss": 1.1386,
+      "step": 5093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9572996792729836e-06,
+      "loss": 1.1497,
+      "step": 5094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.945564800540867e-06,
+      "loss": 1.0507,
+      "step": 5095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9338648599852792e-06,
+      "loss": 1.1859,
+      "step": 5096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.922199861775087e-06,
+      "loss": 1.1431,
+      "step": 5097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.910569810066687e-06,
+      "loss": 1.183,
+      "step": 5098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.898974709004031e-06,
+      "loss": 1.1329,
+      "step": 5099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8874145627186257e-06,
+      "loss": 1.16,
+      "step": 5100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8758893753295092e-06,
+      "loss": 1.0495,
+      "step": 5101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8643991509432523e-06,
+      "loss": 1.0592,
+      "step": 5102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8529438936540021e-06,
+      "loss": 1.273,
+      "step": 5103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8415236075434162e-06,
+      "loss": 1.0229,
+      "step": 5104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8301382966807278e-06,
+      "loss": 1.0721,
+      "step": 5105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8187879651226592e-06,
+      "loss": 1.1227,
+      "step": 5106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.807472616913497e-06,
+      "loss": 1.0432,
+      "step": 5107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7961922560850609e-06,
+      "loss": 1.1071,
+      "step": 5108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.784946886656702e-06,
+      "loss": 1.1478,
+      "step": 5109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7737365126353045e-06,
+      "loss": 1.1612,
+      "step": 5110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.762561138015284e-06,
+      "loss": 1.1226,
+      "step": 5111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7514207667785997e-06,
+      "loss": 1.2875,
+      "step": 5112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7403154028946877e-06,
+      "loss": 1.1155,
+      "step": 5113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7292450503205715e-06,
+      "loss": 1.1022,
+      "step": 5114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7182097130007734e-06,
+      "loss": 1.1877,
+      "step": 5115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7072093948673263e-06,
+      "loss": 1.0387,
+      "step": 5116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6962440998397944e-06,
+      "loss": 1.1317,
+      "step": 5117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.685313831825297e-06,
+      "loss": 1.2064,
+      "step": 5118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6744185947184077e-06,
+      "loss": 1.0981,
+      "step": 5119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6635583924012765e-06,
+      "loss": 1.0213,
+      "step": 5120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6527332287435305e-06,
+      "loss": 1.1555,
+      "step": 5121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6419431076023506e-06,
+      "loss": 1.1812,
+      "step": 5122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6311880328223728e-06,
+      "loss": 1.057,
+      "step": 5123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6204680082358203e-06,
+      "loss": 1.0278,
+      "step": 5124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097830376623713e-06,
+      "loss": 1.0467,
+      "step": 5125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5991331249092246e-06,
+      "loss": 1.0334,
+      "step": 5126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5885182737711003e-06,
+      "loss": 1.1094,
+      "step": 5127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5779384880302283e-06,
+      "loss": 1.0795,
+      "step": 5128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.567393771456316e-06,
+      "loss": 0.9051,
+      "step": 5129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5568841278066237e-06,
+      "loss": 1.0881,
+      "step": 5130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5464095608258566e-06,
+      "loss": 1.113,
+      "step": 5131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5359700742462512e-06,
+      "loss": 1.0202,
+      "step": 5132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5255656717875655e-06,
+      "loss": 1.2327,
+      "step": 5133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5151963571570227e-06,
+      "loss": 1.1244,
+      "step": 5134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5048621340493562e-06,
+      "loss": 1.1094,
+      "step": 5135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4945630061467985e-06,
+      "loss": 1.0016,
+      "step": 5136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4842989771190584e-06,
+      "loss": 1.1182,
+      "step": 5137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4740700506233773e-06,
+      "loss": 1.1291,
+      "step": 5138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4638762303044506e-06,
+      "loss": 0.9794,
+      "step": 5139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4537175197944842e-06,
+      "loss": 1.1072,
+      "step": 5140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4435939227131713e-06,
+      "loss": 0.984,
+      "step": 5141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.433505442667682e-06,
+      "loss": 1.0934,
+      "step": 5142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4234520832527076e-06,
+      "loss": 1.0764,
+      "step": 5143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4134338480503829e-06,
+      "loss": 1.2295,
+      "step": 5144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.403450740630341e-06,
+      "loss": 1.1393,
+      "step": 5145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3935027645497146e-06,
+      "loss": 1.0448,
+      "step": 5146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3835899233531013e-06,
+      "loss": 1.1514,
+      "step": 5147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3737122205725982e-06,
+      "loss": 1.0138,
+      "step": 5148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3638696597277679e-06,
+      "loss": 1.1058,
+      "step": 5149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.354062244325638e-06,
+      "loss": 1.0483,
+      "step": 5150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3442899778607353e-06,
+      "loss": 1.121,
+      "step": 5151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3345528638150751e-06,
+      "loss": 1.1022,
+      "step": 5152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3248509056580926e-06,
+      "loss": 1.187,
+      "step": 5153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3151841068467675e-06,
+      "loss": 0.9274,
+      "step": 5154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3055524708254885e-06,
+      "loss": 1.0737,
+      "step": 5155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2959560010261662e-06,
+      "loss": 1.1858,
+      "step": 5156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2863947008681433e-06,
+      "loss": 1.2378,
+      "step": 5157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2768685737582497e-06,
+      "loss": 1.1711,
+      "step": 5158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2673776230907708e-06,
+      "loss": 0.9905,
+      "step": 5159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.25792185224749e-06,
+      "loss": 1.1081,
+      "step": 5160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2485012645976014e-06,
+      "loss": 1.0604,
+      "step": 5161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2391158634978194e-06,
+      "loss": 0.9934,
+      "step": 5162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2297656522922695e-06,
+      "loss": 1.1166,
+      "step": 5163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2204506343125866e-06,
+      "loss": 0.9062,
+      "step": 5164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2111708128778266e-06,
+      "loss": 1.1013,
+      "step": 5165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2019261912945334e-06,
+      "loss": 1.0317,
+      "step": 5166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1927167728566835e-06,
+      "loss": 1.1802,
+      "step": 5167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1835425608457296e-06,
+      "loss": 1.1185,
+      "step": 5168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.174403558530568e-06,
+      "loss": 1.1468,
+      "step": 5169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1652997691675605e-06,
+      "loss": 1.077,
+      "step": 5170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1562311960005234e-06,
+      "loss": 1.068,
+      "step": 5171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1471978422606943e-06,
+      "loss": 1.0773,
+      "step": 5172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1381997111667874e-06,
+      "loss": 1.0322,
+      "step": 5173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1292368059249936e-06,
+      "loss": 1.1575,
+      "step": 5174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1203091297288804e-06,
+      "loss": 1.0104,
+      "step": 5175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1114166857595365e-06,
+      "loss": 1.0404,
+      "step": 5176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1025594771854497e-06,
+      "loss": 1.1183,
+      "step": 5177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0937375071625844e-06,
+      "loss": 1.1172,
+      "step": 5178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.084950778834304e-06,
+      "loss": 0.9908,
+      "step": 5179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0761992953314703e-06,
+      "loss": 1.082,
+      "step": 5180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.067483059772334e-06,
+      "loss": 1.0515,
+      "step": 5181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.058802075262655e-06,
+      "loss": 1.124,
+      "step": 5182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0501563448955365e-06,
+      "loss": 1.1516,
+      "step": 5183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0415458717516146e-06,
+      "loss": 1.0563,
+      "step": 5184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.032970658898913e-06,
+      "loss": 1.093,
+      "step": 5185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0244307093928873e-06,
+      "loss": 1.2772,
+      "step": 5186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0159260262764591e-06,
+      "loss": 0.967,
+      "step": 5187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0074566125799601e-06,
+      "loss": 1.1734,
+      "step": 5188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.990224713211538e-07,
+      "loss": 1.0395,
+      "step": 5189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.906236055052586e-07,
+      "loss": 1.2157,
+      "step": 5190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.82260018124881e-07,
+      "loss": 1.0924,
+      "step": 5191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.739317121601254e-07,
+      "loss": 1.0267,
+      "step": 5192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.656386905784632e-07,
+      "loss": 1.1127,
+      "step": 5193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.573809563348079e-07,
+      "loss": 1.2254,
+      "step": 5194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.491585123715063e-07,
+      "loss": 1.0336,
+      "step": 5195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.40971361618348e-07,
+      "loss": 1.1747,
+      "step": 5196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.328195069925217e-07,
+      "loss": 1.146,
+      "step": 5197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.247029513986482e-07,
+      "loss": 1.1022,
+      "step": 5198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.16621697728759e-07,
+      "loss": 1.1704,
+      "step": 5199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.085757488623392e-07,
+      "loss": 1.0252,
+      "step": 5200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.005651076662624e-07,
+      "loss": 1.112,
+      "step": 5201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.925897769948344e-07,
+      "loss": 1.0904,
+      "step": 5202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.846497596897707e-07,
+      "loss": 1.1268,
+      "step": 5203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.767450585802306e-07,
+      "loss": 1.1993,
+      "step": 5204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.688756764827388e-07,
+      "loss": 0.9887,
+      "step": 5205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.610416162013079e-07,
+      "loss": 1.0434,
+      "step": 5206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.532428805272829e-07,
+      "loss": 0.9683,
+      "step": 5207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.454794722394965e-07,
+      "loss": 0.9725,
+      "step": 5208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.377513941041359e-07,
+      "loss": 1.1331,
+      "step": 5209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.300586488748541e-07,
+      "loss": 1.1535,
+      "step": 5210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.224012392926472e-07,
+      "loss": 1.2134,
+      "step": 5211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.147791680859773e-07,
+      "loss": 1.1166,
+      "step": 5212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.071924379706941e-07,
+      "loss": 1.1223,
+      "step": 5213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.996410516500685e-07,
+      "loss": 1.2418,
+      "step": 5214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.921250118147483e-07,
+      "loss": 1.2182,
+      "step": 5215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.846443211428023e-07,
+      "loss": 1.1497,
+      "step": 5216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.771989822997206e-07,
+      "loss": 1.1815,
+      "step": 5217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.69788997938381e-07,
+      "loss": 1.145,
+      "step": 5218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.624143706990494e-07,
+      "loss": 1.0281,
+      "step": 5219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.550751032094239e-07,
+      "loss": 1.042,
+      "step": 5220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.477711980845903e-07,
+      "loss": 1.0341,
+      "step": 5221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.405026579270224e-07,
+      "loss": 1.1248,
+      "step": 5222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.332694853266042e-07,
+      "loss": 1.1705,
+      "step": 5223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.260716828606295e-07,
+      "loss": 0.9899,
+      "step": 5224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.18909253093758e-07,
+      "loss": 1.0665,
+      "step": 5225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.117821985780926e-07,
+      "loss": 1.066,
+      "step": 5226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.046905218530686e-07,
+      "loss": 1.1695,
+      "step": 5227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.976342254455759e-07,
+      "loss": 1.1695,
+      "step": 5228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.906133118698588e-07,
+      "loss": 1.0596,
+      "step": 5229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.836277836275828e-07,
+      "loss": 1.0463,
+      "step": 5230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.766776432077682e-07,
+      "loss": 1.147,
+      "step": 5231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.697628930868671e-07,
+      "loss": 0.9972,
+      "step": 5232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.628835357286978e-07,
+      "loss": 1.1396,
+      "step": 5233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.560395735844549e-07,
+      "loss": 1.0292,
+      "step": 5234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.492310090927656e-07,
+      "loss": 1.1397,
+      "step": 5235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.424578446796003e-07,
+      "loss": 1.1494,
+      "step": 5236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.357200827583398e-07,
+      "loss": 0.9675,
+      "step": 5237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.290177257297414e-07,
+      "loss": 1.1603,
+      "step": 5238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.223507759819392e-07,
+      "loss": 1.0995,
+      "step": 5239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.157192358904774e-07,
+      "loss": 1.1194,
+      "step": 5240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.091231078182547e-07,
+      "loss": 1.1495,
+      "step": 5241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.0256239411558e-07,
+      "loss": 1.0307,
+      "step": 5242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.960370971201168e-07,
+      "loss": 1.0493,
+      "step": 5243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.895472191569274e-07,
+      "loss": 1.0051,
+      "step": 5244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.830927625384286e-07,
+      "loss": 1.0457,
+      "step": 5245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.766737295644586e-07,
+      "loss": 1.0513,
+      "step": 5246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.702901225221879e-07,
+      "loss": 1.0672,
+      "step": 5247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.63941943686197e-07,
+      "loss": 1.1393,
+      "step": 5248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.576291953184321e-07,
+      "loss": 1.1236,
+      "step": 5249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.51351879668216e-07,
+      "loss": 1.1455,
+      "step": 5250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.451099989722375e-07,
+      "loss": 1.0831,
+      "step": 5251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.38903555454573e-07,
+      "loss": 1.097,
+      "step": 5252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.327325513266534e-07,
+      "loss": 1.1015,
+      "step": 5253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.26596988787309e-07,
+      "loss": 1.0808,
+      "step": 5254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.204968700227242e-07,
+      "loss": 1.0423,
+      "step": 5255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.144321972064603e-07,
+      "loss": 1.0624,
+      "step": 5256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.084029724994332e-07,
+      "loss": 1.0655,
+      "step": 5257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.024091980499468e-07,
+      "loss": 1.0891,
+      "step": 5258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.964508759936704e-07,
+      "loss": 1.0787,
+      "step": 5259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.905280084536279e-07,
+      "loss": 1.0543,
+      "step": 5260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.846405975402424e-07,
+      "loss": 1.2689,
+      "step": 5261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.78788645351258e-07,
+      "loss": 1.0734,
+      "step": 5262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7297215397181795e-07,
+      "loss": 1.1216,
+      "step": 5263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.671911254744199e-07,
+      "loss": 1.0562,
+      "step": 5264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.614455619189273e-07,
+      "loss": 0.9704,
+      "step": 5265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.557354653525581e-07,
+      "loss": 1.0406,
+      "step": 5266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5006083780991804e-07,
+      "loss": 1.1506,
+      "step": 5267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4442168131293427e-07,
+      "loss": 1.2013,
+      "step": 5268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.388179978709217e-07,
+      "loss": 1.1047,
+      "step": 5269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3324978948057205e-07,
+      "loss": 1.0833,
+      "step": 5270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2771705812588714e-07,
+      "loss": 1.0509,
+      "step": 5271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.222198057782789e-07,
+      "loss": 1.1156,
+      "step": 5272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.167580343964916e-07,
+      "loss": 1.0356,
+      "step": 5273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.113317459266242e-07,
+      "loss": 1.0635,
+      "step": 5274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0594094230214096e-07,
+      "loss": 1.1529,
+      "step": 5275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0058562544387224e-07,
+      "loss": 0.9414,
+      "step": 5276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952657972599805e-07,
+      "loss": 0.9776,
+      "step": 5277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8998145964599386e-07,
+      "loss": 1.134,
+      "step": 5278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8473261448480623e-07,
+      "loss": 0.9647,
+      "step": 5279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.795192636466549e-07,
+      "loss": 1.0463,
+      "step": 5280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7434140898912063e-07,
+      "loss": 1.1119,
+      "step": 5281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.691990523571498e-07,
+      "loss": 1.0147,
+      "step": 5282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.640921955830434e-07,
+      "loss": 1.175,
+      "step": 5283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.590208404864348e-07,
+      "loss": 1.1244,
+      "step": 5284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.539849888743341e-07,
+      "loss": 1.0701,
+      "step": 5285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4898464254107257e-07,
+      "loss": 0.9664,
+      "step": 5286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4401980326835836e-07,
+      "loss": 1.1356,
+      "step": 5287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390904728252098e-07,
+      "loss": 1.0698,
+      "step": 5288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3419665296804404e-07,
+      "loss": 1.0924,
+      "step": 5289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.293383454405774e-07,
+      "loss": 0.9682,
+      "step": 5290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.245155519739029e-07,
+      "loss": 1.1202,
+      "step": 5291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.197282742864571e-07,
+      "loss": 1.0619,
+      "step": 5292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1497651408399776e-07,
+      "loss": 1.1743,
+      "step": 5293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1026027305964823e-07,
+      "loss": 1.0713,
+      "step": 5294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0557955289387543e-07,
+      "loss": 0.9807,
+      "step": 5295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.009343552544897e-07,
+      "loss": 1.1561,
+      "step": 5296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9632468179664474e-07,
+      "loss": 0.9726,
+      "step": 5297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.917505341628157e-07,
+      "loss": 1.0475,
+      "step": 5298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.872119139828433e-07,
+      "loss": 1.1383,
+      "step": 5299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8270882287390055e-07,
+      "loss": 1.0857,
+      "step": 5300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.78241262440504e-07,
+      "loss": 1.1581,
+      "step": 5301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7380923427451355e-07,
+      "loss": 1.0199,
+      "step": 5302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.694127399551216e-07,
+      "loss": 1.1851,
+      "step": 5303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6505178104885285e-07,
+      "loss": 1.0126,
+      "step": 5304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.607263591095754e-07,
+      "loss": 1.0344,
+      "step": 5305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5643647567851205e-07,
+      "loss": 1.0242,
+      "step": 5306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5218213228420664e-07,
+      "loss": 1.0795,
+      "step": 5307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.479633304425355e-07,
+      "loss": 0.976,
+      "step": 5308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4378007165671843e-07,
+      "loss": 1.1856,
+      "step": 5309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.396323574173076e-07,
+      "loss": 1.2325,
+      "step": 5310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3552018920219855e-07,
+      "loss": 1.118,
+      "step": 5311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.314435684766081e-07,
+      "loss": 1.0427,
+      "step": 5312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2740249669309655e-07,
+      "loss": 1.016,
+      "step": 5313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2339697529155656e-07,
+      "loss": 1.1043,
+      "step": 5314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1942700569921314e-07,
+      "loss": 1.0229,
+      "step": 5315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1549258933061255e-07,
+      "loss": 1.0361,
+      "step": 5316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.115937275876445e-07,
+      "loss": 1.2238,
+      "step": 5317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0773042185954218e-07,
+      "loss": 1.0374,
+      "step": 5318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0390267352284888e-07,
+      "loss": 1.2312,
+      "step": 5319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0011048394142917e-07,
+      "loss": 0.9094,
+      "step": 5320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9635385446652442e-07,
+      "loss": 1.0541,
+      "step": 5321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9263278643664175e-07,
+      "loss": 1.0582,
+      "step": 5322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.88947281177676e-07,
+      "loss": 1.157,
+      "step": 5323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8529734000281017e-07,
+      "loss": 1.0366,
+      "step": 5324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8168296421258169e-07,
+      "loss": 1.0346,
+      "step": 5325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7810415509483812e-07,
+      "loss": 1.0187,
+      "step": 5326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7456091392474838e-07,
+      "loss": 1.0703,
+      "step": 5327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7105324196482475e-07,
+      "loss": 1.1149,
+      "step": 5328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.675811404649119e-07,
+      "loss": 1.0587,
+      "step": 5329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6414461066216468e-07,
+      "loss": 1.2965,
+      "step": 5330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6074365378105915e-07,
+      "loss": 0.957,
+      "step": 5331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5737827103340375e-07,
+      "loss": 1.0557,
+      "step": 5332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5404846361833926e-07,
+      "loss": 1.2173,
+      "step": 5333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5075423272231658e-07,
+      "loss": 1.0139,
+      "step": 5334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4749557951911906e-07,
+      "loss": 1.1486,
+      "step": 5335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4427250516985124e-07,
+      "loss": 1.0918,
+      "step": 5336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4108501082295e-07,
+      "loss": 0.9742,
+      "step": 5337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3793309761414018e-07,
+      "loss": 1.0624,
+      "step": 5338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3481676666651234e-07,
+      "loss": 1.0353,
+      "step": 5339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3173601909045597e-07,
+      "loss": 0.9924,
+      "step": 5340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2869085598368191e-07,
+      "loss": 1.1221,
+      "step": 5341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2568127843122223e-07,
+      "loss": 0.9961,
+      "step": 5342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2270728750544137e-07,
+      "loss": 1.0793,
+      "step": 5343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1976888426600275e-07,
+      "loss": 1.1007,
+      "step": 5344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1686606975991333e-07,
+      "loss": 1.0587,
+      "step": 5345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1399884502146796e-07,
+      "loss": 1.2819,
+      "step": 5346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.111672110723272e-07,
+      "loss": 1.0973,
+      "step": 5347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.083711689214062e-07,
+      "loss": 1.0853,
+      "step": 5348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0561071956500801e-07,
+      "loss": 1.051,
+      "step": 5349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0288586398670142e-07,
+      "loss": 1.145,
+      "step": 5350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0019660315738755e-07,
+      "loss": 1.0119,
+      "step": 5351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.75429380352999e-08,
+      "loss": 1.0718,
+      "step": 5352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.492486956597102e-08,
+      "loss": 1.1555,
+      "step": 5353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.23423986822547e-08,
+      "loss": 1.1669,
+      "step": 5354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.979552630433708e-08,
+      "loss": 1.0573,
+      "step": 5355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.728425333967005e-08,
+      "loss": 0.9589,
+      "step": 5356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.480858068309339e-08,
+      "loss": 1.0101,
+      "step": 5357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.236850921670148e-08,
+      "loss": 1.012,
+      "step": 5358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.996403980993217e-08,
+      "loss": 1.0173,
+      "step": 5359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.759517331952238e-08,
+      "loss": 1.1157,
+      "step": 5360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.526191058955245e-08,
+      "loss": 1.1358,
+      "step": 5361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.296425245137961e-08,
+      "loss": 1.1957,
+      "step": 5362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.070219972370451e-08,
+      "loss": 1.0647,
+      "step": 5363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.847575321251576e-08,
+      "loss": 1.2166,
+      "step": 5364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.628491371114543e-08,
+      "loss": 1.2078,
+      "step": 5365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.412968200019131e-08,
+      "loss": 1.0642,
+      "step": 5366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.201005884762801e-08,
+      "loss": 0.9913,
+      "step": 5367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.99260450086736e-08,
+      "loss": 1.0939,
+      "step": 5368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.787764122592298e-08,
+      "loss": 1.1051,
+      "step": 5369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5864848229225666e-08,
+      "loss": 1.1082,
+      "step": 5370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3887666735785716e-08,
+      "loss": 1.1491,
+      "step": 5371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1946097450084054e-08,
+      "loss": 1.0936,
+      "step": 5372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.004014106394506e-08,
+      "loss": 1.0384,
+      "step": 5373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.816979825648105e-08,
+      "loss": 1.0223,
+      "step": 5374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6335069694125597e-08,
+      "loss": 0.9985,
+      "step": 5375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4535956030611336e-08,
+      "loss": 1.1231,
+      "step": 5376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.277245790698103e-08,
+      "loss": 1.3097,
+      "step": 5377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1044575951620924e-08,
+      "loss": 1.1772,
+      "step": 5378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.935231078017187e-08,
+      "loss": 1.1011,
+      "step": 5379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7695662995618216e-08,
+      "loss": 1.1034,
+      "step": 5380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6074633188265536e-08,
+      "loss": 1.0844,
+      "step": 5381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4489221935685156e-08,
+      "loss": 1.113,
+      "step": 5382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2939429802791856e-08,
+      "loss": 1.0624,
+      "step": 5383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1425257341799464e-08,
+      "loss": 1.0436,
+      "step": 5384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.994670509223196e-08,
+      "loss": 1.1982,
+      "step": 5385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8503773580912386e-08,
+      "loss": 1.1944,
+      "step": 5386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7096463321973908e-08,
+      "loss": 1.0543,
+      "step": 5387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5724774816870965e-08,
+      "loss": 1.2243,
+      "step": 5388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.438870855434594e-08,
+      "loss": 1.1814,
+      "step": 5389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3088265010473565e-08,
+      "loss": 1.054,
+      "step": 5390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1823444648605416e-08,
+      "loss": 1.2051,
+      "step": 5391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0594247919414333e-08,
+      "loss": 1.1925,
+      "step": 5392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9400675260883294e-08,
+      "loss": 1.2997,
+      "step": 5393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8242727098305435e-08,
+      "loss": 1.005,
+      "step": 5394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7120403844272937e-08,
+      "loss": 1.0594,
+      "step": 5395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603370589867703e-08,
+      "loss": 1.108,
+      "step": 5396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4982633648730204e-08,
+      "loss": 1.1185,
+      "step": 5397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3967187468932885e-08,
+      "loss": 1.1278,
+      "step": 5398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2987367721128963e-08,
+      "loss": 1.0359,
+      "step": 5399
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2043174754405861e-08,
+      "loss": 1.045,
+      "step": 5400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1134608905227773e-08,
+      "loss": 1.131,
+      "step": 5401
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0261670497313525e-08,
+      "loss": 1.0967,
+      "step": 5402
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.4243598417032e-09,
+      "loss": 1.1691,
+      "step": 5403
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.622677236735932e-09,
+      "loss": 1.1894,
+      "step": 5404
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.856622968072103e-09,
+      "loss": 1.0784,
+      "step": 5405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.126197308671145e-09,
+      "loss": 0.9691,
+      "step": 5406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.4314005187804394e-09,
+      "loss": 1.1681,
+      "step": 5407
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.7722328459797195e-09,
+      "loss": 0.9371,
+      "step": 5408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.148694525125564e-09,
+      "loss": 1.1532,
+      "step": 5409
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5607857784069065e-09,
+      "loss": 1.0514,
+      "step": 5410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0085068153006274e-09,
+      "loss": 0.9596,
+      "step": 5411
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.491857832593759e-09,
+      "loss": 1.0504,
+      "step": 5412
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0108390143612774e-09,
+      "loss": 1.0813,
+      "step": 5413
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.565450532010516e-09,
+      "loss": 1.1206,
+      "step": 5414
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1556925442367535e-09,
+      "loss": 1.0285,
+      "step": 5415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7815651970343182e-09,
+      "loss": 1.1434,
+      "step": 5416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.443068623729893e-09,
+      "loss": 1.1967,
+      "step": 5417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1402029449159024e-09,
+      "loss": 1.0487,
+      "step": 5418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.729682685171269e-10,
+      "loss": 1.0647,
+      "step": 5419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.413646897462933e-10,
+      "loss": 1.1945,
+      "step": 5420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.453922911262787e-10,
+      "loss": 1.0537,
+      "step": 5421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8505114250121367e-10,
+      "loss": 1.1131,
+      "step": 5422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6034130098097067e-10,
+      "loss": 1.135,
+      "step": 5423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.126281101887955e-11,
+      "loss": 0.9845,
+      "step": 5424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.781570434511437e-11,
+      "loss": 1.0743,
+      "step": 5425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.9744,
+      "step": 5426
+    },
+    {
+      "epoch": 1.0,
+      "step": 5426,
+      "total_flos": 4836567874666496.0,
+      "train_loss": 1.177212331827127,
+      "train_runtime": 28954.0997,
+      "train_samples_per_second": 23.99,
+      "train_steps_per_second": 0.187
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5426,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 4836567874666496.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}