diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,103590 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 17260,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.861003861003861e-08,
+      "loss": 2.1562,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.722007722007723e-08,
+      "loss": 1.7266,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1583011583011584e-07,
+      "loss": 1.7734,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5444015444015445e-07,
+      "loss": 1.6562,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9305019305019306e-07,
+      "loss": 1.7422,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3166023166023168e-07,
+      "loss": 1.5,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-07,
+      "loss": 1.9453,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.088803088803089e-07,
+      "loss": 1.8828,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.474903474903475e-07,
+      "loss": 1.625,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8610038610038613e-07,
+      "loss": 1.8594,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2471042471042474e-07,
+      "loss": 1.9141,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6332046332046336e-07,
+      "loss": 1.7969,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.019305019305019e-07,
+      "loss": 1.8594,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 1.75,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.791505791505792e-07,
+      "loss": 1.8516,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.177606177606178e-07,
+      "loss": 1.7188,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.563706563706564e-07,
+      "loss": 1.7422,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.94980694980695e-07,
+      "loss": 1.6406,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.335907335907337e-07,
+      "loss": 1.6562,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.722007722007723e-07,
+      "loss": 1.6328,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.108108108108109e-07,
+      "loss": 1.6719,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.494208494208495e-07,
+      "loss": 1.6328,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.880308880308882e-07,
+      "loss": 1.3984,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.266409266409267e-07,
+      "loss": 1.5625,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.652509652509653e-07,
+      "loss": 1.5469,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0038610038610038e-06,
+      "loss": 1.3984,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0424710424710426e-06,
+      "loss": 1.4766,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 1.4844,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1196911196911197e-06,
+      "loss": 1.4062,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1583011583011585e-06,
+      "loss": 1.4531,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.196911196911197e-06,
+      "loss": 1.4375,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2355212355212356e-06,
+      "loss": 1.4844,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2741312741312742e-06,
+      "loss": 1.3438,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3127413127413127e-06,
+      "loss": 1.3203,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3513513513513515e-06,
+      "loss": 1.3906,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.38996138996139e-06,
+      "loss": 1.3594,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.3906,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4671814671814674e-06,
+      "loss": 1.3047,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.505791505791506e-06,
+      "loss": 1.2266,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5444015444015445e-06,
+      "loss": 1.3516,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.583011583011583e-06,
+      "loss": 1.3281,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 1.2422,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6602316602316604e-06,
+      "loss": 1.1094,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.698841698841699e-06,
+      "loss": 1.2578,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7374517374517377e-06,
+      "loss": 1.1328,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7760617760617763e-06,
+      "loss": 1.1953,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8146718146718149e-06,
+      "loss": 1.2734,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8532818532818534e-06,
+      "loss": 1.2969,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8918918918918922e-06,
+      "loss": 1.1172,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9305019305019305e-06,
+      "loss": 1.2969,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9691119691119693e-06,
+      "loss": 1.2656,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0077220077220077e-06,
+      "loss": 1.1562,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0463320463320464e-06,
+      "loss": 1.2812,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.084942084942085e-06,
+      "loss": 1.1797,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1235521235521236e-06,
+      "loss": 1.1562,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 1.1797,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.200772200772201e-06,
+      "loss": 1.2578,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2393822393822394e-06,
+      "loss": 1.1797,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2779922779922782e-06,
+      "loss": 1.1875,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.316602316602317e-06,
+      "loss": 1.2969,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3552123552123553e-06,
+      "loss": 1.0703,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.393822393822394e-06,
+      "loss": 1.2031,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.432432432432433e-06,
+      "loss": 1.0625,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4710424710424712e-06,
+      "loss": 1.0625,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.50965250965251e-06,
+      "loss": 1.2109,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5482625482625484e-06,
+      "loss": 1.0859,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5868725868725867e-06,
+      "loss": 1.0938,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6254826254826255e-06,
+      "loss": 1.1328,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6640926640926647e-06,
+      "loss": 1.1875,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 1.2031,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7413127413127418e-06,
+      "loss": 1.1094,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.77992277992278e-06,
+      "loss": 1.0234,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8185328185328185e-06,
+      "loss": 1.1797,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.0078,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8957528957528956e-06,
+      "loss": 1.0547,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.934362934362935e-06,
+      "loss": 1.1094,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9729729729729736e-06,
+      "loss": 1.1406,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.011583011583012e-06,
+      "loss": 1.2188,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0501930501930503e-06,
+      "loss": 1.1953,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.088803088803089e-06,
+      "loss": 1.2422,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1274131274131274e-06,
+      "loss": 1.125,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.166023166023166e-06,
+      "loss": 1.1016,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2046332046332054e-06,
+      "loss": 1.1328,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 1.0625,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.281853281853282e-06,
+      "loss": 1.0781,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.320463320463321e-06,
+      "loss": 1.0703,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.359073359073359e-06,
+      "loss": 1.2109,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.397683397683398e-06,
+      "loss": 1.1172,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4362934362934363e-06,
+      "loss": 1.0547,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4749034749034755e-06,
+      "loss": 1.125,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.513513513513514e-06,
+      "loss": 1.1875,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5521235521235526e-06,
+      "loss": 1.3047,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.590733590733591e-06,
+      "loss": 1.2344,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6293436293436297e-06,
+      "loss": 1.0469,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.667953667953668e-06,
+      "loss": 1.1094,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.706563706563707e-06,
+      "loss": 1.0156,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.745173745173745e-06,
+      "loss": 1.0703,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 1.0547,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.822393822393823e-06,
+      "loss": 1.1641,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.861003861003861e-06,
+      "loss": 1.1172,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8996138996139e-06,
+      "loss": 1.125,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.938223938223939e-06,
+      "loss": 1.0938,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.976833976833977e-06,
+      "loss": 1.1094,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.015444015444015e-06,
+      "loss": 0.9688,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0540540540540545e-06,
+      "loss": 1.125,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.092664092664093e-06,
+      "loss": 1.0781,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.131274131274132e-06,
+      "loss": 1.1562,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.16988416988417e-06,
+      "loss": 1.0703,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.208494208494209e-06,
+      "loss": 0.9805,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.247104247104247e-06,
+      "loss": 1.1172,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.8945,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 1.0547,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.362934362934364e-06,
+      "loss": 1.125,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.401544401544402e-06,
+      "loss": 1.1016,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4401544401544405e-06,
+      "loss": 1.0078,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.478764478764479e-06,
+      "loss": 1.0938,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.517374517374517e-06,
+      "loss": 1.0859,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5559845559845564e-06,
+      "loss": 0.9297,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.594594594594596e-06,
+      "loss": 1.1094,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.633204633204634e-06,
+      "loss": 1.0312,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.671814671814672e-06,
+      "loss": 1.1094,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.710424710424711e-06,
+      "loss": 1.1484,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.749034749034749e-06,
+      "loss": 1.0625,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.787644787644788e-06,
+      "loss": 0.9883,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8262548262548266e-06,
+      "loss": 1.0859,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 1.1562,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.903474903474904e-06,
+      "loss": 1.0781,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9420849420849425e-06,
+      "loss": 1.0547,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980694980694981e-06,
+      "loss": 1.0938,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.01930501930502e-06,
+      "loss": 1.0781,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.057915057915058e-06,
+      "loss": 1.125,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.096525096525097e-06,
+      "loss": 1.0703,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.135135135135135e-06,
+      "loss": 1.125,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.173745173745173e-06,
+      "loss": 1.0391,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.212355212355213e-06,
+      "loss": 1.1328,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.250965250965251e-06,
+      "loss": 1.0781,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.28957528957529e-06,
+      "loss": 1.1641,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.328185328185329e-06,
+      "loss": 0.9844,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.366795366795368e-06,
+      "loss": 1.1484,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.1953,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.444015444015444e-06,
+      "loss": 1.0859,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4826254826254836e-06,
+      "loss": 1.0938,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.521235521235522e-06,
+      "loss": 0.9688,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.55984555984556e-06,
+      "loss": 1.0078,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.598455598455599e-06,
+      "loss": 1.1875,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.637065637065637e-06,
+      "loss": 1.0,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.675675675675676e-06,
+      "loss": 0.9844,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.2031,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.752895752895753e-06,
+      "loss": 1.0078,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.791505791505791e-06,
+      "loss": 1.0234,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.83011583011583e-06,
+      "loss": 1.0,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.86872586872587e-06,
+      "loss": 0.9766,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.907335907335908e-06,
+      "loss": 1.0781,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 1.0312,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9845559845559855e-06,
+      "loss": 1.1016,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.023166023166024e-06,
+      "loss": 1.0156,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.061776061776062e-06,
+      "loss": 0.9688,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1003861003861005e-06,
+      "loss": 1.1641,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.13899613899614e-06,
+      "loss": 1.125,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.177606177606178e-06,
+      "loss": 1.0625,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2162162162162164e-06,
+      "loss": 1.0391,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.254826254826255e-06,
+      "loss": 1.2188,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.293436293436294e-06,
+      "loss": 1.0859,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.332046332046332e-06,
+      "loss": 1.0625,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.370656370656371e-06,
+      "loss": 1.125,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.409266409266411e-06,
+      "loss": 0.9375,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.447876447876449e-06,
+      "loss": 0.9102,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 1.0781,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.525096525096526e-06,
+      "loss": 1.0859,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.563706563706564e-06,
+      "loss": 0.8906,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.602316602316603e-06,
+      "loss": 1.0156,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.640926640926642e-06,
+      "loss": 1.0703,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.67953667953668e-06,
+      "loss": 0.918,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.718146718146718e-06,
+      "loss": 1.0,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7567567567567575e-06,
+      "loss": 0.9766,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.795366795366796e-06,
+      "loss": 1.0,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.833976833976834e-06,
+      "loss": 1.0156,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.872586872586873e-06,
+      "loss": 1.1172,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.911196911196911e-06,
+      "loss": 1.0312,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.949806949806951e-06,
+      "loss": 1.1094,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.988416988416989e-06,
+      "loss": 1.0156,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 0.9805,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.065637065637067e-06,
+      "loss": 1.0,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.104247104247105e-06,
+      "loss": 1.0703,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 1.0156,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.181467181467182e-06,
+      "loss": 1.0156,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.22007722007722e-06,
+      "loss": 0.9336,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2586872586872595e-06,
+      "loss": 1.0938,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.297297297297298e-06,
+      "loss": 1.0,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.335907335907336e-06,
+      "loss": 0.9609,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3745173745173745e-06,
+      "loss": 1.1719,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.413127413127414e-06,
+      "loss": 1.0469,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.451737451737452e-06,
+      "loss": 1.0625,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.49034749034749e-06,
+      "loss": 1.0625,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5289575289575304e-06,
+      "loss": 1.0,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 1.0625,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.606177606177607e-06,
+      "loss": 0.9961,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.644787644787645e-06,
+      "loss": 1.0781,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.683397683397685e-06,
+      "loss": 0.9453,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.722007722007722e-06,
+      "loss": 1.1016,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.760617760617761e-06,
+      "loss": 1.0234,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.7992277992278e-06,
+      "loss": 1.0078,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.837837837837838e-06,
+      "loss": 1.0547,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.876447876447877e-06,
+      "loss": 1.0,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.915057915057915e-06,
+      "loss": 0.9883,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.953667953667954e-06,
+      "loss": 1.0625,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.992277992277993e-06,
+      "loss": 0.9297,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.03088803088803e-06,
+      "loss": 1.0078,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.06949806949807e-06,
+      "loss": 1.0703,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 0.8984,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.146718146718148e-06,
+      "loss": 0.9844,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.185328185328186e-06,
+      "loss": 1.1406,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.223938223938225e-06,
+      "loss": 1.0938,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.262548262548264e-06,
+      "loss": 1.0781,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.301158301158302e-06,
+      "loss": 1.0547,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.33976833976834e-06,
+      "loss": 1.0547,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.378378378378378e-06,
+      "loss": 1.0859,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.416988416988418e-06,
+      "loss": 0.9844,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.455598455598457e-06,
+      "loss": 1.0078,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.494208494208494e-06,
+      "loss": 1.1172,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.532818532818533e-06,
+      "loss": 1.0078,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.0469,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.61003861003861e-06,
+      "loss": 0.9609,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 1.1094,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.687258687258689e-06,
+      "loss": 0.9531,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.725868725868728e-06,
+      "loss": 1.0859,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.764478764478765e-06,
+      "loss": 1.0469,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.803088803088804e-06,
+      "loss": 0.9805,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.841698841698842e-06,
+      "loss": 1.0703,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.880308880308881e-06,
+      "loss": 0.8984,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.91891891891892e-06,
+      "loss": 1.1484,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.957528957528958e-06,
+      "loss": 1.0469,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.996138996138997e-06,
+      "loss": 1.0078,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.034749034749034e-06,
+      "loss": 1.0156,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.073359073359074e-06,
+      "loss": 1.1641,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.111969111969113e-06,
+      "loss": 1.125,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.15057915057915e-06,
+      "loss": 1.1016,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 1.0234,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.227799227799229e-06,
+      "loss": 1.0,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.266409266409268e-06,
+      "loss": 1.0938,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.305019305019305e-06,
+      "loss": 0.9883,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.343629343629345e-06,
+      "loss": 0.9961,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.382239382239384e-06,
+      "loss": 1.0547,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.420849420849421e-06,
+      "loss": 1.0078,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.45945945945946e-06,
+      "loss": 1.0234,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.498069498069498e-06,
+      "loss": 1.0,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.536679536679537e-06,
+      "loss": 1.0,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.575289575289576e-06,
+      "loss": 1.0781,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.613899613899614e-06,
+      "loss": 1.1016,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.652509652509653e-06,
+      "loss": 1.1094,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.69111969111969e-06,
+      "loss": 1.0,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 1.0312,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.768339768339769e-06,
+      "loss": 0.8789,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.806949806949808e-06,
+      "loss": 1.2031,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.845559845559847e-06,
+      "loss": 1.0781,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.884169884169885e-06,
+      "loss": 1.1328,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.922779922779924e-06,
+      "loss": 1.0234,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.961389961389962e-06,
+      "loss": 1.0859,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.0625,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.003861003861004e-05,
+      "loss": 0.9883,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0077220077220078e-05,
+      "loss": 1.0391,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0115830115830117e-05,
+      "loss": 1.0234,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0154440154440154e-05,
+      "loss": 1.0547,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0193050193050193e-05,
+      "loss": 1.0625,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0231660231660233e-05,
+      "loss": 1.0312,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 0.9609,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.030888030888031e-05,
+      "loss": 1.0391,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0347490347490347e-05,
+      "loss": 0.9102,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0386100386100386e-05,
+      "loss": 0.9531,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0424710424710425e-05,
+      "loss": 0.9766,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0463320463320463e-05,
+      "loss": 1.0781,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0501930501930502e-05,
+      "loss": 1.125,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0540540540540541e-05,
+      "loss": 1.0156,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.057915057915058e-05,
+      "loss": 1.1406,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.061776061776062e-05,
+      "loss": 1.0469,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0656370656370659e-05,
+      "loss": 0.9531,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0694980694980696e-05,
+      "loss": 0.9531,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0733590733590735e-05,
+      "loss": 0.9414,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0772200772200775e-05,
+      "loss": 1.0312,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.8984,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0849420849420851e-05,
+      "loss": 1.0312,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0888030888030889e-05,
+      "loss": 0.9727,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0926640926640928e-05,
+      "loss": 1.0703,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0965250965250967e-05,
+      "loss": 1.0547,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1003861003861005e-05,
+      "loss": 0.9062,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1042471042471044e-05,
+      "loss": 1.2812,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1081081081081081e-05,
+      "loss": 1.0469,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.111969111969112e-05,
+      "loss": 0.9766,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.115830115830116e-05,
+      "loss": 1.0469,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1196911196911197e-05,
+      "loss": 0.9805,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1235521235521236e-05,
+      "loss": 0.9414,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1274131274131274e-05,
+      "loss": 1.0859,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1312741312741313e-05,
+      "loss": 1.1172,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 1.0156,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.138996138996139e-05,
+      "loss": 1.0391,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.0312,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1467181467181468e-05,
+      "loss": 0.9609,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1505791505791506e-05,
+      "loss": 1.0781,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1544401544401545e-05,
+      "loss": 0.9531,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1583011583011582e-05,
+      "loss": 0.8594,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1621621621621622e-05,
+      "loss": 1.0312,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.166023166023166e-05,
+      "loss": 0.9258,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.16988416988417e-05,
+      "loss": 1.0078,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.173745173745174e-05,
+      "loss": 1.0078,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1776061776061778e-05,
+      "loss": 1.0703,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1814671814671816e-05,
+      "loss": 0.9844,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1853281853281855e-05,
+      "loss": 1.0156,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.9688,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1930501930501932e-05,
+      "loss": 0.9844,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1969111969111971e-05,
+      "loss": 1.0156,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2007722007722008e-05,
+      "loss": 1.1406,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2046332046332048e-05,
+      "loss": 1.0859,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2084942084942087e-05,
+      "loss": 0.9453,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2123552123552124e-05,
+      "loss": 0.9219,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2162162162162164e-05,
+      "loss": 1.1484,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2200772200772201e-05,
+      "loss": 1.0703,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.223938223938224e-05,
+      "loss": 1.0469,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.227799227799228e-05,
+      "loss": 1.0312,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2316602316602317e-05,
+      "loss": 0.8594,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2355212355212356e-05,
+      "loss": 0.9375,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2393822393822394e-05,
+      "loss": 0.9766,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 0.9609,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2471042471042472e-05,
+      "loss": 0.9531,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.250965250965251e-05,
+      "loss": 1.0312,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2548262548262549e-05,
+      "loss": 1.0703,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2586872586872588e-05,
+      "loss": 1.0391,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2625482625482625e-05,
+      "loss": 1.0781,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2664092664092665e-05,
+      "loss": 0.9531,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2702702702702702e-05,
+      "loss": 1.0078,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2741312741312741e-05,
+      "loss": 1.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2779922779922782e-05,
+      "loss": 0.9961,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2818532818532821e-05,
+      "loss": 1.1094,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 1.0938,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2895752895752898e-05,
+      "loss": 0.9062,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2934362934362936e-05,
+      "loss": 1.125,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 0.9844,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3011583011583014e-05,
+      "loss": 0.9805,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3050193050193052e-05,
+      "loss": 0.9961,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.308880308880309e-05,
+      "loss": 0.9375,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3127413127413128e-05,
+      "loss": 1.0469,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3166023166023167e-05,
+      "loss": 1.0156,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3204633204633207e-05,
+      "loss": 1.1094,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3243243243243244e-05,
+      "loss": 1.0703,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3281853281853283e-05,
+      "loss": 1.0312,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.332046332046332e-05,
+      "loss": 0.9648,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.335907335907336e-05,
+      "loss": 1.0781,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.33976833976834e-05,
+      "loss": 0.8867,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3436293436293437e-05,
+      "loss": 0.9453,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3474903474903476e-05,
+      "loss": 1.0156,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 1.0859,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3552123552123553e-05,
+      "loss": 0.9766,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3590733590733592e-05,
+      "loss": 1.0156,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.362934362934363e-05,
+      "loss": 1.1328,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3667953667953668e-05,
+      "loss": 1.0547,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3706563706563708e-05,
+      "loss": 1.0938,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3745173745173745e-05,
+      "loss": 1.0078,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3783783783783784e-05,
+      "loss": 1.0391,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3822393822393822e-05,
+      "loss": 0.8281,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3861003861003861e-05,
+      "loss": 1.0938,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3899613899613902e-05,
+      "loss": 1.0469,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3938223938223941e-05,
+      "loss": 0.9375,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3976833976833979e-05,
+      "loss": 1.0703,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4015444015444018e-05,
+      "loss": 0.9883,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 0.9531,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4092664092664095e-05,
+      "loss": 1.2109,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4131274131274134e-05,
+      "loss": 1.0,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4169884169884171e-05,
+      "loss": 1.0469,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.420849420849421e-05,
+      "loss": 0.9727,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4247104247104248e-05,
+      "loss": 1.0938,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.9648,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4324324324324326e-05,
+      "loss": 0.9648,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4362934362934364e-05,
+      "loss": 0.9336,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4401544401544403e-05,
+      "loss": 0.9219,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.444015444015444e-05,
+      "loss": 0.8516,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.447876447876448e-05,
+      "loss": 1.0938,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4517374517374519e-05,
+      "loss": 0.8398,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4555984555984556e-05,
+      "loss": 0.9922,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 0.9766,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4633204633204635e-05,
+      "loss": 1.0078,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4671814671814672e-05,
+      "loss": 1.0547,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4710424710424711e-05,
+      "loss": 1.0078,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4749034749034749e-05,
+      "loss": 1.0547,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4787644787644788e-05,
+      "loss": 1.0859,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4826254826254827e-05,
+      "loss": 0.9453,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4864864864864865e-05,
+      "loss": 1.0703,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4903474903474904e-05,
+      "loss": 0.9688,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4942084942084942e-05,
+      "loss": 1.0469,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.498069498069498e-05,
+      "loss": 1.1797,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5019305019305022e-05,
+      "loss": 0.9531,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5057915057915061e-05,
+      "loss": 0.9766,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5096525096525098e-05,
+      "loss": 1.0312,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 0.9609,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5173745173745175e-05,
+      "loss": 1.0156,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5212355212355214e-05,
+      "loss": 0.9922,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5250965250965253e-05,
+      "loss": 1.1094,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.528957528957529e-05,
+      "loss": 1.0156,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.532818532818533e-05,
+      "loss": 1.0938,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.536679536679537e-05,
+      "loss": 1.0,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.540540540540541e-05,
+      "loss": 0.9648,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5444015444015444e-05,
+      "loss": 1.0234,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5482625482625484e-05,
+      "loss": 0.9648,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5521235521235523e-05,
+      "loss": 1.0781,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5559845559845562e-05,
+      "loss": 0.9336,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.55984555984556e-05,
+      "loss": 1.0938,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5637065637065637e-05,
+      "loss": 1.0391,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 1.0234,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 1.0156,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5752895752895755e-05,
+      "loss": 1.0469,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5791505791505794e-05,
+      "loss": 0.9688,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.583011583011583e-05,
+      "loss": 1.0156,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.586872586872587e-05,
+      "loss": 0.9883,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5907335907335908e-05,
+      "loss": 0.9609,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5945945945945947e-05,
+      "loss": 0.9883,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5984555984555986e-05,
+      "loss": 0.8281,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6023166023166022e-05,
+      "loss": 0.9961,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.606177606177606e-05,
+      "loss": 1.0312,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.61003861003861e-05,
+      "loss": 1.0938,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.613899613899614e-05,
+      "loss": 0.9141,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.617760617760618e-05,
+      "loss": 0.9883,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.9961,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6254826254826257e-05,
+      "loss": 1.0547,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6293436293436296e-05,
+      "loss": 1.1094,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6332046332046336e-05,
+      "loss": 0.9609,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.637065637065637e-05,
+      "loss": 1.0234,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.640926640926641e-05,
+      "loss": 1.0938,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.644787644787645e-05,
+      "loss": 1.1641,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.648648648648649e-05,
+      "loss": 0.9609,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6525096525096528e-05,
+      "loss": 1.0703,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6563706563706564e-05,
+      "loss": 1.0,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6602316602316603e-05,
+      "loss": 0.9336,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6640926640926642e-05,
+      "loss": 1.1406,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.667953667953668e-05,
+      "loss": 1.0781,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.671814671814672e-05,
+      "loss": 1.0156,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 0.9961,
+      "step": 434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6795366795366796e-05,
+      "loss": 1.0703,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6833976833976835e-05,
+      "loss": 1.0234,
+      "step": 436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6872586872586874e-05,
+      "loss": 1.0547,
+      "step": 437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6911196911196913e-05,
+      "loss": 0.9453,
+      "step": 438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.694980694980695e-05,
+      "loss": 0.9219,
+      "step": 439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.698841698841699e-05,
+      "loss": 1.0391,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7027027027027028e-05,
+      "loss": 1.0391,
+      "step": 441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7065637065637067e-05,
+      "loss": 1.0781,
+      "step": 442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7104247104247106e-05,
+      "loss": 0.9844,
+      "step": 443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.0234,
+      "step": 444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.718146718146718e-05,
+      "loss": 0.9609,
+      "step": 445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.722007722007722e-05,
+      "loss": 1.0781,
+      "step": 446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7258687258687263e-05,
+      "loss": 1.0391,
+      "step": 447
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 1.0156,
+      "step": 448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7335907335907338e-05,
+      "loss": 0.8867,
+      "step": 449
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7374517374517377e-05,
+      "loss": 1.0938,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7413127413127416e-05,
+      "loss": 1.0391,
+      "step": 451
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7451737451737455e-05,
+      "loss": 1.0234,
+      "step": 452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.749034749034749e-05,
+      "loss": 0.8828,
+      "step": 453
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.752895752895753e-05,
+      "loss": 1.0078,
+      "step": 454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.756756756756757e-05,
+      "loss": 1.0859,
+      "step": 455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.760617760617761e-05,
+      "loss": 1.0859,
+      "step": 456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7644787644787648e-05,
+      "loss": 0.9453,
+      "step": 457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7683397683397684e-05,
+      "loss": 0.9766,
+      "step": 458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7722007722007723e-05,
+      "loss": 1.0312,
+      "step": 459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7760617760617762e-05,
+      "loss": 0.9219,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.77992277992278e-05,
+      "loss": 1.0859,
+      "step": 461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 1.0078,
+      "step": 462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7876447876447876e-05,
+      "loss": 1.0078,
+      "step": 463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7915057915057916e-05,
+      "loss": 0.8789,
+      "step": 464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7953667953667955e-05,
+      "loss": 1.0703,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7992277992277994e-05,
+      "loss": 1.1328,
+      "step": 466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8030888030888033e-05,
+      "loss": 1.0781,
+      "step": 467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.806949806949807e-05,
+      "loss": 1.0312,
+      "step": 468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8108108108108108e-05,
+      "loss": 0.9883,
+      "step": 469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8146718146718147e-05,
+      "loss": 1.0938,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8185328185328187e-05,
+      "loss": 1.1016,
+      "step": 471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8223938223938226e-05,
+      "loss": 1.0156,
+      "step": 472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.826254826254826e-05,
+      "loss": 0.9844,
+      "step": 473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.83011583011583e-05,
+      "loss": 1.0469,
+      "step": 474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8339768339768343e-05,
+      "loss": 0.9492,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 0.9219,
+      "step": 476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.841698841698842e-05,
+      "loss": 0.9414,
+      "step": 477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8455598455598458e-05,
+      "loss": 0.9844,
+      "step": 478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8494208494208497e-05,
+      "loss": 1.0625,
+      "step": 479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8532818532818536e-05,
+      "loss": 1.0156,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.9414,
+      "step": 481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.861003861003861e-05,
+      "loss": 0.9727,
+      "step": 482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.864864864864865e-05,
+      "loss": 1.1016,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.868725868725869e-05,
+      "loss": 1.0234,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.872586872586873e-05,
+      "loss": 0.9023,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8764478764478768e-05,
+      "loss": 1.1406,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8803088803088804e-05,
+      "loss": 1.0156,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8841698841698843e-05,
+      "loss": 1.0938,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8880308880308882e-05,
+      "loss": 1.0391,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.9961,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.895752895752896e-05,
+      "loss": 0.9688,
+      "step": 491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8996138996138996e-05,
+      "loss": 1.0703,
+      "step": 492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9034749034749035e-05,
+      "loss": 1.0078,
+      "step": 493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9073359073359075e-05,
+      "loss": 1.0938,
+      "step": 494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9111969111969114e-05,
+      "loss": 0.9844,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9150579150579153e-05,
+      "loss": 0.9961,
+      "step": 496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.918918918918919e-05,
+      "loss": 1.0156,
+      "step": 497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9227799227799228e-05,
+      "loss": 1.2031,
+      "step": 498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9266409266409267e-05,
+      "loss": 1.1562,
+      "step": 499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9305019305019306e-05,
+      "loss": 1.0391,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9343629343629345e-05,
+      "loss": 0.9609,
+      "step": 501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.938223938223938e-05,
+      "loss": 1.2969,
+      "step": 502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.942084942084942e-05,
+      "loss": 1.1016,
+      "step": 503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 1.0391,
+      "step": 504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9498069498069502e-05,
+      "loss": 0.9844,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9536679536679538e-05,
+      "loss": 1.125,
+      "step": 506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9575289575289577e-05,
+      "loss": 0.9883,
+      "step": 507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9613899613899616e-05,
+      "loss": 1.1484,
+      "step": 508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9652509652509656e-05,
+      "loss": 0.8398,
+      "step": 509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9691119691119695e-05,
+      "loss": 0.9961,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.972972972972973e-05,
+      "loss": 1.0156,
+      "step": 511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.976833976833977e-05,
+      "loss": 0.9961,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.980694980694981e-05,
+      "loss": 1.0547,
+      "step": 513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9845559845559848e-05,
+      "loss": 1.1719,
+      "step": 514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9884169884169887e-05,
+      "loss": 0.9648,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9922779922779923e-05,
+      "loss": 0.9414,
+      "step": 516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961389961389962e-05,
+      "loss": 1.0312,
+      "step": 517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9961,
+      "step": 518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999823942282e-05,
+      "loss": 1.1172,
+      "step": 519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999929576913e-05,
+      "loss": 0.9062,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999841548056e-05,
+      "loss": 0.9414,
+      "step": 521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999997183076604e-05,
+      "loss": 0.9219,
+      "step": 522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999559855731e-05,
+      "loss": 1.0078,
+      "step": 523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993661922735e-05,
+      "loss": 0.9688,
+      "step": 524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999137317294e-05,
+      "loss": 0.8633,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999988732308005e-05,
+      "loss": 1.1797,
+      "step": 526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998573932803e-05,
+      "loss": 1.0781,
+      "step": 527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999982394233117e-05,
+      "loss": 0.957,
+      "step": 528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999978697023387e-05,
+      "loss": 0.9766,
+      "step": 529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997464769896e-05,
+      "loss": 0.9883,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999970246259992e-05,
+      "loss": 1.1094,
+      "step": 531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996549270663e-05,
+      "loss": 0.9922,
+      "step": 532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999960387039043e-05,
+      "loss": 1.0078,
+      "step": 533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995492925741e-05,
+      "loss": 1.0,
+      "step": 534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999949119361927e-05,
+      "loss": 1.0078,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999994295735279e-05,
+      "loss": 0.9258,
+      "step": 536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999936443230227e-05,
+      "loss": 1.0859,
+      "step": 537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999992957699446e-05,
+      "loss": 0.9766,
+      "step": 538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999922358645732e-05,
+      "loss": 0.9531,
+      "step": 539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999147881843e-05,
+      "loss": 1.0078,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999906865610427e-05,
+      "loss": 0.9492,
+      "step": 541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999898590924393e-05,
+      "loss": 0.957,
+      "step": 542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999889964126492e-05,
+      "loss": 1.1016,
+      "step": 543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999988098521702e-05,
+      "loss": 1.0391,
+      "step": 544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999871654196307e-05,
+      "loss": 1.0469,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999861971064668e-05,
+      "loss": 0.9336,
+      "step": 546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999851935822448e-05,
+      "loss": 1.0234,
+      "step": 547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999841548470005e-05,
+      "loss": 0.9766,
+      "step": 548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99998308090077e-05,
+      "loss": 1.0938,
+      "step": 549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999981971743591e-05,
+      "loss": 0.9688,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999980827375503e-05,
+      "loss": 0.9531,
+      "step": 551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999796477965466e-05,
+      "loss": 1.0,
+      "step": 552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999784330067622e-05,
+      "loss": 0.9883,
+      "step": 553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999771830061933e-05,
+      "loss": 0.9648,
+      "step": 554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999975897794884e-05,
+      "loss": 1.0391,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999745773728792e-05,
+      "loss": 1.0,
+      "step": 556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999973221740226e-05,
+      "loss": 1.1641,
+      "step": 557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999718308969715e-05,
+      "loss": 1.1172,
+      "step": 558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999970404843165e-05,
+      "loss": 1.1328,
+      "step": 559
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999689435788562e-05,
+      "loss": 1.0469,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999967447104097e-05,
+      "loss": 1.125,
+      "step": 561
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999659154189404e-05,
+      "loss": 0.9609,
+      "step": 562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999643485234396e-05,
+      "loss": 1.0703,
+      "step": 563
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999627464176503e-05,
+      "loss": 0.9844,
+      "step": 564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999611091016286e-05,
+      "loss": 1.0469,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999594365754323e-05,
+      "loss": 1.0859,
+      "step": 566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999577288391206e-05,
+      "loss": 0.9688,
+      "step": 567
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999559858927535e-05,
+      "loss": 0.8828,
+      "step": 568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999954207736392e-05,
+      "loss": 1.0078,
+      "step": 569
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999523943700986e-05,
+      "loss": 0.9297,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999950545793938e-05,
+      "loss": 1.0234,
+      "step": 571
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999948662007974e-05,
+      "loss": 0.9453,
+      "step": 572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999467430122745e-05,
+      "loss": 1.0859,
+      "step": 573
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999944788806906e-05,
+      "loss": 0.9961,
+      "step": 574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999427993919374e-05,
+      "loss": 0.8828,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999940774767439e-05,
+      "loss": 0.9492,
+      "step": 576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999938714933482e-05,
+      "loss": 1.0625,
+      "step": 577
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999366198901392e-05,
+      "loss": 1.0391,
+      "step": 578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999344896374836e-05,
+      "loss": 0.8867,
+      "step": 579
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999323241755912e-05,
+      "loss": 0.8945,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999301235045378e-05,
+      "loss": 0.9961,
+      "step": 581
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999278876244003e-05,
+      "loss": 0.9688,
+      "step": 582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999256165352588e-05,
+      "loss": 0.9531,
+      "step": 583
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999233102371918e-05,
+      "loss": 0.9688,
+      "step": 584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999209687302813e-05,
+      "loss": 0.9727,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99991859201461e-05,
+      "loss": 0.9805,
+      "step": 586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999161800902608e-05,
+      "loss": 1.0781,
+      "step": 587
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999137329573192e-05,
+      "loss": 0.9805,
+      "step": 588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999911250615871e-05,
+      "loss": 1.1172,
+      "step": 589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999908733066004e-05,
+      "loss": 1.0703,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999061803078066e-05,
+      "loss": 1.1328,
+      "step": 591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999035923413688e-05,
+      "loss": 0.9453,
+      "step": 592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999009691667813e-05,
+      "loss": 1.0625,
+      "step": 593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998983107841375e-05,
+      "loss": 1.1328,
+      "step": 594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99989561719353e-05,
+      "loss": 0.9844,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999892888395054e-05,
+      "loss": 0.9492,
+      "step": 596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998901243888055e-05,
+      "loss": 0.9961,
+      "step": 597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999887325174882e-05,
+      "loss": 1.0156,
+      "step": 598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999884490753382e-05,
+      "loss": 1.0703,
+      "step": 599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999881621124405e-05,
+      "loss": 0.8867,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998787162880523e-05,
+      "loss": 1.0781,
+      "step": 601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998757762444264e-05,
+      "loss": 0.9141,
+      "step": 602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998728009936305e-05,
+      "loss": 1.0547,
+      "step": 603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99986979053577e-05,
+      "loss": 0.8594,
+      "step": 604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998667448709495e-05,
+      "loss": 0.9766,
+      "step": 605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998636639992776e-05,
+      "loss": 1.0703,
+      "step": 606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998605479208625e-05,
+      "loss": 1.1719,
+      "step": 607
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998573966358132e-05,
+      "loss": 1.0,
+      "step": 608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998542101442417e-05,
+      "loss": 1.0078,
+      "step": 609
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998509884462594e-05,
+      "loss": 1.1016,
+      "step": 610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99984773154198e-05,
+      "loss": 1.0469,
+      "step": 611
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998444394315184e-05,
+      "loss": 0.9727,
+      "step": 612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998411121149904e-05,
+      "loss": 0.9805,
+      "step": 613
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998377495925128e-05,
+      "loss": 1.0859,
+      "step": 614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998343518642043e-05,
+      "loss": 1.0156,
+      "step": 615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998309189301848e-05,
+      "loss": 0.9219,
+      "step": 616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998274507905746e-05,
+      "loss": 1.0156,
+      "step": 617
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998239474454963e-05,
+      "loss": 0.9062,
+      "step": 618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998204088950733e-05,
+      "loss": 1.0156,
+      "step": 619
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998168351394295e-05,
+      "loss": 1.0312,
+      "step": 620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998132261786915e-05,
+      "loss": 1.0078,
+      "step": 621
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999809582012986e-05,
+      "loss": 1.1094,
+      "step": 622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998059026424418e-05,
+      "loss": 1.0547,
+      "step": 623
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998021880671876e-05,
+      "loss": 0.8594,
+      "step": 624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997984382873546e-05,
+      "loss": 1.0156,
+      "step": 625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997946533030755e-05,
+      "loss": 1.0156,
+      "step": 626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997908331144823e-05,
+      "loss": 1.0312,
+      "step": 627
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997869777217106e-05,
+      "loss": 1.0469,
+      "step": 628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997830871248957e-05,
+      "loss": 1.0781,
+      "step": 629
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997791613241746e-05,
+      "loss": 1.0469,
+      "step": 630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997752003196854e-05,
+      "loss": 0.9062,
+      "step": 631
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997712041115683e-05,
+      "loss": 1.1016,
+      "step": 632
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997671726999628e-05,
+      "loss": 0.9727,
+      "step": 633
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997631060850116e-05,
+      "loss": 1.125,
+      "step": 634
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997590042668582e-05,
+      "loss": 0.9922,
+      "step": 635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997548672456463e-05,
+      "loss": 0.9141,
+      "step": 636
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997506950215224e-05,
+      "loss": 0.9219,
+      "step": 637
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997464875946323e-05,
+      "loss": 1.0391,
+      "step": 638
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999742244965125e-05,
+      "loss": 1.0547,
+      "step": 639
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99973796713315e-05,
+      "loss": 0.9453,
+      "step": 640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999733654098857e-05,
+      "loss": 0.9648,
+      "step": 641
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997293058623987e-05,
+      "loss": 0.9805,
+      "step": 642
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999724922423928e-05,
+      "loss": 1.0312,
+      "step": 643
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997205037835995e-05,
+      "loss": 0.9766,
+      "step": 644
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999716049941568e-05,
+      "loss": 0.918,
+      "step": 645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999711560897991e-05,
+      "loss": 1.0625,
+      "step": 646
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997070366530265e-05,
+      "loss": 1.0469,
+      "step": 647
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997024772068333e-05,
+      "loss": 1.0,
+      "step": 648
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999697882559573e-05,
+      "loss": 1.0625,
+      "step": 649
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996932527114064e-05,
+      "loss": 0.9141,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996885876624965e-05,
+      "loss": 1.0703,
+      "step": 651
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996838874130082e-05,
+      "loss": 0.9648,
+      "step": 652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996791519631067e-05,
+      "loss": 0.8789,
+      "step": 653
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999674381312959e-05,
+      "loss": 1.0938,
+      "step": 654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996695754627325e-05,
+      "loss": 0.9844,
+      "step": 655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999664734412597e-05,
+      "loss": 1.0469,
+      "step": 656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996598581627227e-05,
+      "loss": 1.1094,
+      "step": 657
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996549467132814e-05,
+      "loss": 1.0,
+      "step": 658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996500000644457e-05,
+      "loss": 1.2266,
+      "step": 659
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99964501821639e-05,
+      "loss": 1.1016,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99964000116929e-05,
+      "loss": 0.9648,
+      "step": 661
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999634948923322e-05,
+      "loss": 1.0859,
+      "step": 662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996298614786642e-05,
+      "loss": 1.0547,
+      "step": 663
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996247388354956e-05,
+      "loss": 1.1406,
+      "step": 664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999619580993996e-05,
+      "loss": 0.9609,
+      "step": 665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999614387954348e-05,
+      "loss": 0.9844,
+      "step": 666
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996091597167337e-05,
+      "loss": 0.9297,
+      "step": 667
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996038962813376e-05,
+      "loss": 0.8789,
+      "step": 668
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995985976483453e-05,
+      "loss": 1.0469,
+      "step": 669
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995932638179427e-05,
+      "loss": 1.0781,
+      "step": 670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995878947903176e-05,
+      "loss": 1.0469,
+      "step": 671
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99958249056566e-05,
+      "loss": 1.0469,
+      "step": 672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999577051144159e-05,
+      "loss": 1.0469,
+      "step": 673
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999571576526007e-05,
+      "loss": 1.0312,
+      "step": 674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995660667113964e-05,
+      "loss": 1.0469,
+      "step": 675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995605217005215e-05,
+      "loss": 0.8984,
+      "step": 676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995549414935768e-05,
+      "loss": 0.9258,
+      "step": 677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995493260907594e-05,
+      "loss": 0.9453,
+      "step": 678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999543675492267e-05,
+      "loss": 1.0312,
+      "step": 679
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995379896982986e-05,
+      "loss": 0.9258,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995322687090547e-05,
+      "loss": 1.0625,
+      "step": 681
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995265125247356e-05,
+      "loss": 0.9883,
+      "step": 682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995207211455453e-05,
+      "loss": 0.9961,
+      "step": 683
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995148945716865e-05,
+      "loss": 1.0156,
+      "step": 684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999509032803366e-05,
+      "loss": 0.9883,
+      "step": 685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995031358407882e-05,
+      "loss": 0.9727,
+      "step": 686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994972036841624e-05,
+      "loss": 1.0781,
+      "step": 687
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994912363336968e-05,
+      "loss": 0.9531,
+      "step": 688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999485233789601e-05,
+      "loss": 0.957,
+      "step": 689
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994791960520877e-05,
+      "loss": 1.0703,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999473123121368e-05,
+      "loss": 0.9922,
+      "step": 691
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999467014997657e-05,
+      "loss": 1.0859,
+      "step": 692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994608716811686e-05,
+      "loss": 1.1016,
+      "step": 693
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994546931721204e-05,
+      "loss": 1.0,
+      "step": 694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994484794707293e-05,
+      "loss": 0.9766,
+      "step": 695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994422305772138e-05,
+      "loss": 0.9102,
+      "step": 696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999435946491794e-05,
+      "loss": 1.0703,
+      "step": 697
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994296272146917e-05,
+      "loss": 0.8438,
+      "step": 698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999423272746129e-05,
+      "loss": 1.0312,
+      "step": 699
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99941688308633e-05,
+      "loss": 0.9492,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994104582355192e-05,
+      "loss": 1.0469,
+      "step": 701
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994039981939233e-05,
+      "loss": 0.9375,
+      "step": 702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993975029617693e-05,
+      "loss": 1.0547,
+      "step": 703
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993909725392863e-05,
+      "loss": 0.9453,
+      "step": 704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999384406926704e-05,
+      "loss": 0.9844,
+      "step": 705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999377806124254e-05,
+      "loss": 1.0156,
+      "step": 706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993711701321682e-05,
+      "loss": 1.0391,
+      "step": 707
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993644989506804e-05,
+      "loss": 1.1719,
+      "step": 708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993577925800257e-05,
+      "loss": 0.9492,
+      "step": 709
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99935105102044e-05,
+      "loss": 1.0156,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993442742721608e-05,
+      "loss": 0.9844,
+      "step": 711
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993374623354266e-05,
+      "loss": 1.0625,
+      "step": 712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993306152104776e-05,
+      "loss": 1.0312,
+      "step": 713
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993237328975547e-05,
+      "loss": 0.9961,
+      "step": 714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993168153969003e-05,
+      "loss": 0.9805,
+      "step": 715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993098627087576e-05,
+      "loss": 0.9922,
+      "step": 716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993028748333717e-05,
+      "loss": 1.1094,
+      "step": 717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999295851770989e-05,
+      "loss": 1.0391,
+      "step": 718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992887935218558e-05,
+      "loss": 0.9727,
+      "step": 719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992817000862218e-05,
+      "loss": 0.8945,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992745714643364e-05,
+      "loss": 0.9336,
+      "step": 721
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992674076564502e-05,
+      "loss": 0.957,
+      "step": 722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992602086628158e-05,
+      "loss": 1.1172,
+      "step": 723
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992529744836865e-05,
+      "loss": 0.9922,
+      "step": 724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999245705119317e-05,
+      "loss": 1.0391,
+      "step": 725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992384005699635e-05,
+      "loss": 0.9531,
+      "step": 726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999231060835883e-05,
+      "loss": 0.9375,
+      "step": 727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992236859173345e-05,
+      "loss": 0.9727,
+      "step": 728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992162758145768e-05,
+      "loss": 1.0391,
+      "step": 729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992088305278713e-05,
+      "loss": 1.0625,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99920135005748e-05,
+      "loss": 0.8516,
+      "step": 731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991938344036667e-05,
+      "loss": 0.9062,
+      "step": 732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991862835666958e-05,
+      "loss": 0.9258,
+      "step": 733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991786975468332e-05,
+      "loss": 0.9336,
+      "step": 734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999171076344345e-05,
+      "loss": 1.0938,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991634199595013e-05,
+      "loss": 0.8281,
+      "step": 736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991557283925707e-05,
+      "loss": 0.9805,
+      "step": 737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999148001643824e-05,
+      "loss": 0.957,
+      "step": 738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991402397135343e-05,
+      "loss": 0.9648,
+      "step": 739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991324426019732e-05,
+      "loss": 0.9375,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991246103094167e-05,
+      "loss": 0.9414,
+      "step": 741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991167428361396e-05,
+      "loss": 1.0234,
+      "step": 742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991088401824193e-05,
+      "loss": 1.0938,
+      "step": 743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991009023485347e-05,
+      "loss": 1.0625,
+      "step": 744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999092929334764e-05,
+      "loss": 0.8984,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990849211413893e-05,
+      "loss": 0.9102,
+      "step": 746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990768777686916e-05,
+      "loss": 0.9258,
+      "step": 747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990687992169545e-05,
+      "loss": 0.9023,
+      "step": 748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990606854864625e-05,
+      "loss": 0.9375,
+      "step": 749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999052536577501e-05,
+      "loss": 1.0703,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990443524903575e-05,
+      "loss": 1.0938,
+      "step": 751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990361332253198e-05,
+      "loss": 1.0625,
+      "step": 752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990278787826768e-05,
+      "loss": 0.9453,
+      "step": 753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990195891627203e-05,
+      "loss": 1.0391,
+      "step": 754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990112643657412e-05,
+      "loss": 0.9414,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999002904392033e-05,
+      "loss": 0.9336,
+      "step": 756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989945092418904e-05,
+      "loss": 1.0078,
+      "step": 757
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989860789156084e-05,
+      "loss": 1.0625,
+      "step": 758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989776134134844e-05,
+      "loss": 1.0,
+      "step": 759
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998969112735816e-05,
+      "loss": 0.9219,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989605768829027e-05,
+      "loss": 0.9375,
+      "step": 761
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989520058550455e-05,
+      "loss": 1.125,
+      "step": 762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989433996525456e-05,
+      "loss": 1.0078,
+      "step": 763
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989347582757062e-05,
+      "loss": 0.9219,
+      "step": 764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989260817248313e-05,
+      "loss": 1.1094,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989173700002272e-05,
+      "loss": 0.9102,
+      "step": 766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989086231022e-05,
+      "loss": 1.0547,
+      "step": 767
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988998410310575e-05,
+      "loss": 1.0078,
+      "step": 768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99889102378711e-05,
+      "loss": 1.0078,
+      "step": 769
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988821713706666e-05,
+      "loss": 0.9922,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988732837820398e-05,
+      "loss": 0.9766,
+      "step": 771
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998864361021543e-05,
+      "loss": 0.875,
+      "step": 772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988554030894894e-05,
+      "loss": 0.9844,
+      "step": 773
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988464099861947e-05,
+      "loss": 1.0,
+      "step": 774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988373817119758e-05,
+      "loss": 0.9023,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988283182671505e-05,
+      "loss": 0.9141,
+      "step": 776
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988192196520376e-05,
+      "loss": 0.9609,
+      "step": 777
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988100858669583e-05,
+      "loss": 0.8906,
+      "step": 778
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988009169122336e-05,
+      "loss": 1.0547,
+      "step": 779
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987917127881865e-05,
+      "loss": 0.875,
+      "step": 780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987824734951407e-05,
+      "loss": 0.9648,
+      "step": 781
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987731990334225e-05,
+      "loss": 0.9883,
+      "step": 782
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987638894033578e-05,
+      "loss": 0.9414,
+      "step": 783
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998754544605274e-05,
+      "loss": 1.0781,
+      "step": 784
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998745164639501e-05,
+      "loss": 1.0156,
+      "step": 785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987357495063683e-05,
+      "loss": 0.9648,
+      "step": 786
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998726299206208e-05,
+      "loss": 1.0547,
+      "step": 787
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987168137393528e-05,
+      "loss": 0.9805,
+      "step": 788
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987072931061364e-05,
+      "loss": 1.0,
+      "step": 789
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986977373068944e-05,
+      "loss": 0.9844,
+      "step": 790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986881463419628e-05,
+      "loss": 0.9453,
+      "step": 791
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986785202116795e-05,
+      "loss": 1.0156,
+      "step": 792
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998668858916384e-05,
+      "loss": 0.9961,
+      "step": 793
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998659162456416e-05,
+      "loss": 0.9844,
+      "step": 794
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986494308321168e-05,
+      "loss": 0.9727,
+      "step": 795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986396640438292e-05,
+      "loss": 1.0,
+      "step": 796
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986298620918972e-05,
+      "loss": 0.9297,
+      "step": 797
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986200249766657e-05,
+      "loss": 0.957,
+      "step": 798
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986101526984815e-05,
+      "loss": 0.9609,
+      "step": 799
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986002452576917e-05,
+      "loss": 0.8789,
+      "step": 800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998590302654646e-05,
+      "loss": 1.0625,
+      "step": 801
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985803248896934e-05,
+      "loss": 1.0469,
+      "step": 802
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998570311963186e-05,
+      "loss": 1.0,
+      "step": 803
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985602638754758e-05,
+      "loss": 0.9688,
+      "step": 804
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998550180626917e-05,
+      "loss": 0.8516,
+      "step": 805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998540062217865e-05,
+      "loss": 0.9375,
+      "step": 806
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985299086486754e-05,
+      "loss": 1.0156,
+      "step": 807
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985197199197058e-05,
+      "loss": 0.8672,
+      "step": 808
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985094960313155e-05,
+      "loss": 1.0234,
+      "step": 809
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998499236983864e-05,
+      "loss": 1.0234,
+      "step": 810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984889427777126e-05,
+      "loss": 1.0234,
+      "step": 811
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998478613413224e-05,
+      "loss": 0.9531,
+      "step": 812
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998468248890762e-05,
+      "loss": 1.0,
+      "step": 813
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998457849210691e-05,
+      "loss": 0.9492,
+      "step": 814
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984474143733773e-05,
+      "loss": 0.9766,
+      "step": 815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984369443791888e-05,
+      "loss": 0.9375,
+      "step": 816
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998426439228494e-05,
+      "loss": 0.9922,
+      "step": 817
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984158989216626e-05,
+      "loss": 0.9766,
+      "step": 818
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998405323459066e-05,
+      "loss": 1.0234,
+      "step": 819
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983947128410763e-05,
+      "loss": 1.0078,
+      "step": 820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998384067068067e-05,
+      "loss": 1.0156,
+      "step": 821
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983733861404135e-05,
+      "loss": 0.8633,
+      "step": 822
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983626700584915e-05,
+      "loss": 1.0625,
+      "step": 823
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983519188226785e-05,
+      "loss": 1.1016,
+      "step": 824
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998341132433353e-05,
+      "loss": 0.9102,
+      "step": 825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983303108908946e-05,
+      "loss": 0.9727,
+      "step": 826
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998319454195685e-05,
+      "loss": 0.9883,
+      "step": 827
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983085623481055e-05,
+      "loss": 0.9258,
+      "step": 828
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99829763534854e-05,
+      "loss": 1.0625,
+      "step": 829
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998286673197374e-05,
+      "loss": 0.8594,
+      "step": 830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982756758949928e-05,
+      "loss": 0.7969,
+      "step": 831
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998264643441783e-05,
+      "loss": 0.9453,
+      "step": 832
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982535758381344e-05,
+      "loss": 1.1016,
+      "step": 833
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998242473084436e-05,
+      "loss": 0.9336,
+      "step": 834
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998231335181079e-05,
+      "loss": 1.125,
+      "step": 835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998220162128455e-05,
+      "loss": 1.0469,
+      "step": 836
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982089539269583e-05,
+      "loss": 1.0312,
+      "step": 837
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998197710576983e-05,
+      "loss": 0.9883,
+      "step": 838
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981864320789247e-05,
+      "loss": 0.9141,
+      "step": 839
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981751184331813e-05,
+      "loss": 0.9609,
+      "step": 840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981637696401507e-05,
+      "loss": 0.9219,
+      "step": 841
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981523857002326e-05,
+      "loss": 1.0938,
+      "step": 842
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998140966613828e-05,
+      "loss": 1.0156,
+      "step": 843
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981295123813385e-05,
+      "loss": 0.9727,
+      "step": 844
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998118023003168e-05,
+      "loss": 0.9727,
+      "step": 845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998106498479721e-05,
+      "loss": 1.0312,
+      "step": 846
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998094938811403e-05,
+      "loss": 0.957,
+      "step": 847
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998083343998621e-05,
+      "loss": 0.9375,
+      "step": 848
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980717140417835e-05,
+      "loss": 0.9648,
+      "step": 849
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980600489412997e-05,
+      "loss": 0.9492,
+      "step": 850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998048348697581e-05,
+      "loss": 0.9414,
+      "step": 851
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998036613311039e-05,
+      "loss": 1.0391,
+      "step": 852
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980248427820865e-05,
+      "loss": 0.9688,
+      "step": 853
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980130371111386e-05,
+      "loss": 0.9336,
+      "step": 854
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980011962986105e-05,
+      "loss": 0.8828,
+      "step": 855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979893203449195e-05,
+      "loss": 1.0234,
+      "step": 856
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979774092504834e-05,
+      "loss": 1.1641,
+      "step": 857
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997965463015722e-05,
+      "loss": 1.0156,
+      "step": 858
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997953481641056e-05,
+      "loss": 0.9805,
+      "step": 859
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997941465126907e-05,
+      "loss": 1.0312,
+      "step": 860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979294134736983e-05,
+      "loss": 1.0859,
+      "step": 861
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997917326681854e-05,
+      "loss": 0.9844,
+      "step": 862
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979052047517998e-05,
+      "loss": 0.9844,
+      "step": 863
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978930476839625e-05,
+      "loss": 0.9492,
+      "step": 864
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978808554787706e-05,
+      "loss": 0.9414,
+      "step": 865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978686281366528e-05,
+      "loss": 1.0,
+      "step": 866
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978563656580395e-05,
+      "loss": 0.9609,
+      "step": 867
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978440680433636e-05,
+      "loss": 0.9531,
+      "step": 868
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997831735293057e-05,
+      "loss": 0.8945,
+      "step": 869
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978193674075547e-05,
+      "loss": 1.0234,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978069643872914e-05,
+      "loss": 0.9336,
+      "step": 871
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977945262327046e-05,
+      "loss": 0.9688,
+      "step": 872
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977820529442316e-05,
+      "loss": 0.9492,
+      "step": 873
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977695445223123e-05,
+      "loss": 0.9375,
+      "step": 874
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977570009673866e-05,
+      "loss": 0.9805,
+      "step": 875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977444222798962e-05,
+      "loss": 1.0625,
+      "step": 876
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977318084602845e-05,
+      "loss": 0.8711,
+      "step": 877
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997719159508995e-05,
+      "loss": 1.125,
+      "step": 878
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997706475426474e-05,
+      "loss": 1.0,
+      "step": 879
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997693756213167e-05,
+      "loss": 1.1328,
+      "step": 880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976810018695226e-05,
+      "loss": 0.957,
+      "step": 881
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976682123959896e-05,
+      "loss": 0.9766,
+      "step": 882
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976553877930188e-05,
+      "loss": 0.8633,
+      "step": 883
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997642528061061e-05,
+      "loss": 1.0312,
+      "step": 884
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976296332005693e-05,
+      "loss": 0.9922,
+      "step": 885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976167032119984e-05,
+      "loss": 0.918,
+      "step": 886
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976037380958026e-05,
+      "loss": 0.957,
+      "step": 887
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975907378524393e-05,
+      "loss": 0.9766,
+      "step": 888
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975777024823656e-05,
+      "loss": 0.9336,
+      "step": 889
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975646319860408e-05,
+      "loss": 1.0625,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975515263639252e-05,
+      "loss": 0.9297,
+      "step": 891
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975383856164798e-05,
+      "loss": 0.9688,
+      "step": 892
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997525209744168e-05,
+      "loss": 0.8633,
+      "step": 893
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997511998747453e-05,
+      "loss": 0.9258,
+      "step": 894
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997498752626801e-05,
+      "loss": 1.1484,
+      "step": 895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974854713826772e-05,
+      "loss": 1.0859,
+      "step": 896
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99747215501555e-05,
+      "loss": 0.9805,
+      "step": 897
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974588035258882e-05,
+      "loss": 1.0859,
+      "step": 898
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974454169141618e-05,
+      "loss": 1.0312,
+      "step": 899
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974319951808424e-05,
+      "loss": 0.875,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974185383264023e-05,
+      "loss": 0.9961,
+      "step": 901
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997405046351315e-05,
+      "loss": 0.918,
+      "step": 902
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997391519256057e-05,
+      "loss": 0.9688,
+      "step": 903
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973779570411026e-05,
+      "loss": 1.0625,
+      "step": 904
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973643597069312e-05,
+      "loss": 0.9961,
+      "step": 905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973507272540203e-05,
+      "loss": 0.8789,
+      "step": 906
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973370596828504e-05,
+      "loss": 0.9453,
+      "step": 907
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997323356993903e-05,
+      "loss": 0.8438,
+      "step": 908
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99730961918766e-05,
+      "loss": 0.8672,
+      "step": 909
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997295846264606e-05,
+      "loss": 0.8711,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997282038225225e-05,
+      "loss": 0.9062,
+      "step": 911
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997268195070004e-05,
+      "loss": 1.0156,
+      "step": 912
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972543167994297e-05,
+      "loss": 1.0781,
+      "step": 913
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972404034139913e-05,
+      "loss": 0.9141,
+      "step": 914
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972264549141783e-05,
+      "loss": 1.0234,
+      "step": 915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972124713004824e-05,
+      "loss": 0.9727,
+      "step": 916
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971984525733957e-05,
+      "loss": 1.0391,
+      "step": 917
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971843987334118e-05,
+      "loss": 0.9219,
+      "step": 918
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971703097810254e-05,
+      "loss": 0.8672,
+      "step": 919
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971561857167327e-05,
+      "loss": 0.9609,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971420265410315e-05,
+      "loss": 0.9375,
+      "step": 921
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971278322544196e-05,
+      "loss": 1.0391,
+      "step": 922
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971136028573973e-05,
+      "loss": 0.9727,
+      "step": 923
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970993383504656e-05,
+      "loss": 0.9219,
+      "step": 924
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970850387341264e-05,
+      "loss": 1.0078,
+      "step": 925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970707040088837e-05,
+      "loss": 1.0,
+      "step": 926
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997056334175242e-05,
+      "loss": 0.9648,
+      "step": 927
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970419292337074e-05,
+      "loss": 0.9922,
+      "step": 928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970274891847868e-05,
+      "loss": 1.0547,
+      "step": 929
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970130140289892e-05,
+      "loss": 1.0312,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996998503766824e-05,
+      "loss": 0.9453,
+      "step": 931
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996983958398802e-05,
+      "loss": 0.9922,
+      "step": 932
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969693779254354e-05,
+      "loss": 1.0,
+      "step": 933
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969547623472375e-05,
+      "loss": 0.9414,
+      "step": 934
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969401116647237e-05,
+      "loss": 0.8555,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996925425878409e-05,
+      "loss": 1.0391,
+      "step": 936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969107049888107e-05,
+      "loss": 0.9805,
+      "step": 937
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968959489964473e-05,
+      "loss": 0.9609,
+      "step": 938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968811579018385e-05,
+      "loss": 0.9922,
+      "step": 939
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968663317055047e-05,
+      "loss": 0.9219,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968514704079683e-05,
+      "loss": 0.957,
+      "step": 941
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968365740097526e-05,
+      "loss": 0.8477,
+      "step": 942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996821642511382e-05,
+      "loss": 0.9648,
+      "step": 943
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996806675913382e-05,
+      "loss": 1.0625,
+      "step": 944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967916742162804e-05,
+      "loss": 1.0156,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967766374206047e-05,
+      "loss": 1.0078,
+      "step": 946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996761565526884e-05,
+      "loss": 0.9688,
+      "step": 947
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967464585356504e-05,
+      "loss": 0.9727,
+      "step": 948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967313164474347e-05,
+      "loss": 0.9609,
+      "step": 949
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99671613926277e-05,
+      "loss": 0.8867,
+      "step": 950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967009269821917e-05,
+      "loss": 0.9297,
+      "step": 951
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966856796062346e-05,
+      "loss": 1.0469,
+      "step": 952
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996670397135436e-05,
+      "loss": 1.0234,
+      "step": 953
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966550795703336e-05,
+      "loss": 0.957,
+      "step": 954
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966397269114674e-05,
+      "loss": 1.0469,
+      "step": 955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966243391593773e-05,
+      "loss": 1.0078,
+      "step": 956
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966089163146055e-05,
+      "loss": 0.9375,
+      "step": 957
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965934583776948e-05,
+      "loss": 0.9141,
+      "step": 958
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99657796534919e-05,
+      "loss": 1.1016,
+      "step": 959
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996562437229636e-05,
+      "loss": 1.0859,
+      "step": 960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965468740195804e-05,
+      "loss": 1.0156,
+      "step": 961
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965312757195704e-05,
+      "loss": 0.8984,
+      "step": 962
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965156423301554e-05,
+      "loss": 1.1094,
+      "step": 963
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964999738518865e-05,
+      "loss": 0.8906,
+      "step": 964
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964842702853144e-05,
+      "loss": 0.957,
+      "step": 965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964685316309927e-05,
+      "loss": 1.0391,
+      "step": 966
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964527578894756e-05,
+      "loss": 1.0,
+      "step": 967
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964369490613186e-05,
+      "loss": 0.9688,
+      "step": 968
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964211051470778e-05,
+      "loss": 1.0078,
+      "step": 969
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964052261473114e-05,
+      "loss": 0.9883,
+      "step": 970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963893120625787e-05,
+      "loss": 0.9492,
+      "step": 971
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963733628934398e-05,
+      "loss": 1.0781,
+      "step": 972
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996357378640456e-05,
+      "loss": 1.0156,
+      "step": 973
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996341359304191e-05,
+      "loss": 0.9062,
+      "step": 974
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963253048852084e-05,
+      "loss": 0.9531,
+      "step": 975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963092153840736e-05,
+      "loss": 1.0859,
+      "step": 976
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996293090801353e-05,
+      "loss": 0.9922,
+      "step": 977
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996276931137614e-05,
+      "loss": 0.9102,
+      "step": 978
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962607363934262e-05,
+      "loss": 0.9414,
+      "step": 979
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99624450656936e-05,
+      "loss": 0.9375,
+      "step": 980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962282416659858e-05,
+      "loss": 0.9766,
+      "step": 981
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962119416838776e-05,
+      "loss": 0.9648,
+      "step": 982
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996195606623609e-05,
+      "loss": 0.9727,
+      "step": 983
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961792364857545e-05,
+      "loss": 1.0,
+      "step": 984
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996162831270891e-05,
+      "loss": 1.0391,
+      "step": 985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996146390979596e-05,
+      "loss": 1.0,
+      "step": 986
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961299156124488e-05,
+      "loss": 1.0156,
+      "step": 987
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961134051700297e-05,
+      "loss": 0.9688,
+      "step": 988
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960968596529187e-05,
+      "loss": 1.0,
+      "step": 989
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960802790616998e-05,
+      "loss": 0.918,
+      "step": 990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960636633969563e-05,
+      "loss": 1.0859,
+      "step": 991
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960470126592734e-05,
+      "loss": 1.0312,
+      "step": 992
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996030326849237e-05,
+      "loss": 0.9961,
+      "step": 993
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996013605967435e-05,
+      "loss": 0.9609,
+      "step": 994
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995996850014456e-05,
+      "loss": 1.0469,
+      "step": 995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959800589908905e-05,
+      "loss": 0.9609,
+      "step": 996
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959632328973294e-05,
+      "loss": 1.0312,
+      "step": 997
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959463717343644e-05,
+      "loss": 0.8672,
+      "step": 998
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959294755025905e-05,
+      "loss": 1.0547,
+      "step": 999
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995912544202602e-05,
+      "loss": 1.0391,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995895577834995e-05,
+      "loss": 1.0156,
+      "step": 1001
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995878576400367e-05,
+      "loss": 0.9922,
+      "step": 1002
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958615398993166e-05,
+      "loss": 0.9258,
+      "step": 1003
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958444683324442e-05,
+      "loss": 0.9492,
+      "step": 1004
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99582736170035e-05,
+      "loss": 1.0625,
+      "step": 1005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995810220003637e-05,
+      "loss": 0.9766,
+      "step": 1006
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995793043242909e-05,
+      "loss": 0.9453,
+      "step": 1007
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957758314187697e-05,
+      "loss": 0.9375,
+      "step": 1008
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957585845318267e-05,
+      "loss": 1.0391,
+      "step": 1009
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995741302582686e-05,
+      "loss": 0.9688,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957239855719564e-05,
+      "loss": 1.0469,
+      "step": 1011
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957066335002485e-05,
+      "loss": 0.9023,
+      "step": 1012
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956892463681722e-05,
+      "loss": 0.9648,
+      "step": 1013
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956718241763404e-05,
+      "loss": 1.0234,
+      "step": 1014
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956543669253666e-05,
+      "loss": 1.0625,
+      "step": 1015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995636874615865e-05,
+      "loss": 0.8867,
+      "step": 1016
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956193472484517e-05,
+      "loss": 0.9766,
+      "step": 1017
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956017848237438e-05,
+      "loss": 1.0625,
+      "step": 1018
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955841873423602e-05,
+      "loss": 1.1641,
+      "step": 1019
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955665548049203e-05,
+      "loss": 0.918,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955488872120443e-05,
+      "loss": 1.1094,
+      "step": 1021
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995531184564355e-05,
+      "loss": 0.9531,
+      "step": 1022
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995513446862476e-05,
+      "loss": 1.0156,
+      "step": 1023
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954956741070312e-05,
+      "loss": 0.8789,
+      "step": 1024
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954778662986468e-05,
+      "loss": 0.9609,
+      "step": 1025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954600234379495e-05,
+      "loss": 0.8516,
+      "step": 1026
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995442145525568e-05,
+      "loss": 0.8594,
+      "step": 1027
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954242325621313e-05,
+      "loss": 1.0078,
+      "step": 1028
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995406284548271e-05,
+      "loss": 0.9727,
+      "step": 1029
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995388301484618e-05,
+      "loss": 0.8672,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995370283371806e-05,
+      "loss": 1.1328,
+      "step": 1031
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99535223021047e-05,
+      "loss": 0.9102,
+      "step": 1032
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953341420012447e-05,
+      "loss": 0.9453,
+      "step": 1033
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953160187447674e-05,
+      "loss": 0.9453,
+      "step": 1034
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952978604416765e-05,
+      "loss": 0.9414,
+      "step": 1035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952796670926113e-05,
+      "loss": 1.0,
+      "step": 1036
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952614386982124e-05,
+      "loss": 0.9336,
+      "step": 1037
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952431752591213e-05,
+      "loss": 1.0234,
+      "step": 1038
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952248767759815e-05,
+      "loss": 0.8516,
+      "step": 1039
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952065432494368e-05,
+      "loss": 1.0,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951881746801336e-05,
+      "loss": 1.1484,
+      "step": 1041
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995169771068718e-05,
+      "loss": 1.1172,
+      "step": 1042
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951513324158383e-05,
+      "loss": 0.9102,
+      "step": 1043
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951328587221437e-05,
+      "loss": 0.9102,
+      "step": 1044
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951143499882844e-05,
+      "loss": 1.0938,
+      "step": 1045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950958062149126e-05,
+      "loss": 0.9062,
+      "step": 1046
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995077227402681e-05,
+      "loss": 0.9258,
+      "step": 1047
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995058613552244e-05,
+      "loss": 1.2266,
+      "step": 1048
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950399646642565e-05,
+      "loss": 1.0469,
+      "step": 1049
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950212807393757e-05,
+      "loss": 0.8867,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995002561778259e-05,
+      "loss": 1.1953,
+      "step": 1051
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949838077815664e-05,
+      "loss": 0.9883,
+      "step": 1052
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949650187499575e-05,
+      "loss": 0.918,
+      "step": 1053
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994946194684094e-05,
+      "loss": 0.8906,
+      "step": 1054
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949273355846386e-05,
+      "loss": 0.9062,
+      "step": 1055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949084414522554e-05,
+      "loss": 0.8789,
+      "step": 1056
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948895122876103e-05,
+      "loss": 1.1094,
+      "step": 1057
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948705480913694e-05,
+      "loss": 0.9102,
+      "step": 1058
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948515488642003e-05,
+      "loss": 1.0,
+      "step": 1059
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948325146067717e-05,
+      "loss": 1.0078,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948134453197548e-05,
+      "loss": 0.9414,
+      "step": 1061
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99479434100382e-05,
+      "loss": 0.9883,
+      "step": 1062
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994775201659641e-05,
+      "loss": 0.9922,
+      "step": 1063
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994756027287891e-05,
+      "loss": 1.0703,
+      "step": 1064
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947368178892454e-05,
+      "loss": 0.8828,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947175734643804e-05,
+      "loss": 1.0703,
+      "step": 1066
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946982940139742e-05,
+      "loss": 0.9805,
+      "step": 1067
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946789795387048e-05,
+      "loss": 0.9648,
+      "step": 1068
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994659630039253e-05,
+      "loss": 0.9883,
+      "step": 1069
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946402455162998e-05,
+      "loss": 0.9609,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994620825970528e-05,
+      "loss": 1.0234,
+      "step": 1071
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994601371402621e-05,
+      "loss": 0.8789,
+      "step": 1072
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994581881813264e-05,
+      "loss": 0.9023,
+      "step": 1073
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945623572031436e-05,
+      "loss": 0.9922,
+      "step": 1074
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994542797572947e-05,
+      "loss": 0.9531,
+      "step": 1075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994523202923363e-05,
+      "loss": 1.0312,
+      "step": 1076
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994503573255081e-05,
+      "loss": 1.0703,
+      "step": 1077
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994483908568793e-05,
+      "loss": 0.9258,
+      "step": 1078
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994464208865191e-05,
+      "loss": 0.9844,
+      "step": 1079
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994444474144969e-05,
+      "loss": 1.0547,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994424704408821e-05,
+      "loss": 0.9688,
+      "step": 1081
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944048996574444e-05,
+      "loss": 1.0781,
+      "step": 1082
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943850598915357e-05,
+      "loss": 0.9414,
+      "step": 1083
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994365185111794e-05,
+      "loss": 1.0859,
+      "step": 1084
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943452753189186e-05,
+      "loss": 0.8789,
+      "step": 1085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994325330513611e-05,
+      "loss": 1.0391,
+      "step": 1086
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943053506965726e-05,
+      "loss": 0.9492,
+      "step": 1087
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994285335868508e-05,
+      "loss": 0.9531,
+      "step": 1088
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994265286030122e-05,
+      "loss": 1.0156,
+      "step": 1089
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942452011821195e-05,
+      "loss": 0.9102,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942250813252088e-05,
+      "loss": 1.0547,
+      "step": 1091
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942049264600974e-05,
+      "loss": 0.9961,
+      "step": 1092
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994184736587496e-05,
+      "loss": 0.9062,
+      "step": 1093
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941645117081148e-05,
+      "loss": 1.1562,
+      "step": 1094
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941442518226663e-05,
+      "loss": 1.0078,
+      "step": 1095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941239569318635e-05,
+      "loss": 0.8477,
+      "step": 1096
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941036270364212e-05,
+      "loss": 1.0,
+      "step": 1097
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940832621370558e-05,
+      "loss": 0.875,
+      "step": 1098
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940628622344834e-05,
+      "loss": 0.9688,
+      "step": 1099
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994042427329423e-05,
+      "loss": 1.0078,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940219574225937e-05,
+      "loss": 1.0312,
+      "step": 1101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994001452514717e-05,
+      "loss": 0.9531,
+      "step": 1102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993980912606514e-05,
+      "loss": 1.0156,
+      "step": 1103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939603376987083e-05,
+      "loss": 1.0078,
+      "step": 1104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939397277920247e-05,
+      "loss": 0.9062,
+      "step": 1105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939190828871886e-05,
+      "loss": 0.9023,
+      "step": 1106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938984029849268e-05,
+      "loss": 0.9688,
+      "step": 1107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993877688085968e-05,
+      "loss": 0.8672,
+      "step": 1108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993856938191041e-05,
+      "loss": 1.0312,
+      "step": 1109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993836153300877e-05,
+      "loss": 0.8633,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993815333416207e-05,
+      "loss": 0.9219,
+      "step": 1111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993794478537765e-05,
+      "loss": 0.9219,
+      "step": 1112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993773588666285e-05,
+      "loss": 0.8945,
+      "step": 1113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937526638025028e-05,
+      "loss": 1.1875,
+      "step": 1114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937317039471545e-05,
+      "loss": 1.0,
+      "step": 1115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937107091009785e-05,
+      "loss": 0.8984,
+      "step": 1116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936896792647148e-05,
+      "loss": 1.0312,
+      "step": 1117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993668614439103e-05,
+      "loss": 1.0938,
+      "step": 1118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936475146248847e-05,
+      "loss": 0.9336,
+      "step": 1119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936263798228033e-05,
+      "loss": 1.0,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936052100336032e-05,
+      "loss": 1.0547,
+      "step": 1121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935840052580294e-05,
+      "loss": 0.9375,
+      "step": 1122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935627654968287e-05,
+      "loss": 1.0703,
+      "step": 1123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993541490750749e-05,
+      "loss": 0.9609,
+      "step": 1124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935201810205393e-05,
+      "loss": 0.9414,
+      "step": 1125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99349883630695e-05,
+      "loss": 0.9375,
+      "step": 1126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934774566107328e-05,
+      "loss": 1.0547,
+      "step": 1127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934560419326403e-05,
+      "loss": 1.0547,
+      "step": 1128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993434592273427e-05,
+      "loss": 1.0078,
+      "step": 1129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934131076338476e-05,
+      "loss": 1.0312,
+      "step": 1130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993391588014659e-05,
+      "loss": 0.9141,
+      "step": 1131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933700334166185e-05,
+      "loss": 0.9844,
+      "step": 1132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933484438404854e-05,
+      "loss": 0.9062,
+      "step": 1133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933268192870202e-05,
+      "loss": 0.9844,
+      "step": 1134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993305159756984e-05,
+      "loss": 0.875,
+      "step": 1135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993283465251139e-05,
+      "loss": 1.0078,
+      "step": 1136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99326173577025e-05,
+      "loss": 1.0234,
+      "step": 1137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932399713150814e-05,
+      "loss": 0.9297,
+      "step": 1138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932181718864e-05,
+      "loss": 1.0,
+      "step": 1139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931963374849732e-05,
+      "loss": 1.0234,
+      "step": 1140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931744681115698e-05,
+      "loss": 0.8555,
+      "step": 1141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931525637669603e-05,
+      "loss": 1.0,
+      "step": 1142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993130624451915e-05,
+      "loss": 1.0469,
+      "step": 1143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931086501672076e-05,
+      "loss": 1.0156,
+      "step": 1144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993086640913611e-05,
+      "loss": 0.9766,
+      "step": 1145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930645966919004e-05,
+      "loss": 1.0547,
+      "step": 1146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930425175028523e-05,
+      "loss": 0.9922,
+      "step": 1147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930204033472437e-05,
+      "loss": 0.9922,
+      "step": 1148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929982542258534e-05,
+      "loss": 1.0156,
+      "step": 1149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929760701394615e-05,
+      "loss": 0.8633,
+      "step": 1150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992953851088849e-05,
+      "loss": 0.9023,
+      "step": 1151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992931597074798e-05,
+      "loss": 0.9297,
+      "step": 1152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992909308098093e-05,
+      "loss": 1.0,
+      "step": 1153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992886984159518e-05,
+      "loss": 0.9492,
+      "step": 1154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992864625259859e-05,
+      "loss": 0.9688,
+      "step": 1155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928422313999038e-05,
+      "loss": 0.9531,
+      "step": 1156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992819802580441e-05,
+      "loss": 1.0,
+      "step": 1157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927973388022594e-05,
+      "loss": 1.0547,
+      "step": 1158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992774840066151e-05,
+      "loss": 1.125,
+      "step": 1159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927523063729077e-05,
+      "loss": 1.0391,
+      "step": 1160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927297377233224e-05,
+      "loss": 0.9414,
+      "step": 1161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927071341181908e-05,
+      "loss": 1.0234,
+      "step": 1162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992684495558308e-05,
+      "loss": 0.9414,
+      "step": 1163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926618220444715e-05,
+      "loss": 0.9844,
+      "step": 1164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926391135774795e-05,
+      "loss": 1.0547,
+      "step": 1165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926163701581318e-05,
+      "loss": 1.0391,
+      "step": 1166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992593591787229e-05,
+      "loss": 0.8828,
+      "step": 1167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925707784655734e-05,
+      "loss": 0.9805,
+      "step": 1168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992547930193968e-05,
+      "loss": 0.8672,
+      "step": 1169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925250469732177e-05,
+      "loss": 0.9844,
+      "step": 1170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925021288041276e-05,
+      "loss": 0.8828,
+      "step": 1171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924791756875053e-05,
+      "loss": 1.1562,
+      "step": 1172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924561876241593e-05,
+      "loss": 1.0156,
+      "step": 1173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992433164614898e-05,
+      "loss": 0.9844,
+      "step": 1174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924101066605332e-05,
+      "loss": 1.0859,
+      "step": 1175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923870137618756e-05,
+      "loss": 0.9766,
+      "step": 1176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923638859197395e-05,
+      "loss": 0.9062,
+      "step": 1177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923407231349386e-05,
+      "loss": 0.9219,
+      "step": 1178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992317525408289e-05,
+      "loss": 0.9844,
+      "step": 1179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922942927406067e-05,
+      "loss": 1.0234,
+      "step": 1180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922710251327105e-05,
+      "loss": 1.0938,
+      "step": 1181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922477225854195e-05,
+      "loss": 0.9219,
+      "step": 1182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992224385099554e-05,
+      "loss": 1.0156,
+      "step": 1183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922010126759362e-05,
+      "loss": 0.9102,
+      "step": 1184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921776053153886e-05,
+      "loss": 0.9531,
+      "step": 1185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921541630187354e-05,
+      "loss": 0.9727,
+      "step": 1186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921306857868025e-05,
+      "loss": 1.0312,
+      "step": 1187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921071736204163e-05,
+      "loss": 0.9844,
+      "step": 1188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920836265204047e-05,
+      "loss": 0.918,
+      "step": 1189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920600444875968e-05,
+      "loss": 1.0078,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920364275228226e-05,
+      "loss": 1.0078,
+      "step": 1191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992012775626915e-05,
+      "loss": 0.9062,
+      "step": 1192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991989088800705e-05,
+      "loss": 0.9609,
+      "step": 1193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919653670450278e-05,
+      "loss": 1.0156,
+      "step": 1194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919416103607188e-05,
+      "loss": 1.0156,
+      "step": 1195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991917818748614e-05,
+      "loss": 1.0234,
+      "step": 1196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991893992209551e-05,
+      "loss": 1.0234,
+      "step": 1197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918701307443692e-05,
+      "loss": 0.9531,
+      "step": 1198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918462343539086e-05,
+      "loss": 1.0234,
+      "step": 1199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991822303039011e-05,
+      "loss": 1.0781,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917983368005185e-05,
+      "loss": 0.9297,
+      "step": 1201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991774335639275e-05,
+      "loss": 0.8945,
+      "step": 1202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917502995561265e-05,
+      "loss": 0.8594,
+      "step": 1203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991726228551918e-05,
+      "loss": 1.1641,
+      "step": 1204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917021226274982e-05,
+      "loss": 1.0,
+      "step": 1205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916779817837153e-05,
+      "loss": 1.1094,
+      "step": 1206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916538060214195e-05,
+      "loss": 1.0547,
+      "step": 1207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991629595341462e-05,
+      "loss": 0.9805,
+      "step": 1208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916053497446958e-05,
+      "loss": 0.8906,
+      "step": 1209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915810692319737e-05,
+      "loss": 1.0625,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915567538041514e-05,
+      "loss": 0.9922,
+      "step": 1211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991532403462085e-05,
+      "loss": 1.1406,
+      "step": 1212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915080182066314e-05,
+      "loss": 1.0469,
+      "step": 1213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914835980386498e-05,
+      "loss": 0.9648,
+      "step": 1214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914591429589997e-05,
+      "loss": 0.8477,
+      "step": 1215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914346529685426e-05,
+      "loss": 1.0078,
+      "step": 1216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914101280681405e-05,
+      "loss": 0.918,
+      "step": 1217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991385568258657e-05,
+      "loss": 1.0234,
+      "step": 1218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991360973540957e-05,
+      "loss": 0.8672,
+      "step": 1219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913363439159065e-05,
+      "loss": 0.8867,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913116793843725e-05,
+      "loss": 1.0,
+      "step": 1221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991286979947224e-05,
+      "loss": 0.9922,
+      "step": 1222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99126224560533e-05,
+      "loss": 0.9844,
+      "step": 1223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912374763595623e-05,
+      "loss": 0.9141,
+      "step": 1224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991212672210792e-05,
+      "loss": 0.9375,
+      "step": 1225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911878331598936e-05,
+      "loss": 0.9453,
+      "step": 1226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991162959207741e-05,
+      "loss": 0.9258,
+      "step": 1227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99113805035521e-05,
+      "loss": 0.9609,
+      "step": 1228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991113106603179e-05,
+      "loss": 1.0469,
+      "step": 1229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910881279525244e-05,
+      "loss": 0.9297,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910631144041263e-05,
+      "loss": 1.0234,
+      "step": 1231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910380659588663e-05,
+      "loss": 0.9844,
+      "step": 1232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910129826176255e-05,
+      "loss": 0.9453,
+      "step": 1233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909878643812877e-05,
+      "loss": 0.9297,
+      "step": 1234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990962711250737e-05,
+      "loss": 0.9609,
+      "step": 1235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909375232268592e-05,
+      "loss": 1.0234,
+      "step": 1236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990912300310541e-05,
+      "loss": 0.9648,
+      "step": 1237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908870425026713e-05,
+      "loss": 0.9219,
+      "step": 1238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908617498041384e-05,
+      "loss": 1.0312,
+      "step": 1239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908364222158333e-05,
+      "loss": 0.9219,
+      "step": 1240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908110597386484e-05,
+      "loss": 1.0625,
+      "step": 1241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990785662373476e-05,
+      "loss": 0.9531,
+      "step": 1242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907602301212107e-05,
+      "loss": 1.0312,
+      "step": 1243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990734762982748e-05,
+      "loss": 1.0391,
+      "step": 1244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907092609589845e-05,
+      "loss": 0.9141,
+      "step": 1245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906837240508187e-05,
+      "loss": 0.9609,
+      "step": 1246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990658152259149e-05,
+      "loss": 1.0547,
+      "step": 1247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990632545584876e-05,
+      "loss": 0.8672,
+      "step": 1248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906069040289018e-05,
+      "loss": 0.9805,
+      "step": 1249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990581227592129e-05,
+      "loss": 0.9102,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905555162754618e-05,
+      "loss": 0.9141,
+      "step": 1251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990529770079805e-05,
+      "loss": 0.9258,
+      "step": 1252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905039890060662e-05,
+      "loss": 0.9453,
+      "step": 1253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904781730551522e-05,
+      "loss": 0.9531,
+      "step": 1254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990452322227973e-05,
+      "loss": 1.0547,
+      "step": 1255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904264365254376e-05,
+      "loss": 0.9805,
+      "step": 1256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904005159484586e-05,
+      "loss": 0.8711,
+      "step": 1257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990374560497948e-05,
+      "loss": 1.0312,
+      "step": 1258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903485701748202e-05,
+      "loss": 0.8984,
+      "step": 1259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99032254497999e-05,
+      "loss": 0.9766,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990296484914374e-05,
+      "loss": 0.9688,
+      "step": 1261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902703899788897e-05,
+      "loss": 0.9141,
+      "step": 1262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990244260174456e-05,
+      "loss": 0.918,
+      "step": 1263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902180955019933e-05,
+      "loss": 1.0312,
+      "step": 1264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990191895962422e-05,
+      "loss": 0.9062,
+      "step": 1265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901656615566655e-05,
+      "loss": 1.0156,
+      "step": 1266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901393922856473e-05,
+      "loss": 1.0078,
+      "step": 1267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901130881502923e-05,
+      "loss": 0.9961,
+      "step": 1268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900867491515266e-05,
+      "loss": 0.9219,
+      "step": 1269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900603752902778e-05,
+      "loss": 0.8945,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900339665674747e-05,
+      "loss": 0.7812,
+      "step": 1271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900075229840467e-05,
+      "loss": 0.8672,
+      "step": 1272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899810445409256e-05,
+      "loss": 0.9688,
+      "step": 1273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989954531239043e-05,
+      "loss": 0.9844,
+      "step": 1274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899279830793334e-05,
+      "loss": 0.9961,
+      "step": 1275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899014000627308e-05,
+      "loss": 1.0156,
+      "step": 1276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898747821901715e-05,
+      "loss": 0.9219,
+      "step": 1277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898481294625927e-05,
+      "loss": 0.9805,
+      "step": 1278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.9805,
+      "step": 1279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989794719446132e-05,
+      "loss": 0.9531,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897679621591307e-05,
+      "loss": 1.0,
+      "step": 1281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897411700208714e-05,
+      "loss": 0.8984,
+      "step": 1282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989714343032297e-05,
+      "loss": 1.0078,
+      "step": 1283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896874811943528e-05,
+      "loss": 1.0156,
+      "step": 1284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989660584507984e-05,
+      "loss": 0.9609,
+      "step": 1285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896336529741384e-05,
+      "loss": 0.8633,
+      "step": 1286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896066865937636e-05,
+      "loss": 0.9531,
+      "step": 1287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895796853678096e-05,
+      "loss": 1.0078,
+      "step": 1288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895526492972268e-05,
+      "loss": 0.9883,
+      "step": 1289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895255783829676e-05,
+      "loss": 0.8867,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894984726259843e-05,
+      "loss": 1.1016,
+      "step": 1291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989471332027233e-05,
+      "loss": 1.0938,
+      "step": 1292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894441565876677e-05,
+      "loss": 0.918,
+      "step": 1293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989416946308246e-05,
+      "loss": 1.0156,
+      "step": 1294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893897011899262e-05,
+      "loss": 0.918,
+      "step": 1295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989362421233667e-05,
+      "loss": 0.9258,
+      "step": 1296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893351064404297e-05,
+      "loss": 0.918,
+      "step": 1297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989307756811176e-05,
+      "loss": 1.125,
+      "step": 1298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989280372346868e-05,
+      "loss": 0.9844,
+      "step": 1299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892529530484713e-05,
+      "loss": 0.8711,
+      "step": 1300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892254989169504e-05,
+      "loss": 0.9062,
+      "step": 1301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891980099532725e-05,
+      "loss": 0.9766,
+      "step": 1302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989170486158405e-05,
+      "loss": 0.9375,
+      "step": 1303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989142927533318e-05,
+      "loss": 1.0469,
+      "step": 1304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891153340789807e-05,
+      "loss": 0.9336,
+      "step": 1305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890877057963657e-05,
+      "loss": 0.9375,
+      "step": 1306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989060042686445e-05,
+      "loss": 0.9219,
+      "step": 1307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890323447501937e-05,
+      "loss": 0.8984,
+      "step": 1308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989004611988586e-05,
+      "loss": 1.0312,
+      "step": 1309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988976844402599e-05,
+      "loss": 1.1016,
+      "step": 1310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889490419932104e-05,
+      "loss": 0.9688,
+      "step": 1311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988921204761399e-05,
+      "loss": 1.0547,
+      "step": 1312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888933327081453e-05,
+      "loss": 0.8242,
+      "step": 1313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888654258344303e-05,
+      "loss": 0.9609,
+      "step": 1314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988837484141237e-05,
+      "loss": 1.0078,
+      "step": 1315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988809507629549e-05,
+      "loss": 0.9766,
+      "step": 1316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887814963003516e-05,
+      "loss": 0.9375,
+      "step": 1317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988753450154631e-05,
+      "loss": 0.9922,
+      "step": 1318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887253691933747e-05,
+      "loss": 1.0703,
+      "step": 1319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886972534175715e-05,
+      "loss": 0.9141,
+      "step": 1320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886691028282118e-05,
+      "loss": 0.957,
+      "step": 1321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988640917426286e-05,
+      "loss": 0.9453,
+      "step": 1322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886126972127876e-05,
+      "loss": 0.957,
+      "step": 1323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885844421887093e-05,
+      "loss": 1.0547,
+      "step": 1324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885561523550466e-05,
+      "loss": 0.9883,
+      "step": 1325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885278277127954e-05,
+      "loss": 0.9297,
+      "step": 1326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988499468262953e-05,
+      "loss": 0.9961,
+      "step": 1327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884710740065185e-05,
+      "loss": 0.918,
+      "step": 1328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884426449444908e-05,
+      "loss": 1.0781,
+      "step": 1329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884141810778718e-05,
+      "loss": 0.9102,
+      "step": 1330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988385682407663e-05,
+      "loss": 0.9688,
+      "step": 1331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988357148934869e-05,
+      "loss": 0.9102,
+      "step": 1332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883285806604927e-05,
+      "loss": 0.9141,
+      "step": 1333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882999775855417e-05,
+      "loss": 0.9297,
+      "step": 1334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882713397110226e-05,
+      "loss": 0.9531,
+      "step": 1335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882426670379434e-05,
+      "loss": 0.9297,
+      "step": 1336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882139595673146e-05,
+      "loss": 0.8477,
+      "step": 1337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988185217300146e-05,
+      "loss": 0.918,
+      "step": 1338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881564402374502e-05,
+      "loss": 0.9336,
+      "step": 1339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881276283802407e-05,
+      "loss": 0.9492,
+      "step": 1340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880987817295314e-05,
+      "loss": 1.0078,
+      "step": 1341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880699002863386e-05,
+      "loss": 0.9961,
+      "step": 1342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988040984051679e-05,
+      "loss": 0.8945,
+      "step": 1343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988012033026571e-05,
+      "loss": 0.9375,
+      "step": 1344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879830472120333e-05,
+      "loss": 0.9102,
+      "step": 1345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879540266090875e-05,
+      "loss": 0.8086,
+      "step": 1346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987924971218755e-05,
+      "loss": 1.0781,
+      "step": 1347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987895881042059e-05,
+      "loss": 0.9258,
+      "step": 1348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878667560800236e-05,
+      "loss": 0.8203,
+      "step": 1349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987837596333674e-05,
+      "loss": 1.0312,
+      "step": 1350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987808401804038e-05,
+      "loss": 1.0,
+      "step": 1351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987779172492143e-05,
+      "loss": 0.9375,
+      "step": 1352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987749908399018e-05,
+      "loss": 0.9805,
+      "step": 1353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877206095256935e-05,
+      "loss": 0.9531,
+      "step": 1354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876912758732018e-05,
+      "loss": 1.0938,
+      "step": 1355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876619074425747e-05,
+      "loss": 1.0234,
+      "step": 1356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876325042348472e-05,
+      "loss": 0.9023,
+      "step": 1357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876030662510543e-05,
+      "loss": 1.0391,
+      "step": 1358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875735934922325e-05,
+      "loss": 1.0938,
+      "step": 1359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875440859594197e-05,
+      "loss": 0.9727,
+      "step": 1360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875145436536548e-05,
+      "loss": 0.8867,
+      "step": 1361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874849665759784e-05,
+      "loss": 0.9219,
+      "step": 1362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874553547274312e-05,
+      "loss": 0.957,
+      "step": 1363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987425708109057e-05,
+      "loss": 0.9961,
+      "step": 1364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873960267218987e-05,
+      "loss": 0.9258,
+      "step": 1365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873663105670014e-05,
+      "loss": 0.9375,
+      "step": 1366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873365596454123e-05,
+      "loss": 1.1484,
+      "step": 1367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987306773958178e-05,
+      "loss": 1.1484,
+      "step": 1368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987276953506348e-05,
+      "loss": 0.8555,
+      "step": 1369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872470982909725e-05,
+      "loss": 0.8828,
+      "step": 1370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872172083131018e-05,
+      "loss": 0.9062,
+      "step": 1371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871872835737894e-05,
+      "loss": 0.9883,
+      "step": 1372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871573240740886e-05,
+      "loss": 1.1172,
+      "step": 1373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987127329815054e-05,
+      "loss": 0.9336,
+      "step": 1374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870973007977417e-05,
+      "loss": 1.0156,
+      "step": 1375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870672370232097e-05,
+      "loss": 1.0859,
+      "step": 1376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987037138492516e-05,
+      "loss": 0.8906,
+      "step": 1377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987007005206721e-05,
+      "loss": 0.9492,
+      "step": 1378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986976837166885e-05,
+      "loss": 0.9062,
+      "step": 1379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869466343740712e-05,
+      "loss": 0.918,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869163968293423e-05,
+      "loss": 0.9141,
+      "step": 1381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868861245337634e-05,
+      "loss": 0.9141,
+      "step": 1382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868558174884e-05,
+      "loss": 0.8906,
+      "step": 1383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98682547569432e-05,
+      "loss": 0.8906,
+      "step": 1384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986795099152591e-05,
+      "loss": 0.9023,
+      "step": 1385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986764687864283e-05,
+      "loss": 0.9922,
+      "step": 1386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986734241830467e-05,
+      "loss": 0.9648,
+      "step": 1387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867037610522147e-05,
+      "loss": 1.0234,
+      "step": 1388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986673245530599e-05,
+      "loss": 0.9453,
+      "step": 1389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866426952666956e-05,
+      "loss": 1.1172,
+      "step": 1390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866121102615795e-05,
+      "loss": 0.9648,
+      "step": 1391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865814905163274e-05,
+      "loss": 0.9375,
+      "step": 1392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986550836032018e-05,
+      "loss": 0.8906,
+      "step": 1393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865201468097302e-05,
+      "loss": 1.0781,
+      "step": 1394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986489422850545e-05,
+      "loss": 0.9609,
+      "step": 1395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864586641555437e-05,
+      "loss": 0.9102,
+      "step": 1396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864278707258103e-05,
+      "loss": 0.8672,
+      "step": 1397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986397042562428e-05,
+      "loss": 0.9688,
+      "step": 1398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863661796664835e-05,
+      "loss": 0.9961,
+      "step": 1399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863352820390626e-05,
+      "loss": 0.8867,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863043496812532e-05,
+      "loss": 1.0078,
+      "step": 1401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986273382594145e-05,
+      "loss": 0.9023,
+      "step": 1402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986242380778828e-05,
+      "loss": 0.9453,
+      "step": 1403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862113442363942e-05,
+      "loss": 0.9688,
+      "step": 1404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861802729679364e-05,
+      "loss": 1.0312,
+      "step": 1405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861491669745484e-05,
+      "loss": 0.9219,
+      "step": 1406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861180262573255e-05,
+      "loss": 1.0234,
+      "step": 1407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860868508173646e-05,
+      "loss": 1.0,
+      "step": 1408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986055640655763e-05,
+      "loss": 0.9062,
+      "step": 1409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860243957736197e-05,
+      "loss": 0.9922,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859931161720352e-05,
+      "loss": 0.9609,
+      "step": 1411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985961801852111e-05,
+      "loss": 1.0547,
+      "step": 1412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985930452814949e-05,
+      "loss": 1.0625,
+      "step": 1413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858990690616536e-05,
+      "loss": 0.9648,
+      "step": 1414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858676505933296e-05,
+      "loss": 0.9766,
+      "step": 1415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858361974110837e-05,
+      "loss": 0.9844,
+      "step": 1416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985804709516023e-05,
+      "loss": 0.9961,
+      "step": 1417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857731869092565e-05,
+      "loss": 0.9453,
+      "step": 1418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985741629591894e-05,
+      "loss": 0.9922,
+      "step": 1419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985710037565047e-05,
+      "loss": 1.0,
+      "step": 1420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856784108298274e-05,
+      "loss": 0.9297,
+      "step": 1421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985646749387349e-05,
+      "loss": 0.9258,
+      "step": 1422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856150532387266e-05,
+      "loss": 0.8672,
+      "step": 1423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855833223850768e-05,
+      "loss": 0.8945,
+      "step": 1424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855515568275157e-05,
+      "loss": 0.9727,
+      "step": 1425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855197565671632e-05,
+      "loss": 1.0469,
+      "step": 1426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854879216051387e-05,
+      "loss": 0.9688,
+      "step": 1427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854560519425626e-05,
+      "loss": 0.9375,
+      "step": 1428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854241475805575e-05,
+      "loss": 0.8867,
+      "step": 1429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853922085202465e-05,
+      "loss": 0.9297,
+      "step": 1430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853602347627544e-05,
+      "loss": 0.9688,
+      "step": 1431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985328226309207e-05,
+      "loss": 1.0156,
+      "step": 1432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852961831607313e-05,
+      "loss": 0.9844,
+      "step": 1433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852641053184558e-05,
+      "loss": 0.9219,
+      "step": 1434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98523199278351e-05,
+      "loss": 0.9141,
+      "step": 1435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851998455570247e-05,
+      "loss": 0.9531,
+      "step": 1436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851676636401314e-05,
+      "loss": 1.1797,
+      "step": 1437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985135447033964e-05,
+      "loss": 0.9062,
+      "step": 1438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851031957396558e-05,
+      "loss": 1.1016,
+      "step": 1439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850709097583434e-05,
+      "loss": 1.0234,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850385890911634e-05,
+      "loss": 0.9375,
+      "step": 1441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850062337392536e-05,
+      "loss": 0.8984,
+      "step": 1442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849738437037537e-05,
+      "loss": 0.9805,
+      "step": 1443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984941418985804e-05,
+      "loss": 1.0156,
+      "step": 1444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849089595865462e-05,
+      "loss": 0.8594,
+      "step": 1445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984876465507123e-05,
+      "loss": 0.8594,
+      "step": 1446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984843936748679e-05,
+      "loss": 0.9023,
+      "step": 1447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984811373312359e-05,
+      "loss": 1.0234,
+      "step": 1448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847787751993104e-05,
+      "loss": 1.0312,
+      "step": 1449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847461424106807e-05,
+      "loss": 0.9336,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847134749476183e-05,
+      "loss": 0.9805,
+      "step": 1451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846807728112747e-05,
+      "loss": 0.9141,
+      "step": 1452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846480360028006e-05,
+      "loss": 0.9648,
+      "step": 1453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846152645233485e-05,
+      "loss": 1.0078,
+      "step": 1454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845824583740733e-05,
+      "loss": 0.9922,
+      "step": 1455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984549617556129e-05,
+      "loss": 1.2031,
+      "step": 1456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845167420706727e-05,
+      "loss": 0.9609,
+      "step": 1457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984483831918862e-05,
+      "loss": 0.8516,
+      "step": 1458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844508871018553e-05,
+      "loss": 0.9844,
+      "step": 1459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844179076208128e-05,
+      "loss": 0.9102,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984384893476896e-05,
+      "loss": 1.0312,
+      "step": 1461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843518446712674e-05,
+      "loss": 0.8867,
+      "step": 1462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98431876120509e-05,
+      "loss": 1.0156,
+      "step": 1463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842856430795298e-05,
+      "loss": 1.0312,
+      "step": 1464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842524902957523e-05,
+      "loss": 1.1953,
+      "step": 1465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842193028549248e-05,
+      "loss": 0.8438,
+      "step": 1466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984186080758216e-05,
+      "loss": 0.9727,
+      "step": 1467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841528240067955e-05,
+      "loss": 1.0156,
+      "step": 1468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984119532601835e-05,
+      "loss": 0.9102,
+      "step": 1469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840862065445057e-05,
+      "loss": 0.8633,
+      "step": 1470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984052845835982e-05,
+      "loss": 0.9805,
+      "step": 1471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984019450477438e-05,
+      "loss": 0.9766,
+      "step": 1472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839860204700504e-05,
+      "loss": 0.9961,
+      "step": 1473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839525558149952e-05,
+      "loss": 1.0078,
+      "step": 1474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839190565134515e-05,
+      "loss": 1.1953,
+      "step": 1475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838855225665983e-05,
+      "loss": 0.9492,
+      "step": 1476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838519539756172e-05,
+      "loss": 0.9727,
+      "step": 1477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98381835074169e-05,
+      "loss": 1.0625,
+      "step": 1478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837847128659992e-05,
+      "loss": 1.1016,
+      "step": 1479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837510403497296e-05,
+      "loss": 1.0469,
+      "step": 1480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837173331940672e-05,
+      "loss": 0.9883,
+      "step": 1481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836835914001983e-05,
+      "loss": 0.9648,
+      "step": 1482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836498149693117e-05,
+      "loss": 0.9062,
+      "step": 1483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983616003902596e-05,
+      "loss": 1.0312,
+      "step": 1484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835821582012426e-05,
+      "loss": 0.957,
+      "step": 1485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835482778664426e-05,
+      "loss": 0.957,
+      "step": 1486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983514362899389e-05,
+      "loss": 0.9766,
+      "step": 1487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983480413301276e-05,
+      "loss": 0.9297,
+      "step": 1488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834464290732996e-05,
+      "loss": 1.0312,
+      "step": 1489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983412410216656e-05,
+      "loss": 0.957,
+      "step": 1490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833783567325426e-05,
+      "loss": 0.9609,
+      "step": 1491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983344268622159e-05,
+      "loss": 0.8555,
+      "step": 1492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833101458867056e-05,
+      "loss": 1.0312,
+      "step": 1493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832759885273834e-05,
+      "loss": 1.0625,
+      "step": 1494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983241796545396e-05,
+      "loss": 0.9375,
+      "step": 1495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832075699419466e-05,
+      "loss": 1.0078,
+      "step": 1496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831733087182406e-05,
+      "loss": 0.9922,
+      "step": 1497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831390128754843e-05,
+      "loss": 0.9727,
+      "step": 1498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831046824148858e-05,
+      "loss": 1.1562,
+      "step": 1499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983070317337653e-05,
+      "loss": 0.9375,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830359176449967e-05,
+      "loss": 0.8867,
+      "step": 1501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830014833381276e-05,
+      "loss": 0.9141,
+      "step": 1502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982967014418259e-05,
+      "loss": 0.875,
+      "step": 1503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829325108866036e-05,
+      "loss": 0.7852,
+      "step": 1504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828979727443776e-05,
+      "loss": 0.9414,
+      "step": 1505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982863399992796e-05,
+      "loss": 0.9766,
+      "step": 1506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982828792633076e-05,
+      "loss": 1.0078,
+      "step": 1507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827941506664378e-05,
+      "loss": 0.957,
+      "step": 1508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827594740940993e-05,
+      "loss": 1.0469,
+      "step": 1509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827247629172825e-05,
+      "loss": 0.9453,
+      "step": 1510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98269001713721e-05,
+      "loss": 1.0234,
+      "step": 1511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982655236755104e-05,
+      "loss": 0.9766,
+      "step": 1512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826204217721902e-05,
+      "loss": 0.918,
+      "step": 1513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825855721896944e-05,
+      "loss": 0.8672,
+      "step": 1514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825506880088433e-05,
+      "loss": 0.9844,
+      "step": 1515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982515769230865e-05,
+      "loss": 1.0781,
+      "step": 1516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824808158569903e-05,
+      "loss": 1.0312,
+      "step": 1517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824458278884485e-05,
+      "loss": 0.9219,
+      "step": 1518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824108053264726e-05,
+      "loss": 0.8828,
+      "step": 1519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823757481722952e-05,
+      "loss": 0.9688,
+      "step": 1520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823406564271507e-05,
+      "loss": 0.8828,
+      "step": 1521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823055300922753e-05,
+      "loss": 0.9688,
+      "step": 1522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822703691689057e-05,
+      "loss": 1.0703,
+      "step": 1523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822351736582793e-05,
+      "loss": 0.9648,
+      "step": 1524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982199943561636e-05,
+      "loss": 0.9531,
+      "step": 1525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821646788802162e-05,
+      "loss": 1.0625,
+      "step": 1526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821293796152617e-05,
+      "loss": 1.0781,
+      "step": 1527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820940457680153e-05,
+      "loss": 0.9062,
+      "step": 1528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982058677339721e-05,
+      "loss": 1.0859,
+      "step": 1529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820232743316244e-05,
+      "loss": 0.9102,
+      "step": 1530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819878367449723e-05,
+      "loss": 1.1016,
+      "step": 1531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819523645810125e-05,
+      "loss": 0.8711,
+      "step": 1532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819168578409937e-05,
+      "loss": 0.9648,
+      "step": 1533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981881316526166e-05,
+      "loss": 0.9844,
+      "step": 1534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818457406377808e-05,
+      "loss": 0.8516,
+      "step": 1535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981810130177092e-05,
+      "loss": 1.0156,
+      "step": 1536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817744851453516e-05,
+      "loss": 0.9883,
+      "step": 1537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817388055438162e-05,
+      "loss": 0.9688,
+      "step": 1538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817030913737418e-05,
+      "loss": 0.8555,
+      "step": 1539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981667342636386e-05,
+      "loss": 0.9688,
+      "step": 1540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981631559333007e-05,
+      "loss": 0.9141,
+      "step": 1541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981595741464865e-05,
+      "loss": 1.0391,
+      "step": 1542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815598890332216e-05,
+      "loss": 1.0703,
+      "step": 1543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815240020393388e-05,
+      "loss": 0.9258,
+      "step": 1544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981488080484481e-05,
+      "loss": 1.0156,
+      "step": 1545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981452124369912e-05,
+      "loss": 0.9375,
+      "step": 1546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814161336968982e-05,
+      "loss": 0.9297,
+      "step": 1547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981380108466707e-05,
+      "loss": 0.9023,
+      "step": 1548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813440486806076e-05,
+      "loss": 0.9023,
+      "step": 1549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813079543398684e-05,
+      "loss": 0.8633,
+      "step": 1550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981271825445761e-05,
+      "loss": 0.8984,
+      "step": 1551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981235661999558e-05,
+      "loss": 1.0469,
+      "step": 1552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981199464002532e-05,
+      "loss": 0.9336,
+      "step": 1553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811632314559582e-05,
+      "loss": 1.0156,
+      "step": 1554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811269643611117e-05,
+      "loss": 0.9688,
+      "step": 1555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98109066271927e-05,
+      "loss": 0.8984,
+      "step": 1556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810543265317116e-05,
+      "loss": 0.9336,
+      "step": 1557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810179557997155e-05,
+      "loss": 0.9219,
+      "step": 1558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809815505245625e-05,
+      "loss": 0.9531,
+      "step": 1559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809451107075342e-05,
+      "loss": 0.9961,
+      "step": 1560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809086363499143e-05,
+      "loss": 0.8984,
+      "step": 1561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808721274529866e-05,
+      "loss": 1.0312,
+      "step": 1562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808355840180368e-05,
+      "loss": 0.9297,
+      "step": 1563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980799006046352e-05,
+      "loss": 1.0,
+      "step": 1564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980762393539219e-05,
+      "loss": 1.0547,
+      "step": 1565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807257464979287e-05,
+      "loss": 1.0078,
+      "step": 1566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806890649237704e-05,
+      "loss": 0.8008,
+      "step": 1567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806523488180362e-05,
+      "loss": 0.9023,
+      "step": 1568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806155981820183e-05,
+      "loss": 0.8867,
+      "step": 1569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805788130170112e-05,
+      "loss": 0.8828,
+      "step": 1570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805419933243097e-05,
+      "loss": 1.0234,
+      "step": 1571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805051391052114e-05,
+      "loss": 0.9414,
+      "step": 1572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804682503610128e-05,
+      "loss": 1.0234,
+      "step": 1573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804313270930134e-05,
+      "loss": 0.9531,
+      "step": 1574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803943693025132e-05,
+      "loss": 0.8672,
+      "step": 1575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980357376990813e-05,
+      "loss": 0.9219,
+      "step": 1576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803203501592168e-05,
+      "loss": 1.0391,
+      "step": 1577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980283288809027e-05,
+      "loss": 0.9453,
+      "step": 1578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980246192941549e-05,
+      "loss": 0.9961,
+      "step": 1579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802090625580893e-05,
+      "loss": 0.9414,
+      "step": 1580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980171897659955e-05,
+      "loss": 1.0,
+      "step": 1581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801346982484543e-05,
+      "loss": 0.9688,
+      "step": 1582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980097464324898e-05,
+      "loss": 0.957,
+      "step": 1583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800601958905966e-05,
+      "loss": 0.9492,
+      "step": 1584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800228929468627e-05,
+      "loss": 0.9375,
+      "step": 1585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799855554950095e-05,
+      "loss": 0.9023,
+      "step": 1586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799481835363517e-05,
+      "loss": 0.8906,
+      "step": 1587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799107770722056e-05,
+      "loss": 1.0859,
+      "step": 1588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798733361038874e-05,
+      "loss": 1.125,
+      "step": 1589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798358606327168e-05,
+      "loss": 1.0938,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797983506600125e-05,
+      "loss": 0.9336,
+      "step": 1591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797608061870953e-05,
+      "loss": 0.8945,
+      "step": 1592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797232272152877e-05,
+      "loss": 1.0078,
+      "step": 1593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796856137459122e-05,
+      "loss": 1.0469,
+      "step": 1594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796479657802936e-05,
+      "loss": 0.9766,
+      "step": 1595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979610283319758e-05,
+      "loss": 0.9648,
+      "step": 1596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795725663656317e-05,
+      "loss": 0.8711,
+      "step": 1597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795348149192424e-05,
+      "loss": 1.0078,
+      "step": 1598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97949702898192e-05,
+      "loss": 0.8906,
+      "step": 1599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979459208554995e-05,
+      "loss": 1.0078,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794213536397993e-05,
+      "loss": 1.0391,
+      "step": 1601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979383464237665e-05,
+      "loss": 1.125,
+      "step": 1602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793455403499272e-05,
+      "loss": 1.0625,
+      "step": 1603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793075819779205e-05,
+      "loss": 0.8555,
+      "step": 1604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979269589122982e-05,
+      "loss": 0.8789,
+      "step": 1605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979231561786449e-05,
+      "loss": 1.1953,
+      "step": 1606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791934999696615e-05,
+      "loss": 1.0703,
+      "step": 1607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791554036739583e-05,
+      "loss": 0.9922,
+      "step": 1608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791172729006815e-05,
+      "loss": 1.0,
+      "step": 1609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979079107651174e-05,
+      "loss": 0.957,
+      "step": 1610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790409079267794e-05,
+      "loss": 0.9023,
+      "step": 1611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790026737288428e-05,
+      "loss": 1.0781,
+      "step": 1612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789644050587106e-05,
+      "loss": 1.0781,
+      "step": 1613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97892610191773e-05,
+      "loss": 1.0391,
+      "step": 1614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97888776430725e-05,
+      "loss": 0.8633,
+      "step": 1615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788493922286203e-05,
+      "loss": 1.0391,
+      "step": 1616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978810985683192e-05,
+      "loss": 1.0312,
+      "step": 1617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978772544672318e-05,
+      "loss": 0.8789,
+      "step": 1618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787340691973513e-05,
+      "loss": 0.8984,
+      "step": 1619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978695559259647e-05,
+      "loss": 0.9961,
+      "step": 1620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786570148605608e-05,
+      "loss": 0.9258,
+      "step": 1621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786184360014498e-05,
+      "loss": 0.875,
+      "step": 1622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785798226836732e-05,
+      "loss": 0.9375,
+      "step": 1623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785411749085898e-05,
+      "loss": 0.9805,
+      "step": 1624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978502492677561e-05,
+      "loss": 0.9062,
+      "step": 1625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784637759919483e-05,
+      "loss": 0.9492,
+      "step": 1626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784250248531155e-05,
+      "loss": 0.8906,
+      "step": 1627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978386239262427e-05,
+      "loss": 0.9414,
+      "step": 1628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783474192212484e-05,
+      "loss": 0.957,
+      "step": 1629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783085647309462e-05,
+      "loss": 1.0078,
+      "step": 1630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782696757928895e-05,
+      "loss": 1.0156,
+      "step": 1631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782307524084466e-05,
+      "loss": 0.8672,
+      "step": 1632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978191794578989e-05,
+      "loss": 1.0625,
+      "step": 1633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781528023058873e-05,
+      "loss": 0.9297,
+      "step": 1634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978113775590516e-05,
+      "loss": 0.9531,
+      "step": 1635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978074714434248e-05,
+      "loss": 0.9492,
+      "step": 1636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978035618838459e-05,
+      "loss": 1.0781,
+      "step": 1637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.977996488804526e-05,
+      "loss": 1.0156,
+      "step": 1638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779573243338267e-05,
+      "loss": 0.7969,
+      "step": 1639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779181254277397e-05,
+      "loss": 0.8086,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778788920876458e-05,
+      "loss": 0.8867,
+      "step": 1641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778396243149264e-05,
+      "loss": 0.8594,
+      "step": 1642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778003221109637e-05,
+      "loss": 0.9844,
+      "step": 1643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777609854771423e-05,
+      "loss": 0.957,
+      "step": 1644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977721614414847e-05,
+      "loss": 0.8867,
+      "step": 1645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977682208925464e-05,
+      "loss": 0.8555,
+      "step": 1646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776427690103807e-05,
+      "loss": 1.0156,
+      "step": 1647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977603294670986e-05,
+      "loss": 1.0469,
+      "step": 1648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97756378590867e-05,
+      "loss": 0.9492,
+      "step": 1649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775242427248235e-05,
+      "loss": 0.8477,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774846651208395e-05,
+      "loss": 0.875,
+      "step": 1651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774450530981107e-05,
+      "loss": 0.9492,
+      "step": 1652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774054066580332e-05,
+      "loss": 0.9375,
+      "step": 1653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773657258020014e-05,
+      "loss": 1.0391,
+      "step": 1654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773260105314137e-05,
+      "loss": 0.9922,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772862608476683e-05,
+      "loss": 0.8477,
+      "step": 1656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772464767521645e-05,
+      "loss": 1.0859,
+      "step": 1657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772066582463037e-05,
+      "loss": 0.9492,
+      "step": 1658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771668053314874e-05,
+      "loss": 1.0469,
+      "step": 1659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771269180091192e-05,
+      "loss": 1.0312,
+      "step": 1660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977086996280604e-05,
+      "loss": 0.9258,
+      "step": 1661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770470401473462e-05,
+      "loss": 0.8945,
+      "step": 1662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977007049610754e-05,
+      "loss": 1.0547,
+      "step": 1663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769670246722354e-05,
+      "loss": 0.9844,
+      "step": 1664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976926965333199e-05,
+      "loss": 0.9961,
+      "step": 1665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976886871595056e-05,
+      "loss": 0.8867,
+      "step": 1666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768467434592177e-05,
+      "loss": 1.0,
+      "step": 1667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976806580927097e-05,
+      "loss": 0.9492,
+      "step": 1668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976766384000109e-05,
+      "loss": 1.0781,
+      "step": 1669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767261526796684e-05,
+      "loss": 1.0781,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766858869671916e-05,
+      "loss": 1.0469,
+      "step": 1671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976645586864097e-05,
+      "loss": 0.9844,
+      "step": 1672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766052523718034e-05,
+      "loss": 0.9258,
+      "step": 1673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765648834917307e-05,
+      "loss": 0.8398,
+      "step": 1674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976524480225301e-05,
+      "loss": 1.0391,
+      "step": 1675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764840425739363e-05,
+      "loss": 0.9219,
+      "step": 1676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764435705390606e-05,
+      "loss": 0.8516,
+      "step": 1677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764030641220998e-05,
+      "loss": 0.9258,
+      "step": 1678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976362523324479e-05,
+      "loss": 0.8516,
+      "step": 1679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763219481476264e-05,
+      "loss": 0.9609,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762813385929707e-05,
+      "loss": 1.1406,
+      "step": 1681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762406946619416e-05,
+      "loss": 0.9375,
+      "step": 1682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762000163559706e-05,
+      "loss": 1.1719,
+      "step": 1683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761593036764894e-05,
+      "loss": 1.0156,
+      "step": 1684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976118556624932e-05,
+      "loss": 0.9336,
+      "step": 1685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760777752027332e-05,
+      "loss": 0.9805,
+      "step": 1686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760369594113287e-05,
+      "loss": 1.1328,
+      "step": 1687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759961092521562e-05,
+      "loss": 0.9766,
+      "step": 1688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759552247266534e-05,
+      "loss": 1.0391,
+      "step": 1689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759143058362606e-05,
+      "loss": 1.0312,
+      "step": 1690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758733525824177e-05,
+      "loss": 0.9297,
+      "step": 1691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758323649665678e-05,
+      "loss": 0.9688,
+      "step": 1692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757913429901535e-05,
+      "loss": 1.0312,
+      "step": 1693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757502866546195e-05,
+      "loss": 0.9414,
+      "step": 1694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757091959614115e-05,
+      "loss": 1.0234,
+      "step": 1695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756680709119756e-05,
+      "loss": 0.9961,
+      "step": 1696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975626911507761e-05,
+      "loss": 0.9062,
+      "step": 1697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755857177502165e-05,
+      "loss": 0.8438,
+      "step": 1698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755444896407925e-05,
+      "loss": 1.0078,
+      "step": 1699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755032271809407e-05,
+      "loss": 1.0547,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754619303721143e-05,
+      "loss": 0.9219,
+      "step": 1701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754205992157672e-05,
+      "loss": 0.9453,
+      "step": 1702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753792337133545e-05,
+      "loss": 0.9922,
+      "step": 1703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975337833866333e-05,
+      "loss": 1.0,
+      "step": 1704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752963996761604e-05,
+      "loss": 1.0156,
+      "step": 1705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975254931144296e-05,
+      "loss": 0.9141,
+      "step": 1706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752134282721996e-05,
+      "loss": 0.9766,
+      "step": 1707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751718910613326e-05,
+      "loss": 0.9727,
+      "step": 1708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975130319513158e-05,
+      "loss": 1.0781,
+      "step": 1709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750887136291388e-05,
+      "loss": 0.9883,
+      "step": 1710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750470734107404e-05,
+      "loss": 0.9297,
+      "step": 1711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750053988594295e-05,
+      "loss": 0.918,
+      "step": 1712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749636899766728e-05,
+      "loss": 0.9883,
+      "step": 1713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749219467639395e-05,
+      "loss": 0.9453,
+      "step": 1714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974880169222699e-05,
+      "loss": 0.9492,
+      "step": 1715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748383573544222e-05,
+      "loss": 1.0078,
+      "step": 1716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747965111605822e-05,
+      "loss": 0.9609,
+      "step": 1717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747546306426518e-05,
+      "loss": 0.9219,
+      "step": 1718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974712715802106e-05,
+      "loss": 1.0,
+      "step": 1719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746707666404202e-05,
+      "loss": 0.9375,
+      "step": 1720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974628783159072e-05,
+      "loss": 1.0781,
+      "step": 1721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745867653595393e-05,
+      "loss": 0.9336,
+      "step": 1722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745447132433023e-05,
+      "loss": 1.0391,
+      "step": 1723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974502626811841e-05,
+      "loss": 0.9414,
+      "step": 1724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744605060666374e-05,
+      "loss": 1.0078,
+      "step": 1725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974418351009175e-05,
+      "loss": 1.0781,
+      "step": 1726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974376161640938e-05,
+      "loss": 0.8516,
+      "step": 1727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743339379634117e-05,
+      "loss": 0.9297,
+      "step": 1728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742916799780832e-05,
+      "loss": 1.0078,
+      "step": 1729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742493876864404e-05,
+      "loss": 1.0781,
+      "step": 1730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742070610899724e-05,
+      "loss": 0.8555,
+      "step": 1731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741647001901697e-05,
+      "loss": 0.9375,
+      "step": 1732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741223049885236e-05,
+      "loss": 1.1484,
+      "step": 1733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974079875486527e-05,
+      "loss": 1.0156,
+      "step": 1734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740374116856744e-05,
+      "loss": 0.8789,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739949135874606e-05,
+      "loss": 1.1094,
+      "step": 1736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739523811933815e-05,
+      "loss": 0.9648,
+      "step": 1737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973909814504936e-05,
+      "loss": 0.9414,
+      "step": 1738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738672135236218e-05,
+      "loss": 0.9961,
+      "step": 1739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738245782509396e-05,
+      "loss": 0.8633,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97378190868839e-05,
+      "loss": 0.9766,
+      "step": 1741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737392048374763e-05,
+      "loss": 0.875,
+      "step": 1742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736964666997015e-05,
+      "loss": 0.9297,
+      "step": 1743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973653694276571e-05,
+      "loss": 0.9648,
+      "step": 1744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736108875695903e-05,
+      "loss": 0.9688,
+      "step": 1745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735680465802673e-05,
+      "loss": 1.0469,
+      "step": 1746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735251713101097e-05,
+      "loss": 0.9375,
+      "step": 1747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973482261760628e-05,
+      "loss": 0.9414,
+      "step": 1748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973439317933333e-05,
+      "loss": 1.0078,
+      "step": 1749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733963398297365e-05,
+      "loss": 1.0,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973353327451352e-05,
+      "loss": 1.0078,
+      "step": 1751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733102807996937e-05,
+      "loss": 0.9727,
+      "step": 1752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732671998762778e-05,
+      "loss": 1.0938,
+      "step": 1753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973224084682621e-05,
+      "loss": 0.9453,
+      "step": 1754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973180935220242e-05,
+      "loss": 0.9922,
+      "step": 1755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973137751490659e-05,
+      "loss": 0.9805,
+      "step": 1756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973094533495394e-05,
+      "loss": 0.875,
+      "step": 1757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730512812359674e-05,
+      "loss": 0.8711,
+      "step": 1758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973007994713903e-05,
+      "loss": 0.918,
+      "step": 1759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729646739307244e-05,
+      "loss": 0.8789,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972921318887958e-05,
+      "loss": 1.0781,
+      "step": 1761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728779295871294e-05,
+      "loss": 0.918,
+      "step": 1762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972834506029767e-05,
+      "loss": 0.8164,
+      "step": 1763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727910482173995e-05,
+      "loss": 0.9648,
+      "step": 1764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972747556151557e-05,
+      "loss": 0.875,
+      "step": 1765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727040298337715e-05,
+      "loss": 0.9727,
+      "step": 1766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726604692655747e-05,
+      "loss": 0.9297,
+      "step": 1767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726168744485016e-05,
+      "loss": 0.9766,
+      "step": 1768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725732453840864e-05,
+      "loss": 1.0703,
+      "step": 1769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725295820738658e-05,
+      "loss": 0.9336,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724858845193768e-05,
+      "loss": 0.8203,
+      "step": 1771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972442152722158e-05,
+      "loss": 0.9531,
+      "step": 1772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723983866837502e-05,
+      "loss": 0.9102,
+      "step": 1773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723545864056932e-05,
+      "loss": 0.918,
+      "step": 1774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723107518895304e-05,
+      "loss": 1.0156,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722668831368044e-05,
+      "loss": 1.0234,
+      "step": 1776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722229801490606e-05,
+      "loss": 1.0469,
+      "step": 1777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721790429278443e-05,
+      "loss": 0.9102,
+      "step": 1778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972135071474703e-05,
+      "loss": 0.9648,
+      "step": 1779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972091065791185e-05,
+      "loss": 0.8008,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720470258788393e-05,
+      "loss": 0.9883,
+      "step": 1781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720029517392173e-05,
+      "loss": 0.8477,
+      "step": 1782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719588433738705e-05,
+      "loss": 1.0156,
+      "step": 1783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971914700784352e-05,
+      "loss": 0.9805,
+      "step": 1784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718705239722164e-05,
+      "loss": 0.957,
+      "step": 1785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971826312939019e-05,
+      "loss": 0.8984,
+      "step": 1786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717820676863165e-05,
+      "loss": 1.1328,
+      "step": 1787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717377882156672e-05,
+      "loss": 0.9688,
+      "step": 1788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97169347452863e-05,
+      "loss": 0.9961,
+      "step": 1789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716491266267652e-05,
+      "loss": 0.9844,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716047445116343e-05,
+      "loss": 0.9883,
+      "step": 1791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715603281848007e-05,
+      "loss": 1.0156,
+      "step": 1792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715158776478273e-05,
+      "loss": 1.0,
+      "step": 1793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97147139290228e-05,
+      "loss": 0.9102,
+      "step": 1794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971426873949725e-05,
+      "loss": 1.0156,
+      "step": 1795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97138232079173e-05,
+      "loss": 1.0625,
+      "step": 1796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971337733429864e-05,
+      "loss": 0.8164,
+      "step": 1797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971293111865696e-05,
+      "loss": 1.0312,
+      "step": 1798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971248456100798e-05,
+      "loss": 0.8945,
+      "step": 1799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712037661367423e-05,
+      "loss": 1.0156,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711590419751028e-05,
+      "loss": 1.0312,
+      "step": 1801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711142836174537e-05,
+      "loss": 0.9766,
+      "step": 1802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971069491065371e-05,
+      "loss": 0.9453,
+      "step": 1803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710246643204327e-05,
+      "loss": 0.8477,
+      "step": 1804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709798033842165e-05,
+      "loss": 0.9609,
+      "step": 1805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709349082583024e-05,
+      "loss": 1.0234,
+      "step": 1806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708899789442707e-05,
+      "loss": 0.9961,
+      "step": 1807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970845015443704e-05,
+      "loss": 0.8867,
+      "step": 1808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708000177581855e-05,
+      "loss": 0.9414,
+      "step": 1809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707549858892992e-05,
+      "loss": 1.0156,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970709919838631e-05,
+      "loss": 1.0547,
+      "step": 1811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9706648196077678e-05,
+      "loss": 0.9375,
+      "step": 1812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706196851982974e-05,
+      "loss": 1.0156,
+      "step": 1813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705745166118096e-05,
+      "loss": 1.0234,
+      "step": 1814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705293138498943e-05,
+      "loss": 0.8906,
+      "step": 1815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704840769141435e-05,
+      "loss": 0.8984,
+      "step": 1816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704388058061497e-05,
+      "loss": 0.9023,
+      "step": 1817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703935005275076e-05,
+      "loss": 0.8906,
+      "step": 1818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703481610798118e-05,
+      "loss": 0.8516,
+      "step": 1819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703027874646592e-05,
+      "loss": 0.9727,
+      "step": 1820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970257379683647e-05,
+      "loss": 1.0156,
+      "step": 1821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702119377383747e-05,
+      "loss": 1.1172,
+      "step": 1822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701664616304418e-05,
+      "loss": 0.875,
+      "step": 1823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97012095136145e-05,
+      "loss": 0.9102,
+      "step": 1824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700754069330016e-05,
+      "loss": 1.0859,
+      "step": 1825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700298283467006e-05,
+      "loss": 1.0156,
+      "step": 1826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969984215604151e-05,
+      "loss": 0.8672,
+      "step": 1827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699385687069598e-05,
+      "loss": 0.9219,
+      "step": 1828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969892887656734e-05,
+      "loss": 0.9531,
+      "step": 1829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698471724550826e-05,
+      "loss": 0.9102,
+      "step": 1830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698014231036146e-05,
+      "loss": 0.9609,
+      "step": 1831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697556396039408e-05,
+      "loss": 0.9414,
+      "step": 1832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697098219576737e-05,
+      "loss": 0.9766,
+      "step": 1833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696639701664266e-05,
+      "loss": 0.9375,
+      "step": 1834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696180842318143e-05,
+      "loss": 1.0469,
+      "step": 1835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695721641554514e-05,
+      "loss": 0.9375,
+      "step": 1836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695262099389562e-05,
+      "loss": 0.8984,
+      "step": 1837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969480221583946e-05,
+      "loss": 0.9219,
+      "step": 1838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694341990920403e-05,
+      "loss": 1.0469,
+      "step": 1839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96938814246486e-05,
+      "loss": 0.918,
+      "step": 1840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969342051704026e-05,
+      "loss": 1.1016,
+      "step": 1841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969295926811162e-05,
+      "loss": 0.9766,
+      "step": 1842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692497677878916e-05,
+      "loss": 0.9297,
+      "step": 1843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692035746358405e-05,
+      "loss": 0.918,
+      "step": 1844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691573473566352e-05,
+      "loss": 0.8086,
+      "step": 1845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969111085951903e-05,
+      "loss": 1.0078,
+      "step": 1846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690647904232733e-05,
+      "loss": 0.8984,
+      "step": 1847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969018460772376e-05,
+      "loss": 0.8477,
+      "step": 1848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968972097000843e-05,
+      "loss": 0.8594,
+      "step": 1849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689256991103058e-05,
+      "loss": 0.9805,
+      "step": 1850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688792671023992e-05,
+      "loss": 1.0859,
+      "step": 1851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688328009787573e-05,
+      "loss": 0.875,
+      "step": 1852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968786300741017e-05,
+      "loss": 1.0703,
+      "step": 1853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687397663908147e-05,
+      "loss": 1.0469,
+      "step": 1854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96869319792979e-05,
+      "loss": 0.9688,
+      "step": 1855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968646595359582e-05,
+      "loss": 1.0703,
+      "step": 1856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685999586818315e-05,
+      "loss": 1.0078,
+      "step": 1857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968553287898181e-05,
+      "loss": 0.9375,
+      "step": 1858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968506583010274e-05,
+      "loss": 0.9062,
+      "step": 1859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968459844019755e-05,
+      "loss": 0.9766,
+      "step": 1860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684130709282692e-05,
+      "loss": 0.9883,
+      "step": 1861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968366263737464e-05,
+      "loss": 0.8203,
+      "step": 1862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968319422448988e-05,
+      "loss": 0.9375,
+      "step": 1863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682725470644898e-05,
+      "loss": 1.0,
+      "step": 1864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96822563758562e-05,
+      "loss": 0.8672,
+      "step": 1865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681786940140303e-05,
+      "loss": 0.8711,
+      "step": 1866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681317163513745e-05,
+      "loss": 0.9805,
+      "step": 1867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968084704599306e-05,
+      "loss": 1.0469,
+      "step": 1868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680376587594802e-05,
+      "loss": 0.9336,
+      "step": 1869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679905788335537e-05,
+      "loss": 0.8516,
+      "step": 1870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679434648231844e-05,
+      "loss": 0.9961,
+      "step": 1871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678963167300315e-05,
+      "loss": 1.0391,
+      "step": 1872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678491345557547e-05,
+      "loss": 0.9766,
+      "step": 1873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678019183020156e-05,
+      "loss": 0.9336,
+      "step": 1874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677546679704765e-05,
+      "loss": 0.9531,
+      "step": 1875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677073835628012e-05,
+      "loss": 1.1016,
+      "step": 1876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967660065080655e-05,
+      "loss": 0.9336,
+      "step": 1877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967612712525704e-05,
+      "loss": 1.0391,
+      "step": 1878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675653258996153e-05,
+      "loss": 0.9141,
+      "step": 1879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675179052040573e-05,
+      "loss": 0.9297,
+      "step": 1880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674704504407003e-05,
+      "loss": 0.9609,
+      "step": 1881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674229616112147e-05,
+      "loss": 0.9766,
+      "step": 1882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673754387172735e-05,
+      "loss": 1.0312,
+      "step": 1883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673278817605488e-05,
+      "loss": 0.9141,
+      "step": 1884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672802907427163e-05,
+      "loss": 1.0,
+      "step": 1885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967232665665451e-05,
+      "loss": 0.8477,
+      "step": 1886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671850065304303e-05,
+      "loss": 0.8945,
+      "step": 1887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671373133393322e-05,
+      "loss": 0.875,
+      "step": 1888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967089586093836e-05,
+      "loss": 1.0625,
+      "step": 1889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670418247956224e-05,
+      "loss": 0.9375,
+      "step": 1890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966994029446373e-05,
+      "loss": 0.9648,
+      "step": 1891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96694620004777e-05,
+      "loss": 0.9648,
+      "step": 1892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668983366014993e-05,
+      "loss": 0.9844,
+      "step": 1893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966850439109245e-05,
+      "loss": 0.9609,
+      "step": 1894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668025075726938e-05,
+      "loss": 0.8672,
+      "step": 1895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667545419935338e-05,
+      "loss": 0.8477,
+      "step": 1896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667065423734538e-05,
+      "loss": 1.125,
+      "step": 1897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666585087141436e-05,
+      "loss": 0.8242,
+      "step": 1898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966610441017295e-05,
+      "loss": 0.8828,
+      "step": 1899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665623392846e-05,
+      "loss": 1.0,
+      "step": 1900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665142035177526e-05,
+      "loss": 0.9648,
+      "step": 1901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664660337184482e-05,
+      "loss": 0.9922,
+      "step": 1902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664178298883826e-05,
+      "loss": 0.8867,
+      "step": 1903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663695920292528e-05,
+      "loss": 0.8164,
+      "step": 1904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663213201427576e-05,
+      "loss": 1.0938,
+      "step": 1905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662730142305966e-05,
+      "loss": 1.0469,
+      "step": 1906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966224674294471e-05,
+      "loss": 0.9844,
+      "step": 1907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661763003360828e-05,
+      "loss": 0.9922,
+      "step": 1908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661278923571353e-05,
+      "loss": 0.9336,
+      "step": 1909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966079450359333e-05,
+      "loss": 0.9414,
+      "step": 1910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660309743443816e-05,
+      "loss": 1.1328,
+      "step": 1911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659824643139882e-05,
+      "loss": 0.9453,
+      "step": 1912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659339202698606e-05,
+      "loss": 0.9961,
+      "step": 1913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658853422137084e-05,
+      "loss": 0.9375,
+      "step": 1914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658367301472417e-05,
+      "loss": 0.9414,
+      "step": 1915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657880840721727e-05,
+      "loss": 0.9766,
+      "step": 1916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965739403990214e-05,
+      "loss": 0.9297,
+      "step": 1917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656906899030797e-05,
+      "loss": 1.0078,
+      "step": 1918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656419418124854e-05,
+      "loss": 1.0234,
+      "step": 1919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965593159720147e-05,
+      "loss": 1.0156,
+      "step": 1920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655443436277834e-05,
+      "loss": 0.918,
+      "step": 1921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965495493537112e-05,
+      "loss": 0.9766,
+      "step": 1922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654466094498533e-05,
+      "loss": 0.9961,
+      "step": 1923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653976913677296e-05,
+      "loss": 0.9531,
+      "step": 1924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965348739292462e-05,
+      "loss": 1.0078,
+      "step": 1925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652997532257752e-05,
+      "loss": 0.9375,
+      "step": 1926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652507331693936e-05,
+      "loss": 0.9688,
+      "step": 1927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652016791250432e-05,
+      "loss": 0.9141,
+      "step": 1928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651525910944512e-05,
+      "loss": 0.9727,
+      "step": 1929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651034690793466e-05,
+      "loss": 0.957,
+      "step": 1930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650543130814585e-05,
+      "loss": 1.0859,
+      "step": 1931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650051231025183e-05,
+      "loss": 1.0312,
+      "step": 1932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649558991442575e-05,
+      "loss": 1.0391,
+      "step": 1933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96490664120841e-05,
+      "loss": 0.8789,
+      "step": 1934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648573492967093e-05,
+      "loss": 1.0312,
+      "step": 1935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964808023410892e-05,
+      "loss": 1.1797,
+      "step": 1936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647586635526946e-05,
+      "loss": 0.9141,
+      "step": 1937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647092697238547e-05,
+      "loss": 0.957,
+      "step": 1938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964659841926112e-05,
+      "loss": 0.9492,
+      "step": 1939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646103801612072e-05,
+      "loss": 1.0234,
+      "step": 1940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645608844308814e-05,
+      "loss": 0.8945,
+      "step": 1941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645113547368772e-05,
+      "loss": 0.9258,
+      "step": 1942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964461791080939e-05,
+      "loss": 0.9023,
+      "step": 1943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644121934648122e-05,
+      "loss": 0.8828,
+      "step": 1944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643625618902432e-05,
+      "loss": 0.9688,
+      "step": 1945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643128963589792e-05,
+      "loss": 1.0234,
+      "step": 1946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642631968727698e-05,
+      "loss": 0.832,
+      "step": 1947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964213463433364e-05,
+      "loss": 0.9492,
+      "step": 1948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964163696042513e-05,
+      "loss": 0.9141,
+      "step": 1949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96411389470197e-05,
+      "loss": 0.8359,
+      "step": 1950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964064059413488e-05,
+      "loss": 1.1797,
+      "step": 1951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640141901788224e-05,
+      "loss": 0.9805,
+      "step": 1952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639642869997283e-05,
+      "loss": 1.0703,
+      "step": 1953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639143498779634e-05,
+      "loss": 1.0625,
+      "step": 1954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963864378815286e-05,
+      "loss": 0.8984,
+      "step": 1955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638143738134555e-05,
+      "loss": 0.9766,
+      "step": 1956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637643348742327e-05,
+      "loss": 0.9922,
+      "step": 1957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96371426199938e-05,
+      "loss": 0.8281,
+      "step": 1958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96366415519066e-05,
+      "loss": 0.9062,
+      "step": 1959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636140144498372e-05,
+      "loss": 0.8477,
+      "step": 1960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963563839778677e-05,
+      "loss": 0.9062,
+      "step": 1961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635136311789463e-05,
+      "loss": 0.8711,
+      "step": 1962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963463388652413e-05,
+      "loss": 0.9102,
+      "step": 1963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634131122008463e-05,
+      "loss": 0.9258,
+      "step": 1964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633628018260165e-05,
+      "loss": 0.8906,
+      "step": 1965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963312457529695e-05,
+      "loss": 1.0391,
+      "step": 1966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632620793136543e-05,
+      "loss": 0.7812,
+      "step": 1967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632116671796682e-05,
+      "loss": 0.8672,
+      "step": 1968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631612211295126e-05,
+      "loss": 1.0078,
+      "step": 1969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631107411649632e-05,
+      "loss": 0.9375,
+      "step": 1970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630602272877976e-05,
+      "loss": 1.0,
+      "step": 1971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630096794997947e-05,
+      "loss": 1.0469,
+      "step": 1972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629590978027336e-05,
+      "loss": 0.8633,
+      "step": 1973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962908482198396e-05,
+      "loss": 0.957,
+      "step": 1974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962857832688564e-05,
+      "loss": 0.8281,
+      "step": 1975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628071492750213e-05,
+      "loss": 1.0391,
+      "step": 1976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627564319595522e-05,
+      "loss": 0.9805,
+      "step": 1977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627056807439426e-05,
+      "loss": 0.9258,
+      "step": 1978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626548956299795e-05,
+      "loss": 1.0312,
+      "step": 1979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626040766194512e-05,
+      "loss": 0.8984,
+      "step": 1980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962553223714147e-05,
+      "loss": 1.0078,
+      "step": 1981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962502336915858e-05,
+      "loss": 0.9688,
+      "step": 1982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624514162263754e-05,
+      "loss": 1.0859,
+      "step": 1983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624004616474924e-05,
+      "loss": 0.9531,
+      "step": 1984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623494731810032e-05,
+      "loss": 0.8594,
+      "step": 1985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962298450828703e-05,
+      "loss": 0.8945,
+      "step": 1986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622473945923888e-05,
+      "loss": 1.0781,
+      "step": 1987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962196304473858e-05,
+      "loss": 0.9453,
+      "step": 1988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621451804749097e-05,
+      "loss": 0.9102,
+      "step": 1989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620940225973445e-05,
+      "loss": 0.9531,
+      "step": 1990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962042830842963e-05,
+      "loss": 0.9375,
+      "step": 1991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619916052135676e-05,
+      "loss": 1.0703,
+      "step": 1992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961940345710963e-05,
+      "loss": 0.957,
+      "step": 1993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618890523369534e-05,
+      "loss": 1.0234,
+      "step": 1994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618377250933452e-05,
+      "loss": 1.0156,
+      "step": 1995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617863639819455e-05,
+      "loss": 0.8984,
+      "step": 1996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961734969004563e-05,
+      "loss": 1.0547,
+      "step": 1997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961683540163007e-05,
+      "loss": 0.8945,
+      "step": 1998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616320774590893e-05,
+      "loss": 1.0156,
+      "step": 1999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615805808946212e-05,
+      "loss": 0.8867,
+      "step": 2000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615290504714158e-05,
+      "loss": 0.8359,
+      "step": 2001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614774861912883e-05,
+      "loss": 0.9961,
+      "step": 2002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961425888056054e-05,
+      "loss": 1.0547,
+      "step": 2003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96137425606753e-05,
+      "loss": 0.9922,
+      "step": 2004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613225902275334e-05,
+      "loss": 0.9375,
+      "step": 2005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612708905378845e-05,
+      "loss": 0.8516,
+      "step": 2006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612191570004032e-05,
+      "loss": 0.9375,
+      "step": 2007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611673896169115e-05,
+      "loss": 0.9609,
+      "step": 2008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611155883892318e-05,
+      "loss": 0.9492,
+      "step": 2009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961063753319188e-05,
+      "loss": 1.0625,
+      "step": 2010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610118844086062e-05,
+      "loss": 1.0156,
+      "step": 2011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609599816593117e-05,
+      "loss": 0.8242,
+      "step": 2012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609080450731327e-05,
+      "loss": 0.9766,
+      "step": 2013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608560746518977e-05,
+      "loss": 0.9297,
+      "step": 2014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960804070397437e-05,
+      "loss": 0.9336,
+      "step": 2015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960752032311581e-05,
+      "loss": 0.8867,
+      "step": 2016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960699960396163e-05,
+      "loss": 0.9766,
+      "step": 2017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960647854653016e-05,
+      "loss": 0.9297,
+      "step": 2018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960595715083975e-05,
+      "loss": 1.0234,
+      "step": 2019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605435416908752e-05,
+      "loss": 0.9375,
+      "step": 2020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604913344755543e-05,
+      "loss": 0.8594,
+      "step": 2021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604390934398508e-05,
+      "loss": 0.9258,
+      "step": 2022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960386818585604e-05,
+      "loss": 1.0703,
+      "step": 2023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603345099146544e-05,
+      "loss": 0.8242,
+      "step": 2024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602821674288437e-05,
+      "loss": 0.9219,
+      "step": 2025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602297911300155e-05,
+      "loss": 0.9609,
+      "step": 2026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960177381020013e-05,
+      "loss": 0.9766,
+      "step": 2027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601249371006833e-05,
+      "loss": 1.0234,
+      "step": 2028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960072459373872e-05,
+      "loss": 0.957,
+      "step": 2029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600199478414268e-05,
+      "loss": 0.9453,
+      "step": 2030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599674025051972e-05,
+      "loss": 0.9453,
+      "step": 2031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959914823367033e-05,
+      "loss": 0.875,
+      "step": 2032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598622104287858e-05,
+      "loss": 0.9453,
+      "step": 2033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959809563692308e-05,
+      "loss": 0.9258,
+      "step": 2034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959756883159454e-05,
+      "loss": 1.0547,
+      "step": 2035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959704168832078e-05,
+      "loss": 0.9609,
+      "step": 2036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959651420712036e-05,
+      "loss": 1.0625,
+      "step": 2037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595986388011866e-05,
+      "loss": 0.9297,
+      "step": 2038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959545823101387e-05,
+      "loss": 0.8359,
+      "step": 2039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.9297,
+      "step": 2040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959440090342379e-05,
+      "loss": 0.9805,
+      "step": 2041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593871732868935e-05,
+      "loss": 0.8945,
+      "step": 2042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593342224499045e-05,
+      "loss": 0.9141,
+      "step": 2043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959281237833276e-05,
+      "loss": 0.9727,
+      "step": 2044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959228219438874e-05,
+      "loss": 1.0312,
+      "step": 2045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959175167268566e-05,
+      "loss": 0.9648,
+      "step": 2046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591220813242185e-05,
+      "loss": 1.0078,
+      "step": 2047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590689616077022e-05,
+      "loss": 0.9961,
+      "step": 2048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590158081208866e-05,
+      "loss": 0.9336,
+      "step": 2049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958962620865644e-05,
+      "loss": 0.8047,
+      "step": 2050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958909399843847e-05,
+      "loss": 1.0078,
+      "step": 2051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958856145057369e-05,
+      "loss": 1.0,
+      "step": 2052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588028565080863e-05,
+      "loss": 0.9297,
+      "step": 2053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587495341978742e-05,
+      "loss": 0.918,
+      "step": 2054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586961781286108e-05,
+      "loss": 1.0078,
+      "step": 2055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586427883021747e-05,
+      "loss": 0.9766,
+      "step": 2056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958589364720446e-05,
+      "loss": 0.918,
+      "step": 2057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585359073853052e-05,
+      "loss": 0.8633,
+      "step": 2058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958482416298636e-05,
+      "loss": 0.9219,
+      "step": 2059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95842889146232e-05,
+      "loss": 0.9141,
+      "step": 2060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583753328782437e-05,
+      "loss": 0.9922,
+      "step": 2061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958321740548292e-05,
+      "loss": 1.0938,
+      "step": 2062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958268114474352e-05,
+      "loss": 0.8633,
+      "step": 2063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582144546583124e-05,
+      "loss": 0.9219,
+      "step": 2064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581607611020625e-05,
+      "loss": 0.8828,
+      "step": 2065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958107033807492e-05,
+      "loss": 0.9297,
+      "step": 2066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580532727764946e-05,
+      "loss": 0.8242,
+      "step": 2067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579994780109615e-05,
+      "loss": 1.0312,
+      "step": 2068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957945649512788e-05,
+      "loss": 0.9375,
+      "step": 2069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578917872838687e-05,
+      "loss": 0.9531,
+      "step": 2070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957837891326101e-05,
+      "loss": 1.0703,
+      "step": 2071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957783961641382e-05,
+      "loss": 1.0,
+      "step": 2072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957729998231611e-05,
+      "loss": 0.9414,
+      "step": 2073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957676001098688e-05,
+      "loss": 0.9453,
+      "step": 2074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576219702445144e-05,
+      "loss": 0.9336,
+      "step": 2075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575679056709925e-05,
+      "loss": 0.8164,
+      "step": 2076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575138073800265e-05,
+      "loss": 0.8047,
+      "step": 2077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574596753735205e-05,
+      "loss": 0.9961,
+      "step": 2078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574055096533814e-05,
+      "loss": 0.9062,
+      "step": 2079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957351310221516e-05,
+      "loss": 0.8633,
+      "step": 2080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572970770798324e-05,
+      "loss": 0.9453,
+      "step": 2081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572428102302412e-05,
+      "loss": 1.0391,
+      "step": 2082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571885096746525e-05,
+      "loss": 0.8984,
+      "step": 2083
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571341754149787e-05,
+      "loss": 0.9609,
+      "step": 2084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570798074531324e-05,
+      "loss": 0.8828,
+      "step": 2085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570254057910283e-05,
+      "loss": 1.0469,
+      "step": 2086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569709704305823e-05,
+      "loss": 0.9375,
+      "step": 2087
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956916501373711e-05,
+      "loss": 0.918,
+      "step": 2088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568619986223316e-05,
+      "loss": 1.0234,
+      "step": 2089
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568074621783643e-05,
+      "loss": 0.9141,
+      "step": 2090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567528920437292e-05,
+      "loss": 1.0156,
+      "step": 2091
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956698288220347e-05,
+      "loss": 0.9023,
+      "step": 2092
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566436507101412e-05,
+      "loss": 0.9375,
+      "step": 2093
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956588979515035e-05,
+      "loss": 0.957,
+      "step": 2094
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565342746369545e-05,
+      "loss": 0.8711,
+      "step": 2095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956479536077825e-05,
+      "loss": 0.9414,
+      "step": 2096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564247638395742e-05,
+      "loss": 0.9961,
+      "step": 2097
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956369957924131e-05,
+      "loss": 0.8984,
+      "step": 2098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563151183334246e-05,
+      "loss": 0.8086,
+      "step": 2099
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562602450693865e-05,
+      "loss": 0.8633,
+      "step": 2100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562053381339486e-05,
+      "loss": 0.9219,
+      "step": 2101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561503975290446e-05,
+      "loss": 0.9102,
+      "step": 2102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560954232566088e-05,
+      "loss": 0.9688,
+      "step": 2103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560404153185767e-05,
+      "loss": 0.9375,
+      "step": 2104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559853737168858e-05,
+      "loss": 0.9531,
+      "step": 2105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559302984534733e-05,
+      "loss": 0.9102,
+      "step": 2106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558751895302796e-05,
+      "loss": 0.9297,
+      "step": 2107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558200469492445e-05,
+      "loss": 0.9648,
+      "step": 2108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557648707123093e-05,
+      "loss": 1.0938,
+      "step": 2109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557096608214178e-05,
+      "loss": 0.9492,
+      "step": 2110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955654417278513e-05,
+      "loss": 0.875,
+      "step": 2111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555991400855414e-05,
+      "loss": 0.9492,
+      "step": 2112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955543829244448e-05,
+      "loss": 0.9258,
+      "step": 2113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554884847571817e-05,
+      "loss": 0.9492,
+      "step": 2114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95543310662569e-05,
+      "loss": 0.8633,
+      "step": 2115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553776948519238e-05,
+      "loss": 1.0078,
+      "step": 2116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553222494378333e-05,
+      "loss": 0.9844,
+      "step": 2117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955266770385372e-05,
+      "loss": 0.9336,
+      "step": 2118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552112576964923e-05,
+      "loss": 0.8359,
+      "step": 2119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551557113731497e-05,
+      "loss": 0.9102,
+      "step": 2120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551001314172996e-05,
+      "loss": 0.9219,
+      "step": 2121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550445178308992e-05,
+      "loss": 0.918,
+      "step": 2122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549888706159067e-05,
+      "loss": 0.9883,
+      "step": 2123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954933189774282e-05,
+      "loss": 0.9883,
+      "step": 2124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548774753079843e-05,
+      "loss": 1.0156,
+      "step": 2125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548217272189772e-05,
+      "loss": 0.9336,
+      "step": 2126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547659455092226e-05,
+      "loss": 0.9102,
+      "step": 2127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954710130180685e-05,
+      "loss": 0.8398,
+      "step": 2128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546542812353296e-05,
+      "loss": 0.875,
+      "step": 2129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954598398675123e-05,
+      "loss": 0.9688,
+      "step": 2130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545424825020326e-05,
+      "loss": 0.8711,
+      "step": 2131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544865327180283e-05,
+      "loss": 0.9258,
+      "step": 2132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544305493250788e-05,
+      "loss": 0.9219,
+      "step": 2133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954374532325156e-05,
+      "loss": 0.8906,
+      "step": 2134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543184817202326e-05,
+      "loss": 0.9883,
+      "step": 2135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542623975122822e-05,
+      "loss": 0.9414,
+      "step": 2136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954206279703279e-05,
+      "loss": 1.0156,
+      "step": 2137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541501282952e-05,
+      "loss": 0.9883,
+      "step": 2138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954093943290021e-05,
+      "loss": 0.9219,
+      "step": 2139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540377246897216e-05,
+      "loss": 0.8203,
+      "step": 2140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539814724962808e-05,
+      "loss": 0.918,
+      "step": 2141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539251867116793e-05,
+      "loss": 0.9609,
+      "step": 2142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538688673378993e-05,
+      "loss": 0.9453,
+      "step": 2143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538125143769237e-05,
+      "loss": 0.9336,
+      "step": 2144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537561278307367e-05,
+      "loss": 0.9336,
+      "step": 2145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536997077013236e-05,
+      "loss": 0.8242,
+      "step": 2146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536432539906718e-05,
+      "loss": 0.9727,
+      "step": 2147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535867667007682e-05,
+      "loss": 0.9414,
+      "step": 2148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535302458336024e-05,
+      "loss": 0.9648,
+      "step": 2149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534736913911643e-05,
+      "loss": 0.8516,
+      "step": 2150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534171033754453e-05,
+      "loss": 1.0547,
+      "step": 2151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953360481788438e-05,
+      "loss": 0.9219,
+      "step": 2152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533038266321363e-05,
+      "loss": 1.0312,
+      "step": 2153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953247137908535e-05,
+      "loss": 0.8867,
+      "step": 2154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95319041561963e-05,
+      "loss": 0.9219,
+      "step": 2155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531336597674192e-05,
+      "loss": 0.9688,
+      "step": 2156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9530768703539002e-05,
+      "loss": 0.8867,
+      "step": 2157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953020047381073e-05,
+      "loss": 0.9688,
+      "step": 2158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529631908509387e-05,
+      "loss": 0.918,
+      "step": 2159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952906300765499e-05,
+      "loss": 0.8594,
+      "step": 2160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952849377126757e-05,
+      "loss": 1.0078,
+      "step": 2161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527924199367174e-05,
+      "loss": 1.0156,
+      "step": 2162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527354291973856e-05,
+      "loss": 0.8789,
+      "step": 2163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526784049107684e-05,
+      "loss": 0.9336,
+      "step": 2164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526213470788734e-05,
+      "loss": 0.918,
+      "step": 2165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525642557037103e-05,
+      "loss": 1.0156,
+      "step": 2166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525071307872886e-05,
+      "loss": 1.0781,
+      "step": 2167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95244997233162e-05,
+      "loss": 0.9062,
+      "step": 2168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523927803387178e-05,
+      "loss": 0.9258,
+      "step": 2169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952335554810595e-05,
+      "loss": 1.0312,
+      "step": 2170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952278295749267e-05,
+      "loss": 0.8281,
+      "step": 2171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95222100315675e-05,
+      "loss": 0.9609,
+      "step": 2172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521636770350612e-05,
+      "loss": 0.9062,
+      "step": 2173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952106317386219e-05,
+      "loss": 1.0312,
+      "step": 2174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520489242122434e-05,
+      "loss": 0.9375,
+      "step": 2175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519914975151554e-05,
+      "loss": 0.9336,
+      "step": 2176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951934037296977e-05,
+      "loss": 1.0078,
+      "step": 2177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518765435597306e-05,
+      "loss": 0.9727,
+      "step": 2178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951819016305442e-05,
+      "loss": 0.8555,
+      "step": 2179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517614555361363e-05,
+      "loss": 0.8945,
+      "step": 2180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95170386125384e-05,
+      "loss": 0.957,
+      "step": 2181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516462334605815e-05,
+      "loss": 0.9141,
+      "step": 2182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515885721583898e-05,
+      "loss": 1.2188,
+      "step": 2183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951530877349295e-05,
+      "loss": 0.9258,
+      "step": 2184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514731490353292e-05,
+      "loss": 0.9492,
+      "step": 2185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514153872185243e-05,
+      "loss": 0.9219,
+      "step": 2186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513575919009152e-05,
+      "loss": 0.9102,
+      "step": 2187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512997630845358e-05,
+      "loss": 0.9414,
+      "step": 2188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512419007714232e-05,
+      "loss": 0.9219,
+      "step": 2189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951184004963615e-05,
+      "loss": 1.0078,
+      "step": 2190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511260756631487e-05,
+      "loss": 0.9062,
+      "step": 2191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951068112872065e-05,
+      "loss": 0.9062,
+      "step": 2192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510101165924045e-05,
+      "loss": 0.8711,
+      "step": 2193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509520868262095e-05,
+      "loss": 0.8672,
+      "step": 2194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508940235755234e-05,
+      "loss": 1.0703,
+      "step": 2195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508359268423903e-05,
+      "loss": 1.0391,
+      "step": 2196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507777966288563e-05,
+      "loss": 1.0469,
+      "step": 2197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950719632936968e-05,
+      "loss": 0.9297,
+      "step": 2198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506614357687735e-05,
+      "loss": 0.9609,
+      "step": 2199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506032051263224e-05,
+      "loss": 0.9102,
+      "step": 2200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950544941011664e-05,
+      "loss": 0.9453,
+      "step": 2201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504866434268514e-05,
+      "loss": 0.9453,
+      "step": 2202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950428312373936e-05,
+      "loss": 0.9453,
+      "step": 2203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503699478549724e-05,
+      "loss": 0.9375,
+      "step": 2204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503115498720156e-05,
+      "loss": 0.918,
+      "step": 2205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502531184271218e-05,
+      "loss": 0.9961,
+      "step": 2206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501946535223487e-05,
+      "loss": 0.8945,
+      "step": 2207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501361551597545e-05,
+      "loss": 0.9844,
+      "step": 2208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500776233413995e-05,
+      "loss": 1.0,
+      "step": 2209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950019058069344e-05,
+      "loss": 0.9766,
+      "step": 2210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949960459345651e-05,
+      "loss": 0.9609,
+      "step": 2211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499018271723836e-05,
+      "loss": 0.9531,
+      "step": 2212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949843161551606e-05,
+      "loss": 0.875,
+      "step": 2213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497844624853842e-05,
+      "loss": 1.0,
+      "step": 2214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949725729975785e-05,
+      "loss": 0.8438,
+      "step": 2215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496669640248763e-05,
+      "loss": 0.8945,
+      "step": 2216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496081646347277e-05,
+      "loss": 0.8633,
+      "step": 2217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495493318074093e-05,
+      "loss": 0.8633,
+      "step": 2218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494904655449934e-05,
+      "loss": 1.1719,
+      "step": 2219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494315658495514e-05,
+      "loss": 0.8672,
+      "step": 2220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493726327231586e-05,
+      "loss": 0.957,
+      "step": 2221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493136661678896e-05,
+      "loss": 0.9844,
+      "step": 2222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492546661858202e-05,
+      "loss": 0.9375,
+      "step": 2223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491956327790287e-05,
+      "loss": 0.9453,
+      "step": 2224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491365659495935e-05,
+      "loss": 0.8906,
+      "step": 2225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949077465699594e-05,
+      "loss": 0.957,
+      "step": 2226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949018332031112e-05,
+      "loss": 1.0,
+      "step": 2227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489591649462288e-05,
+      "loss": 0.9688,
+      "step": 2228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488999644470283e-05,
+      "loss": 0.9844,
+      "step": 2229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948840730535595e-05,
+      "loss": 0.9375,
+      "step": 2230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487814632140148e-05,
+      "loss": 0.8555,
+      "step": 2231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487221624843743e-05,
+      "loss": 0.918,
+      "step": 2232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486628283487613e-05,
+      "loss": 0.9883,
+      "step": 2233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486034608092657e-05,
+      "loss": 0.957,
+      "step": 2234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485440598679774e-05,
+      "loss": 0.9414,
+      "step": 2235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948484625526988e-05,
+      "loss": 0.8359,
+      "step": 2236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484251577883914e-05,
+      "loss": 0.9961,
+      "step": 2237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94836565665428e-05,
+      "loss": 1.0781,
+      "step": 2238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483061221267493e-05,
+      "loss": 0.9062,
+      "step": 2239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482465542078965e-05,
+      "loss": 1.0391,
+      "step": 2240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481869528998177e-05,
+      "loss": 0.8672,
+      "step": 2241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948127318204613e-05,
+      "loss": 0.9531,
+      "step": 2242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948067650124381e-05,
+      "loss": 1.0078,
+      "step": 2243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480079486612235e-05,
+      "loss": 0.8125,
+      "step": 2244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479482138172422e-05,
+      "loss": 0.9883,
+      "step": 2245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478884455945408e-05,
+      "loss": 0.9023,
+      "step": 2246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478286439952238e-05,
+      "loss": 0.8438,
+      "step": 2247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477688090213968e-05,
+      "loss": 1.0,
+      "step": 2248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477089406751662e-05,
+      "loss": 0.9219,
+      "step": 2249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476490389586412e-05,
+      "loss": 1.0859,
+      "step": 2250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475891038739302e-05,
+      "loss": 0.9922,
+      "step": 2251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475291354231433e-05,
+      "loss": 0.9922,
+      "step": 2252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474691336083934e-05,
+      "loss": 0.957,
+      "step": 2253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474090984317916e-05,
+      "loss": 0.9883,
+      "step": 2254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947349029895453e-05,
+      "loss": 0.8906,
+      "step": 2255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472889280014924e-05,
+      "loss": 1.1328,
+      "step": 2256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947228792752026e-05,
+      "loss": 0.875,
+      "step": 2257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471686241491713e-05,
+      "loss": 1.0469,
+      "step": 2258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471084221950466e-05,
+      "loss": 0.9492,
+      "step": 2259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470481868917725e-05,
+      "loss": 0.918,
+      "step": 2260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469879182414692e-05,
+      "loss": 0.9258,
+      "step": 2261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469276162462598e-05,
+      "loss": 0.9805,
+      "step": 2262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946867280908266e-05,
+      "loss": 0.9453,
+      "step": 2263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468069122296138e-05,
+      "loss": 0.8906,
+      "step": 2264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946746510212428e-05,
+      "loss": 0.9922,
+      "step": 2265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466860748588365e-05,
+      "loss": 0.9375,
+      "step": 2266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946625606170966e-05,
+      "loss": 0.9453,
+      "step": 2267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465651041509464e-05,
+      "loss": 0.9805,
+      "step": 2268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946504568800908e-05,
+      "loss": 1.0078,
+      "step": 2269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464440001229826e-05,
+      "loss": 0.8984,
+      "step": 2270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463833981193024e-05,
+      "loss": 1.0547,
+      "step": 2271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946322762792002e-05,
+      "loss": 0.8594,
+      "step": 2272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462620941432154e-05,
+      "loss": 0.957,
+      "step": 2273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94620139217508e-05,
+      "loss": 0.9102,
+      "step": 2274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461406568897322e-05,
+      "loss": 0.9141,
+      "step": 2275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460798882893113e-05,
+      "loss": 0.9336,
+      "step": 2276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460190863759564e-05,
+      "loss": 1.0,
+      "step": 2277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945958251151809e-05,
+      "loss": 1.0469,
+      "step": 2278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458973826190113e-05,
+      "loss": 1.0234,
+      "step": 2279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458364807797058e-05,
+      "loss": 0.9219,
+      "step": 2280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945775545636038e-05,
+      "loss": 1.0859,
+      "step": 2281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457145771901528e-05,
+      "loss": 0.9062,
+      "step": 2282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456535754441968e-05,
+      "loss": 1.0078,
+      "step": 2283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945592540400319e-05,
+      "loss": 0.9219,
+      "step": 2284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945531472060667e-05,
+      "loss": 0.9023,
+      "step": 2285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454703704273926e-05,
+      "loss": 0.9258,
+      "step": 2286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454092355026466e-05,
+      "loss": 0.9219,
+      "step": 2287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453480672885814e-05,
+      "loss": 0.8828,
+      "step": 2288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452868657873513e-05,
+      "loss": 0.9062,
+      "step": 2289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452256310011114e-05,
+      "loss": 0.9883,
+      "step": 2290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945164362932017e-05,
+      "loss": 0.8789,
+      "step": 2291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451030615822266e-05,
+      "loss": 0.8359,
+      "step": 2292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450417269538978e-05,
+      "loss": 0.9336,
+      "step": 2293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449803590491912e-05,
+      "loss": 0.9961,
+      "step": 2294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449189578702664e-05,
+      "loss": 0.8438,
+      "step": 2295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448575234192867e-05,
+      "loss": 0.9023,
+      "step": 2296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447960556984144e-05,
+      "loss": 1.0156,
+      "step": 2297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447345547098143e-05,
+      "loss": 0.8711,
+      "step": 2298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944673020455652e-05,
+      "loss": 0.957,
+      "step": 2299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944611452938094e-05,
+      "loss": 0.9531,
+      "step": 2300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445498521593084e-05,
+      "loss": 1.0234,
+      "step": 2301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944488218121464e-05,
+      "loss": 0.7969,
+      "step": 2302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944426550826731e-05,
+      "loss": 0.8398,
+      "step": 2303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944364850277281e-05,
+      "loss": 0.832,
+      "step": 2304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443031164752867e-05,
+      "loss": 0.8828,
+      "step": 2305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442413494229214e-05,
+      "loss": 0.8633,
+      "step": 2306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944179549122361e-05,
+      "loss": 0.9141,
+      "step": 2307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94411771557578e-05,
+      "loss": 0.9141,
+      "step": 2308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944055848785357e-05,
+      "loss": 0.8828,
+      "step": 2309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439939487532696e-05,
+      "loss": 0.9766,
+      "step": 2310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439320154816982e-05,
+      "loss": 1.0078,
+      "step": 2311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438700489728227e-05,
+      "loss": 1.0312,
+      "step": 2312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438080492288252e-05,
+      "loss": 0.7773,
+      "step": 2313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94374601625189e-05,
+      "loss": 0.8672,
+      "step": 2314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436839500441997e-05,
+      "loss": 0.9023,
+      "step": 2315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436218506079405e-05,
+      "loss": 0.8945,
+      "step": 2316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435597179452992e-05,
+      "loss": 1.0156,
+      "step": 2317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434975520584634e-05,
+      "loss": 0.8945,
+      "step": 2318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434353529496216e-05,
+      "loss": 0.8945,
+      "step": 2319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943373120620965e-05,
+      "loss": 0.8516,
+      "step": 2320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943310855074684e-05,
+      "loss": 0.9258,
+      "step": 2321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943248556312971e-05,
+      "loss": 0.8984,
+      "step": 2322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431862243380205e-05,
+      "loss": 1.0078,
+      "step": 2323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431238591520267e-05,
+      "loss": 0.9336,
+      "step": 2324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430614607571852e-05,
+      "loss": 1.0625,
+      "step": 2325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942999029155694e-05,
+      "loss": 0.8828,
+      "step": 2326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429365643497508e-05,
+      "loss": 0.8125,
+      "step": 2327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428740663415552e-05,
+      "loss": 0.8555,
+      "step": 2328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942811535133308e-05,
+      "loss": 0.8242,
+      "step": 2329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942748970727211e-05,
+      "loss": 0.9219,
+      "step": 2330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426863731254674e-05,
+      "loss": 0.8672,
+      "step": 2331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426237423302806e-05,
+      "loss": 0.9258,
+      "step": 2332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425610783438567e-05,
+      "loss": 0.9766,
+      "step": 2333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942498381168402e-05,
+      "loss": 1.0938,
+      "step": 2334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942435650806124e-05,
+      "loss": 0.9141,
+      "step": 2335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423728872592314e-05,
+      "loss": 0.8438,
+      "step": 2336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423100905299348e-05,
+      "loss": 0.9219,
+      "step": 2337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422472606204444e-05,
+      "loss": 1.0469,
+      "step": 2338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942184397532974e-05,
+      "loss": 0.9922,
+      "step": 2339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421215012697353e-05,
+      "loss": 1.0156,
+      "step": 2340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420585718329446e-05,
+      "loss": 1.1641,
+      "step": 2341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941995609224817e-05,
+      "loss": 1.1172,
+      "step": 2342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419326134475693e-05,
+      "loss": 0.957,
+      "step": 2343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418695845034197e-05,
+      "loss": 0.9766,
+      "step": 2344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418065223945883e-05,
+      "loss": 0.8711,
+      "step": 2345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417434271232948e-05,
+      "loss": 1.0391,
+      "step": 2346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416802986917615e-05,
+      "loss": 0.9766,
+      "step": 2347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416171371022104e-05,
+      "loss": 1.0234,
+      "step": 2348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941553942356867e-05,
+      "loss": 0.9023,
+      "step": 2349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414907144579547e-05,
+      "loss": 1.0625,
+      "step": 2350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414274534077008e-05,
+      "loss": 0.8789,
+      "step": 2351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413641592083327e-05,
+      "loss": 0.9414,
+      "step": 2352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941300831862079e-05,
+      "loss": 1.0312,
+      "step": 2353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94123747137117e-05,
+      "loss": 0.9922,
+      "step": 2354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411740777378364e-05,
+      "loss": 0.9414,
+      "step": 2355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94111065096431e-05,
+      "loss": 0.9648,
+      "step": 2356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410471910528245e-05,
+      "loss": 1.0156,
+      "step": 2357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409836980056148e-05,
+      "loss": 0.8516,
+      "step": 2358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940920171824916e-05,
+      "loss": 0.9375,
+      "step": 2359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940856612512965e-05,
+      "loss": 0.9336,
+      "step": 2360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407930200720005e-05,
+      "loss": 0.9219,
+      "step": 2361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940729394504261e-05,
+      "loss": 1.0,
+      "step": 2362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406657358119866e-05,
+      "loss": 0.9453,
+      "step": 2363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406020439974197e-05,
+      "loss": 1.0078,
+      "step": 2364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405383190628023e-05,
+      "loss": 0.8906,
+      "step": 2365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404745610103785e-05,
+      "loss": 0.9727,
+      "step": 2366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404107698423936e-05,
+      "loss": 0.9414,
+      "step": 2367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940346945561093e-05,
+      "loss": 0.8867,
+      "step": 2368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402830881687253e-05,
+      "loss": 0.9727,
+      "step": 2369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402191976675376e-05,
+      "loss": 0.957,
+      "step": 2370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401552740597805e-05,
+      "loss": 1.0547,
+      "step": 2371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400913173477045e-05,
+      "loss": 0.8516,
+      "step": 2372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940027327533562e-05,
+      "loss": 0.9844,
+      "step": 2373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939963304619606e-05,
+      "loss": 0.9336,
+      "step": 2374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398992486080902e-05,
+      "loss": 0.8945,
+      "step": 2375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939835159501271e-05,
+      "loss": 1.0391,
+      "step": 2376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397710373014048e-05,
+      "loss": 0.9805,
+      "step": 2377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939706882010749e-05,
+      "loss": 1.0859,
+      "step": 2378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939642693631563e-05,
+      "loss": 0.875,
+      "step": 2379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395784721661074e-05,
+      "loss": 0.9766,
+      "step": 2380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939514217616643e-05,
+      "loss": 0.9609,
+      "step": 2381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939449929985432e-05,
+      "loss": 1.0,
+      "step": 2382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939385609274739e-05,
+      "loss": 0.918,
+      "step": 2383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939321255486828e-05,
+      "loss": 0.832,
+      "step": 2384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392568686239653e-05,
+      "loss": 0.9531,
+      "step": 2385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939192448688418e-05,
+      "loss": 1.0625,
+      "step": 2386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391279956824546e-05,
+      "loss": 0.9805,
+      "step": 2387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390635096083443e-05,
+      "loss": 0.8672,
+      "step": 2388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389989904683577e-05,
+      "loss": 0.957,
+      "step": 2389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938934438264767e-05,
+      "loss": 0.9922,
+      "step": 2390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388698529998453e-05,
+      "loss": 1.0391,
+      "step": 2391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388052346758663e-05,
+      "loss": 0.8633,
+      "step": 2392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387405832951055e-05,
+      "loss": 0.8633,
+      "step": 2393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386758988598394e-05,
+      "loss": 0.9219,
+      "step": 2394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386111813723455e-05,
+      "loss": 0.8555,
+      "step": 2395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385464308349027e-05,
+      "loss": 0.8906,
+      "step": 2396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384816472497907e-05,
+      "loss": 0.9336,
+      "step": 2397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384168306192916e-05,
+      "loss": 0.9648,
+      "step": 2398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383519809456862e-05,
+      "loss": 0.8516,
+      "step": 2399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938287098231259e-05,
+      "loss": 0.9141,
+      "step": 2400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382221824782947e-05,
+      "loss": 0.7617,
+      "step": 2401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381572336890784e-05,
+      "loss": 0.9766,
+      "step": 2402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380922518658975e-05,
+      "loss": 1.1484,
+      "step": 2403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380272370110397e-05,
+      "loss": 0.957,
+      "step": 2404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937962189126795e-05,
+      "loss": 0.8906,
+      "step": 2405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937897108215453e-05,
+      "loss": 0.9219,
+      "step": 2406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378319942793064e-05,
+      "loss": 0.9531,
+      "step": 2407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377668473206467e-05,
+      "loss": 0.918,
+      "step": 2408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377016673417683e-05,
+      "loss": 0.8281,
+      "step": 2409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937636454344967e-05,
+      "loss": 0.9883,
+      "step": 2410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375712083325376e-05,
+      "loss": 0.9844,
+      "step": 2411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937505929306779e-05,
+      "loss": 0.7852,
+      "step": 2412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374406172699893e-05,
+      "loss": 0.9922,
+      "step": 2413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373752722244675e-05,
+      "loss": 0.9453,
+      "step": 2414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373098941725153e-05,
+      "loss": 0.9414,
+      "step": 2415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372444831164342e-05,
+      "loss": 1.0078,
+      "step": 2416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371790390585282e-05,
+      "loss": 1.0,
+      "step": 2417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937113562001101e-05,
+      "loss": 0.9258,
+      "step": 2418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370480519464584e-05,
+      "loss": 0.9805,
+      "step": 2419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369825088969072e-05,
+      "loss": 1.0703,
+      "step": 2420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936916932854755e-05,
+      "loss": 0.9922,
+      "step": 2421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368513238223112e-05,
+      "loss": 0.9961,
+      "step": 2422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367856818018858e-05,
+      "loss": 0.9531,
+      "step": 2423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367200067957897e-05,
+      "loss": 0.9375,
+      "step": 2424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366542988063364e-05,
+      "loss": 0.918,
+      "step": 2425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936588557835839e-05,
+      "loss": 0.9492,
+      "step": 2426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936522783886612e-05,
+      "loss": 0.8438,
+      "step": 2427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936456976960972e-05,
+      "loss": 0.8594,
+      "step": 2428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936391137061236e-05,
+      "loss": 0.8672,
+      "step": 2429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363252641897224e-05,
+      "loss": 0.9023,
+      "step": 2430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362593583487503e-05,
+      "loss": 0.9844,
+      "step": 2431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936193419540641e-05,
+      "loss": 0.8438,
+      "step": 2432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361274477677155e-05,
+      "loss": 1.1016,
+      "step": 2433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360614430322973e-05,
+      "loss": 0.8945,
+      "step": 2434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359954053367106e-05,
+      "loss": 0.9102,
+      "step": 2435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359293346832802e-05,
+      "loss": 0.9453,
+      "step": 2436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935863231074333e-05,
+      "loss": 0.7578,
+      "step": 2437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357970945121966e-05,
+      "loss": 0.9609,
+      "step": 2438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357309249991996e-05,
+      "loss": 0.832,
+      "step": 2439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356647225376716e-05,
+      "loss": 0.9062,
+      "step": 2440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355984871299447e-05,
+      "loss": 0.9766,
+      "step": 2441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355322187783502e-05,
+      "loss": 1.0469,
+      "step": 2442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354659174852216e-05,
+      "loss": 0.9844,
+      "step": 2443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935399583252894e-05,
+      "loss": 1.0156,
+      "step": 2444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353332160837026e-05,
+      "loss": 1.0859,
+      "step": 2445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352668159799844e-05,
+      "loss": 0.9453,
+      "step": 2446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935200382944078e-05,
+      "loss": 0.8984,
+      "step": 2447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935133916978322e-05,
+      "loss": 0.9297,
+      "step": 2448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350674180850567e-05,
+      "loss": 1.0156,
+      "step": 2449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935000886266624e-05,
+      "loss": 0.9219,
+      "step": 2450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349343215253666e-05,
+      "loss": 0.9961,
+      "step": 2451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934867723863628e-05,
+      "loss": 0.9453,
+      "step": 2452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348010932837536e-05,
+      "loss": 1.0391,
+      "step": 2453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347344297880892e-05,
+      "loss": 1.0391,
+      "step": 2454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346677333789822e-05,
+      "loss": 0.9453,
+      "step": 2455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346010040587818e-05,
+      "loss": 0.8359,
+      "step": 2456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345342418298365e-05,
+      "loss": 0.9414,
+      "step": 2457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344674466944978e-05,
+      "loss": 0.8984,
+      "step": 2458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344006186551173e-05,
+      "loss": 1.0938,
+      "step": 2459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343337577140484e-05,
+      "loss": 0.9453,
+      "step": 2460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342668638736452e-05,
+      "loss": 1.0391,
+      "step": 2461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341999371362635e-05,
+      "loss": 1.0312,
+      "step": 2462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341329775042594e-05,
+      "loss": 1.0,
+      "step": 2463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934065984979991e-05,
+      "loss": 0.8906,
+      "step": 2464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339989595658167e-05,
+      "loss": 0.9688,
+      "step": 2465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339319012640972e-05,
+      "loss": 0.8555,
+      "step": 2466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338648100771936e-05,
+      "loss": 0.8398,
+      "step": 2467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933797686007468e-05,
+      "loss": 1.0156,
+      "step": 2468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337305290572842e-05,
+      "loss": 0.8828,
+      "step": 2469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336633392290066e-05,
+      "loss": 0.8164,
+      "step": 2470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335961165250015e-05,
+      "loss": 0.8359,
+      "step": 2471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335288609476355e-05,
+      "loss": 0.9336,
+      "step": 2472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334615724992773e-05,
+      "loss": 0.9297,
+      "step": 2473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333942511822955e-05,
+      "loss": 0.9688,
+      "step": 2474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333268969990607e-05,
+      "loss": 0.9688,
+      "step": 2475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332595099519454e-05,
+      "loss": 0.9922,
+      "step": 2476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331920900433213e-05,
+      "loss": 0.9062,
+      "step": 2477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331246372755634e-05,
+      "loss": 0.9922,
+      "step": 2478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933057151651046e-05,
+      "loss": 0.9297,
+      "step": 2479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329896331721455e-05,
+      "loss": 0.8633,
+      "step": 2480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329220818412398e-05,
+      "loss": 0.9688,
+      "step": 2481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932854497660707e-05,
+      "loss": 0.9805,
+      "step": 2482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327868806329268e-05,
+      "loss": 1.0234,
+      "step": 2483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327192307602805e-05,
+      "loss": 0.9727,
+      "step": 2484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93265154804515e-05,
+      "loss": 0.8867,
+      "step": 2485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325838324899183e-05,
+      "loss": 0.9531,
+      "step": 2486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325160840969702e-05,
+      "loss": 1.0078,
+      "step": 2487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932448302868691e-05,
+      "loss": 0.9062,
+      "step": 2488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323804888074672e-05,
+      "loss": 0.9062,
+      "step": 2489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932312641915687e-05,
+      "loss": 0.8906,
+      "step": 2490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932244762195739e-05,
+      "loss": 0.8867,
+      "step": 2491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321768496500133e-05,
+      "loss": 1.1562,
+      "step": 2492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932108904280902e-05,
+      "loss": 1.0469,
+      "step": 2493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320409260907967e-05,
+      "loss": 1.0391,
+      "step": 2494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319729150820915e-05,
+      "loss": 1.0391,
+      "step": 2495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319048712571807e-05,
+      "loss": 0.9023,
+      "step": 2496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318367946184606e-05,
+      "loss": 1.0,
+      "step": 2497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317686851683285e-05,
+      "loss": 0.8828,
+      "step": 2498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931700542909182e-05,
+      "loss": 0.8789,
+      "step": 2499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316323678434213e-05,
+      "loss": 0.8867,
+      "step": 2500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931564159973446e-05,
+      "loss": 0.9492,
+      "step": 2501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9314959193016584e-05,
+      "loss": 0.9336,
+      "step": 2502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314276458304615e-05,
+      "loss": 1.0,
+      "step": 2503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313593395622586e-05,
+      "loss": 0.8672,
+      "step": 2504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312910004994557e-05,
+      "loss": 0.9102,
+      "step": 2505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931222628644459e-05,
+      "loss": 0.8516,
+      "step": 2506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931154223999675e-05,
+      "loss": 1.0156,
+      "step": 2507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931085786567514e-05,
+      "loss": 0.8789,
+      "step": 2508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931017316350384e-05,
+      "loss": 0.9883,
+      "step": 2509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309488133506974e-05,
+      "loss": 0.8945,
+      "step": 2510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308802775708656e-05,
+      "loss": 0.9375,
+      "step": 2511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930811709013302e-05,
+      "loss": 0.9375,
+      "step": 2512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307431076804203e-05,
+      "loss": 0.8672,
+      "step": 2513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306744735746374e-05,
+      "loss": 1.0312,
+      "step": 2514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930605806698369e-05,
+      "loss": 0.9805,
+      "step": 2515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305371070540335e-05,
+      "loss": 0.9453,
+      "step": 2516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304683746440495e-05,
+      "loss": 0.7812,
+      "step": 2517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303996094708376e-05,
+      "loss": 0.9922,
+      "step": 2518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930330811536819e-05,
+      "loss": 0.9375,
+      "step": 2519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302619808444158e-05,
+      "loss": 0.9766,
+      "step": 2520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301931173960524e-05,
+      "loss": 0.9062,
+      "step": 2521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301242211941527e-05,
+      "loss": 0.918,
+      "step": 2522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930055292241143e-05,
+      "loss": 0.8984,
+      "step": 2523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299863305394507e-05,
+      "loss": 0.9961,
+      "step": 2524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929917336091504e-05,
+      "loss": 0.9062,
+      "step": 2525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298483088997317e-05,
+      "loss": 0.9297,
+      "step": 2526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929779248966565e-05,
+      "loss": 0.8359,
+      "step": 2527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297101562944353e-05,
+      "loss": 1.0,
+      "step": 2528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296410308857755e-05,
+      "loss": 0.9531,
+      "step": 2529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295718727430196e-05,
+      "loss": 0.9258,
+      "step": 2530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295026818686026e-05,
+      "loss": 0.8633,
+      "step": 2531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294334582649613e-05,
+      "loss": 1.0156,
+      "step": 2532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929364201934533e-05,
+      "loss": 1.0078,
+      "step": 2533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929294912879756e-05,
+      "loss": 1.0469,
+      "step": 2534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292255911030703e-05,
+      "loss": 0.9336,
+      "step": 2535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929156236606917e-05,
+      "loss": 0.9648,
+      "step": 2536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929086849393738e-05,
+      "loss": 0.9258,
+      "step": 2537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290174294659763e-05,
+      "loss": 0.9453,
+      "step": 2538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289479768260765e-05,
+      "loss": 0.8125,
+      "step": 2539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288784914764843e-05,
+      "loss": 0.9531,
+      "step": 2540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288089734196462e-05,
+      "loss": 0.8555,
+      "step": 2541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287394226580102e-05,
+      "loss": 1.0234,
+      "step": 2542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286698391940248e-05,
+      "loss": 0.9531,
+      "step": 2543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286002230301408e-05,
+      "loss": 0.9336,
+      "step": 2544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285305741688093e-05,
+      "loss": 0.9062,
+      "step": 2545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284608926124825e-05,
+      "loss": 1.0781,
+      "step": 2546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283911783636136e-05,
+      "loss": 0.8438,
+      "step": 2547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283214314246588e-05,
+      "loss": 0.9609,
+      "step": 2548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282516517980726e-05,
+      "loss": 0.9492,
+      "step": 2549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281818394863126e-05,
+      "loss": 1.0938,
+      "step": 2550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281119944918368e-05,
+      "loss": 1.0625,
+      "step": 2551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280421168171046e-05,
+      "loss": 0.8789,
+      "step": 2552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279722064645766e-05,
+      "loss": 0.918,
+      "step": 2553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279022634367148e-05,
+      "loss": 1.0,
+      "step": 2554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278322877359815e-05,
+      "loss": 1.1719,
+      "step": 2555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277622793648404e-05,
+      "loss": 1.0234,
+      "step": 2556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276922383257576e-05,
+      "loss": 0.9922,
+      "step": 2557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927622164621198e-05,
+      "loss": 0.9531,
+      "step": 2558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92755205825363e-05,
+      "loss": 0.9844,
+      "step": 2559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274819192255224e-05,
+      "loss": 0.9531,
+      "step": 2560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274117475393436e-05,
+      "loss": 0.9688,
+      "step": 2561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273415431975658e-05,
+      "loss": 0.9141,
+      "step": 2562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272713062026605e-05,
+      "loss": 0.9297,
+      "step": 2563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272010365571002e-05,
+      "loss": 0.8828,
+      "step": 2564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271307342633605e-05,
+      "loss": 0.8359,
+      "step": 2565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270603993239158e-05,
+      "loss": 1.0547,
+      "step": 2566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926990031741243e-05,
+      "loss": 0.9805,
+      "step": 2567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269196315178203e-05,
+      "loss": 0.8633,
+      "step": 2568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268491986561258e-05,
+      "loss": 0.8984,
+      "step": 2569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267787331586402e-05,
+      "loss": 1.1406,
+      "step": 2570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267082350278445e-05,
+      "loss": 0.8906,
+      "step": 2571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266377042662206e-05,
+      "loss": 0.8359,
+      "step": 2572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926567140876253e-05,
+      "loss": 0.9492,
+      "step": 2573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264965448604255e-05,
+      "loss": 1.0234,
+      "step": 2574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926425916221224e-05,
+      "loss": 1.0391,
+      "step": 2575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263552549611356e-05,
+      "loss": 0.9219,
+      "step": 2576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262845610826484e-05,
+      "loss": 0.8984,
+      "step": 2577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262138345882516e-05,
+      "loss": 0.9023,
+      "step": 2578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926143075480436e-05,
+      "loss": 0.9805,
+      "step": 2579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260722837616926e-05,
+      "loss": 1.0,
+      "step": 2580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926001459434514e-05,
+      "loss": 0.8789,
+      "step": 2581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259306025013944e-05,
+      "loss": 0.832,
+      "step": 2582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258597129648287e-05,
+      "loss": 0.9375,
+      "step": 2583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925788790827313e-05,
+      "loss": 0.8789,
+      "step": 2584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257178360913443e-05,
+      "loss": 0.8984,
+      "step": 2585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256468487594215e-05,
+      "loss": 1.0156,
+      "step": 2586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255758288340442e-05,
+      "loss": 1.0078,
+      "step": 2587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255047763177122e-05,
+      "loss": 0.8438,
+      "step": 2588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254336912129284e-05,
+      "loss": 0.9297,
+      "step": 2589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253625735221955e-05,
+      "loss": 0.8633,
+      "step": 2590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252914232480178e-05,
+      "loss": 0.9609,
+      "step": 2591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252202403929003e-05,
+      "loss": 0.8633,
+      "step": 2592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251490249593497e-05,
+      "loss": 0.9453,
+      "step": 2593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250777769498735e-05,
+      "loss": 0.9258,
+      "step": 2594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250064963669803e-05,
+      "loss": 1.0625,
+      "step": 2595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249351832131803e-05,
+      "loss": 0.8164,
+      "step": 2596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924863837490984e-05,
+      "loss": 0.9219,
+      "step": 2597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247924592029045e-05,
+      "loss": 0.9062,
+      "step": 2598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247210483514543e-05,
+      "loss": 0.9648,
+      "step": 2599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246496049391486e-05,
+      "loss": 0.9141,
+      "step": 2600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245781289685023e-05,
+      "loss": 0.9609,
+      "step": 2601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245066204420326e-05,
+      "loss": 0.9258,
+      "step": 2602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244350793622575e-05,
+      "loss": 0.9961,
+      "step": 2603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924363505731696e-05,
+      "loss": 0.9844,
+      "step": 2604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924291899552868e-05,
+      "loss": 0.9023,
+      "step": 2605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924220260828295e-05,
+      "loss": 0.8867,
+      "step": 2606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241485895605e-05,
+      "loss": 0.9922,
+      "step": 2607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240768857520062e-05,
+      "loss": 0.9766,
+      "step": 2608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924005149405338e-05,
+      "loss": 0.8789,
+      "step": 2609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239333805230226e-05,
+      "loss": 0.9219,
+      "step": 2610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238615791075857e-05,
+      "loss": 0.9297,
+      "step": 2611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237897451615565e-05,
+      "loss": 0.9297,
+      "step": 2612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923717878687464e-05,
+      "loss": 0.8555,
+      "step": 2613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236459796878387e-05,
+      "loss": 0.9609,
+      "step": 2614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235740481652125e-05,
+      "loss": 0.957,
+      "step": 2615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923502084122118e-05,
+      "loss": 0.918,
+      "step": 2616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234300875610893e-05,
+      "loss": 0.8906,
+      "step": 2617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923358058484661e-05,
+      "loss": 0.9062,
+      "step": 2618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232859968953702e-05,
+      "loss": 0.9219,
+      "step": 2619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923213902795754e-05,
+      "loss": 1.0938,
+      "step": 2620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231417761883507e-05,
+      "loss": 0.9414,
+      "step": 2621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230696170756998e-05,
+      "loss": 0.8594,
+      "step": 2622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922997425460343e-05,
+      "loss": 0.9141,
+      "step": 2623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229252013448218e-05,
+      "loss": 0.7969,
+      "step": 2624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228529447316785e-05,
+      "loss": 0.8672,
+      "step": 2625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922780655623459e-05,
+      "loss": 0.9297,
+      "step": 2626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227083340227075e-05,
+      "loss": 0.8516,
+      "step": 2627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226359799319707e-05,
+      "loss": 0.8633,
+      "step": 2628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225635933537966e-05,
+      "loss": 1.0781,
+      "step": 2629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922491174290734e-05,
+      "loss": 0.9531,
+      "step": 2630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922418722745333e-05,
+      "loss": 0.8906,
+      "step": 2631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223462387201443e-05,
+      "loss": 0.9648,
+      "step": 2632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222737222177206e-05,
+      "loss": 0.9453,
+      "step": 2633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922201173240615e-05,
+      "loss": 0.8945,
+      "step": 2634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221285917913824e-05,
+      "loss": 0.9023,
+      "step": 2635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922055977872578e-05,
+      "loss": 1.0156,
+      "step": 2636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219833314867594e-05,
+      "loss": 0.9297,
+      "step": 2637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921910652636484e-05,
+      "loss": 0.9844,
+      "step": 2638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921837941324311e-05,
+      "loss": 1.0078,
+      "step": 2639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217651975528008e-05,
+      "loss": 0.9531,
+      "step": 2640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216924213245147e-05,
+      "loss": 0.9492,
+      "step": 2641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921619612642016e-05,
+      "loss": 0.9336,
+      "step": 2642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215467715078667e-05,
+      "loss": 1.0234,
+      "step": 2643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921473897924633e-05,
+      "loss": 0.9102,
+      "step": 2644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214009918948813e-05,
+      "loss": 0.9453,
+      "step": 2645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213280534211774e-05,
+      "loss": 1.0781,
+      "step": 2646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212550825060905e-05,
+      "loss": 0.9688,
+      "step": 2647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211820791521897e-05,
+      "loss": 0.9453,
+      "step": 2648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211090433620454e-05,
+      "loss": 0.8867,
+      "step": 2649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210359751382297e-05,
+      "loss": 0.8555,
+      "step": 2650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920962874483315e-05,
+      "loss": 1.0391,
+      "step": 2651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208897413998758e-05,
+      "loss": 0.8867,
+      "step": 2652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920816575890487e-05,
+      "loss": 0.8711,
+      "step": 2653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207433779577244e-05,
+      "loss": 0.8828,
+      "step": 2654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206701476041664e-05,
+      "loss": 0.7969,
+      "step": 2655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205968848323905e-05,
+      "loss": 0.9688,
+      "step": 2656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920523589644977e-05,
+      "loss": 0.8281,
+      "step": 2657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204502620445065e-05,
+      "loss": 0.9297,
+      "step": 2658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203769020335614e-05,
+      "loss": 1.0781,
+      "step": 2659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203035096147243e-05,
+      "loss": 0.9258,
+      "step": 2660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202300847905802e-05,
+      "loss": 0.8594,
+      "step": 2661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201566275637135e-05,
+      "loss": 0.8633,
+      "step": 2662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920083137936711e-05,
+      "loss": 0.8164,
+      "step": 2663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200096159121612e-05,
+      "loss": 0.9531,
+      "step": 2664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199360614926518e-05,
+      "loss": 1.0,
+      "step": 2665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919862474680774e-05,
+      "loss": 0.8711,
+      "step": 2666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197888554791176e-05,
+      "loss": 1.0,
+      "step": 2667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197152038902755e-05,
+      "loss": 0.8594,
+      "step": 2668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196415199168412e-05,
+      "loss": 0.9414,
+      "step": 2669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919567803561409e-05,
+      "loss": 0.9688,
+      "step": 2670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194940548265746e-05,
+      "loss": 0.9219,
+      "step": 2671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194202737149347e-05,
+      "loss": 0.8555,
+      "step": 2672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193464602290878e-05,
+      "loss": 0.8203,
+      "step": 2673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192726143716322e-05,
+      "loss": 0.957,
+      "step": 2674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9191987361451686e-05,
+      "loss": 1.0938,
+      "step": 2675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191248255522984e-05,
+      "loss": 0.8672,
+      "step": 2676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190508825956237e-05,
+      "loss": 0.9141,
+      "step": 2677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189769072777483e-05,
+      "loss": 1.0312,
+      "step": 2678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189028996012777e-05,
+      "loss": 1.0156,
+      "step": 2679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188288595688166e-05,
+      "loss": 0.9531,
+      "step": 2680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918754787182973e-05,
+      "loss": 0.9648,
+      "step": 2681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186806824463548e-05,
+      "loss": 0.9805,
+      "step": 2682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918606545361571e-05,
+      "loss": 1.0234,
+      "step": 2683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918532375931233e-05,
+      "loss": 0.9375,
+      "step": 2684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184581741579514e-05,
+      "loss": 0.9375,
+      "step": 2685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183839400443395e-05,
+      "loss": 1.0156,
+      "step": 2686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918309673593011e-05,
+      "loss": 0.9492,
+      "step": 2687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918235374806581e-05,
+      "loss": 0.832,
+      "step": 2688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918161043687666e-05,
+      "loss": 0.9141,
+      "step": 2689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918086680238883e-05,
+      "loss": 0.9609,
+      "step": 2690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180122844628505e-05,
+      "loss": 0.9727,
+      "step": 2691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179378563621877e-05,
+      "loss": 0.8789,
+      "step": 2692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178633959395158e-05,
+      "loss": 1.0156,
+      "step": 2693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177889031974567e-05,
+      "loss": 1.0312,
+      "step": 2694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177143781386333e-05,
+      "loss": 0.9688,
+      "step": 2695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176398207656693e-05,
+      "loss": 0.8828,
+      "step": 2696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175652310811905e-05,
+      "loss": 0.957,
+      "step": 2697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174906090878234e-05,
+      "loss": 0.8984,
+      "step": 2698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174159547881955e-05,
+      "loss": 0.8672,
+      "step": 2699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917341268184935e-05,
+      "loss": 1.0234,
+      "step": 2700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172665492806724e-05,
+      "loss": 0.9453,
+      "step": 2701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917191798078038e-05,
+      "loss": 0.9414,
+      "step": 2702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171170145796645e-05,
+      "loss": 0.9453,
+      "step": 2703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170421987881847e-05,
+      "loss": 1.0234,
+      "step": 2704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916967350706233e-05,
+      "loss": 0.8398,
+      "step": 2705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168924703364454e-05,
+      "loss": 0.9609,
+      "step": 2706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168175576814582e-05,
+      "loss": 0.9805,
+      "step": 2707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167426127439092e-05,
+      "loss": 1.0,
+      "step": 2708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166676355264376e-05,
+      "loss": 0.8125,
+      "step": 2709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916592626031683e-05,
+      "loss": 0.8711,
+      "step": 2710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165175842622866e-05,
+      "loss": 0.9141,
+      "step": 2711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164425102208915e-05,
+      "loss": 0.9062,
+      "step": 2712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163674039101402e-05,
+      "loss": 1.1406,
+      "step": 2713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162922653326778e-05,
+      "loss": 0.9883,
+      "step": 2714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91621709449115e-05,
+      "loss": 1.0703,
+      "step": 2715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161418913882036e-05,
+      "loss": 0.9844,
+      "step": 2716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916066656026487e-05,
+      "loss": 0.8203,
+      "step": 2717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159913884086486e-05,
+      "loss": 0.9141,
+      "step": 2718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159160885373397e-05,
+      "loss": 0.8516,
+      "step": 2719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158407564152107e-05,
+      "loss": 0.9531,
+      "step": 2720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157653920449148e-05,
+      "loss": 1.1172,
+      "step": 2721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156899954291052e-05,
+      "loss": 0.8906,
+      "step": 2722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156145665704373e-05,
+      "loss": 1.0,
+      "step": 2723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155391054715667e-05,
+      "loss": 1.0625,
+      "step": 2724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915463612135151e-05,
+      "loss": 0.9297,
+      "step": 2725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153880865638475e-05,
+      "loss": 0.9102,
+      "step": 2726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153125287603162e-05,
+      "loss": 0.8281,
+      "step": 2727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915236938727218e-05,
+      "loss": 0.8594,
+      "step": 2728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151613164672136e-05,
+      "loss": 0.957,
+      "step": 2729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150856619829664e-05,
+      "loss": 0.9141,
+      "step": 2730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91500997527714e-05,
+      "loss": 0.9258,
+      "step": 2731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149342563523998e-05,
+      "loss": 0.9414,
+      "step": 2732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148585052114116e-05,
+      "loss": 1.0078,
+      "step": 2733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914782721856843e-05,
+      "loss": 0.9375,
+      "step": 2734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147069062913623e-05,
+      "loss": 1.0156,
+      "step": 2735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146310585176392e-05,
+      "loss": 0.8594,
+      "step": 2736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145551785383444e-05,
+      "loss": 0.9648,
+      "step": 2737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91447926635615e-05,
+      "loss": 1.0,
+      "step": 2738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914403321973728e-05,
+      "loss": 0.8203,
+      "step": 2739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143273453937533e-05,
+      "loss": 0.9336,
+      "step": 2740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142513366189013e-05,
+      "loss": 0.9766,
+      "step": 2741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141752956518484e-05,
+      "loss": 0.9453,
+      "step": 2742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140992224952713e-05,
+      "loss": 0.9258,
+      "step": 2743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140231171518495e-05,
+      "loss": 0.8984,
+      "step": 2744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139469796242626e-05,
+      "loss": 0.8438,
+      "step": 2745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913870809915191e-05,
+      "loss": 1.0312,
+      "step": 2746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137946080273175e-05,
+      "loss": 1.0469,
+      "step": 2747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137183739633246e-05,
+      "loss": 1.0078,
+      "step": 2748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136421077258976e-05,
+      "loss": 0.8711,
+      "step": 2749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135658093177208e-05,
+      "loss": 0.9609,
+      "step": 2750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134894787414813e-05,
+      "loss": 0.8086,
+      "step": 2751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913413115999867e-05,
+      "loss": 0.8281,
+      "step": 2752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133367210955668e-05,
+      "loss": 0.9648,
+      "step": 2753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132602940312704e-05,
+      "loss": 1.0312,
+      "step": 2754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131838348096686e-05,
+      "loss": 0.9414,
+      "step": 2755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131073434334545e-05,
+      "loss": 0.9609,
+      "step": 2756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130308199053207e-05,
+      "loss": 1.0391,
+      "step": 2757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912954264227962e-05,
+      "loss": 0.9531,
+      "step": 2758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912877676404074e-05,
+      "loss": 0.8633,
+      "step": 2759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912801056436354e-05,
+      "loss": 0.9023,
+      "step": 2760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912724404327499e-05,
+      "loss": 1.0703,
+      "step": 2761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126477200802084e-05,
+      "loss": 0.9023,
+      "step": 2762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912571003697183e-05,
+      "loss": 0.8906,
+      "step": 2763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912494255181123e-05,
+      "loss": 0.9141,
+      "step": 2764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124174745347317e-05,
+      "loss": 0.9453,
+      "step": 2765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912340661760712e-05,
+      "loss": 0.8945,
+      "step": 2766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122638168617693e-05,
+      "loss": 0.9609,
+      "step": 2767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912186939840609e-05,
+      "loss": 1.0391,
+      "step": 2768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912110030699938e-05,
+      "loss": 0.9219,
+      "step": 2769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120330894424647e-05,
+      "loss": 0.9922,
+      "step": 2770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119561160708983e-05,
+      "loss": 0.9961,
+      "step": 2771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118791105879487e-05,
+      "loss": 0.8711,
+      "step": 2772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118020729963277e-05,
+      "loss": 0.8945,
+      "step": 2773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911725003298748e-05,
+      "loss": 0.9414,
+      "step": 2774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911647901497923e-05,
+      "loss": 0.7852,
+      "step": 2775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115707675965682e-05,
+      "loss": 0.8672,
+      "step": 2776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911493601597399e-05,
+      "loss": 0.8438,
+      "step": 2777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114164035031323e-05,
+      "loss": 0.8906,
+      "step": 2778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113391733164876e-05,
+      "loss": 0.9297,
+      "step": 2779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911261911040183e-05,
+      "loss": 1.0078,
+      "step": 2780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111846166769398e-05,
+      "loss": 0.9102,
+      "step": 2781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911107290229479e-05,
+      "loss": 1.0781,
+      "step": 2782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911029931700524e-05,
+      "loss": 0.9297,
+      "step": 2783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109525410927988e-05,
+      "loss": 0.9062,
+      "step": 2784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108751184090277e-05,
+      "loss": 1.1094,
+      "step": 2785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107976636519373e-05,
+      "loss": 1.0391,
+      "step": 2786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910720176824255e-05,
+      "loss": 0.9297,
+      "step": 2787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910642657928709e-05,
+      "loss": 0.9453,
+      "step": 2788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105651069680293e-05,
+      "loss": 0.8711,
+      "step": 2789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910487523944946e-05,
+      "loss": 0.9492,
+      "step": 2790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104099088621916e-05,
+      "loss": 0.8203,
+      "step": 2791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910332261722498e-05,
+      "loss": 0.8828,
+      "step": 2792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102545825286005e-05,
+      "loss": 1.0,
+      "step": 2793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101768712832337e-05,
+      "loss": 1.0234,
+      "step": 2794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100991279891335e-05,
+      "loss": 0.9141,
+      "step": 2795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910021352649038e-05,
+      "loss": 0.9844,
+      "step": 2796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099435452656858e-05,
+      "loss": 0.957,
+      "step": 2797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909865705841816e-05,
+      "loss": 0.8789,
+      "step": 2798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90978783438017e-05,
+      "loss": 0.9062,
+      "step": 2799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90970993088349e-05,
+      "loss": 1.0078,
+      "step": 2800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.8672,
+      "step": 2801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909554027796e-05,
+      "loss": 1.0625,
+      "step": 2802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90947602821068e-05,
+      "loss": 0.9609,
+      "step": 2803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093979966013046e-05,
+      "loss": 0.9453,
+      "step": 2804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093199329706215e-05,
+      "loss": 0.9492,
+      "step": 2805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092418373213795e-05,
+      "loss": 0.8828,
+      "step": 2806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909163709656329e-05,
+      "loss": 0.9023,
+      "step": 2807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090855499782206e-05,
+      "loss": 1.0625,
+      "step": 2808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909007358289806e-05,
+      "loss": 0.9414,
+      "step": 2809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908929134593839e-05,
+      "loss": 0.9375,
+      "step": 2810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088508788930736e-05,
+      "loss": 0.8867,
+      "step": 2811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087725911902656e-05,
+      "loss": 1.0859,
+      "step": 2812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086942714881713e-05,
+      "loss": 0.9531,
+      "step": 2813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086159197895487e-05,
+      "loss": 0.8516,
+      "step": 2814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908537536097157e-05,
+      "loss": 0.9453,
+      "step": 2815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084591204137558e-05,
+      "loss": 0.8438,
+      "step": 2816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083806727421065e-05,
+      "loss": 0.9883,
+      "step": 2817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083021930849708e-05,
+      "loss": 0.9961,
+      "step": 2818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082236814451124e-05,
+      "loss": 0.9922,
+      "step": 2819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081451378252964e-05,
+      "loss": 1.0312,
+      "step": 2820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080665622282874e-05,
+      "loss": 0.9102,
+      "step": 2821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079879546568527e-05,
+      "loss": 0.8125,
+      "step": 2822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079093151137606e-05,
+      "loss": 0.9531,
+      "step": 2823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078306436017792e-05,
+      "loss": 0.9492,
+      "step": 2824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077519401236797e-05,
+      "loss": 0.7891,
+      "step": 2825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076732046822325e-05,
+      "loss": 0.9844,
+      "step": 2826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075944372802106e-05,
+      "loss": 0.9219,
+      "step": 2827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075156379203867e-05,
+      "loss": 0.9805,
+      "step": 2828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074368066055364e-05,
+      "loss": 0.9453,
+      "step": 2829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907357943338435e-05,
+      "loss": 0.9219,
+      "step": 2830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907279048121859e-05,
+      "loss": 0.9258,
+      "step": 2831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072001209585878e-05,
+      "loss": 0.8906,
+      "step": 2832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907121161851399e-05,
+      "loss": 0.8555,
+      "step": 2833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070421708030735e-05,
+      "loss": 0.8867,
+      "step": 2834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906963147816393e-05,
+      "loss": 0.7969,
+      "step": 2835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068840928941395e-05,
+      "loss": 0.9062,
+      "step": 2836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068050060390967e-05,
+      "loss": 0.957,
+      "step": 2837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90672588725405e-05,
+      "loss": 0.957,
+      "step": 2838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066467365417844e-05,
+      "loss": 0.8984,
+      "step": 2839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065675539050878e-05,
+      "loss": 0.8008,
+      "step": 2840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064883393467473e-05,
+      "loss": 0.9297,
+      "step": 2841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906409092869553e-05,
+      "loss": 0.9883,
+      "step": 2842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906329814476295e-05,
+      "loss": 0.9141,
+      "step": 2843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062505041697653e-05,
+      "loss": 0.9297,
+      "step": 2844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061711619527556e-05,
+      "loss": 0.8398,
+      "step": 2845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90609178782806e-05,
+      "loss": 0.8398,
+      "step": 2846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060123817984742e-05,
+      "loss": 0.8984,
+      "step": 2847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059329438667933e-05,
+      "loss": 0.8516,
+      "step": 2848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058534740358147e-05,
+      "loss": 0.8398,
+      "step": 2849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057739723083366e-05,
+      "loss": 1.0,
+      "step": 2850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056944386871583e-05,
+      "loss": 0.9219,
+      "step": 2851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056148731750804e-05,
+      "loss": 0.918,
+      "step": 2852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055352757749047e-05,
+      "loss": 0.8906,
+      "step": 2853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905455646489434e-05,
+      "loss": 0.8398,
+      "step": 2854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053759853214714e-05,
+      "loss": 0.9297,
+      "step": 2855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905296292273823e-05,
+      "loss": 0.9922,
+      "step": 2856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052165673492943e-05,
+      "loss": 1.0391,
+      "step": 2857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051368105506923e-05,
+      "loss": 0.8906,
+      "step": 2858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905057021880826e-05,
+      "loss": 0.9258,
+      "step": 2859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049772013425044e-05,
+      "loss": 0.9688,
+      "step": 2860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048973489385384e-05,
+      "loss": 0.9336,
+      "step": 2861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048174646717394e-05,
+      "loss": 0.8672,
+      "step": 2862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047375485449203e-05,
+      "loss": 0.9844,
+      "step": 2863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046576005608957e-05,
+      "loss": 0.8984,
+      "step": 2864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90457762072248e-05,
+      "loss": 0.8516,
+      "step": 2865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044976090324892e-05,
+      "loss": 0.9883,
+      "step": 2866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044175654937413e-05,
+      "loss": 0.957,
+      "step": 2867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043374901090546e-05,
+      "loss": 1.0469,
+      "step": 2868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042573828812485e-05,
+      "loss": 0.918,
+      "step": 2869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904177243813144e-05,
+      "loss": 1.0547,
+      "step": 2870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040970729075624e-05,
+      "loss": 0.9961,
+      "step": 2871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904016870167327e-05,
+      "loss": 0.9766,
+      "step": 2872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903936635595262e-05,
+      "loss": 0.9922,
+      "step": 2873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903856369194192e-05,
+      "loss": 0.9609,
+      "step": 2874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903776070966944e-05,
+      "loss": 1.0078,
+      "step": 2875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036957409163453e-05,
+      "loss": 1.0312,
+      "step": 2876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036153790452238e-05,
+      "loss": 1.0859,
+      "step": 2877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035349853564098e-05,
+      "loss": 0.8711,
+      "step": 2878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903454559852734e-05,
+      "loss": 0.918,
+      "step": 2879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033741025370283e-05,
+      "loss": 0.9375,
+      "step": 2880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032936134121254e-05,
+      "loss": 1.0078,
+      "step": 2881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032130924808596e-05,
+      "loss": 0.8164,
+      "step": 2882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031325397460666e-05,
+      "loss": 0.832,
+      "step": 2883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030519552105825e-05,
+      "loss": 0.9922,
+      "step": 2884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029713388772443e-05,
+      "loss": 1.0234,
+      "step": 2885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028906907488914e-05,
+      "loss": 1.0391,
+      "step": 2886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902810010828363e-05,
+      "loss": 0.9297,
+      "step": 2887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027292991185003e-05,
+      "loss": 0.8867,
+      "step": 2888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902648555622145e-05,
+      "loss": 0.9297,
+      "step": 2889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025677803421406e-05,
+      "loss": 0.8984,
+      "step": 2890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024869732813312e-05,
+      "loss": 0.9141,
+      "step": 2891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902406134442562e-05,
+      "loss": 0.8555,
+      "step": 2892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023252638286793e-05,
+      "loss": 0.9258,
+      "step": 2893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902244361442531e-05,
+      "loss": 0.8438,
+      "step": 2894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021634272869655e-05,
+      "loss": 0.832,
+      "step": 2895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902082461364833e-05,
+      "loss": 0.9844,
+      "step": 2896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902001463678984e-05,
+      "loss": 0.9648,
+      "step": 2897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901920434232271e-05,
+      "loss": 0.8477,
+      "step": 2898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901839373027547e-05,
+      "loss": 1.1172,
+      "step": 2899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901758280067666e-05,
+      "loss": 0.9102,
+      "step": 2900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016771553554836e-05,
+      "loss": 1.0078,
+      "step": 2901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015959988938566e-05,
+      "loss": 1.0156,
+      "step": 2902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015148106856423e-05,
+      "loss": 0.918,
+      "step": 2903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014335907336998e-05,
+      "loss": 0.9062,
+      "step": 2904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013523390408886e-05,
+      "loss": 0.9531,
+      "step": 2905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012710556100697e-05,
+      "loss": 0.9766,
+      "step": 2906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011897404441056e-05,
+      "loss": 0.9453,
+      "step": 2907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901108393545859e-05,
+      "loss": 0.9766,
+      "step": 2908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010270149181948e-05,
+      "loss": 1.0469,
+      "step": 2909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009456045639783e-05,
+      "loss": 0.9688,
+      "step": 2910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900864162486076e-05,
+      "loss": 0.8359,
+      "step": 2911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007826886873554e-05,
+      "loss": 0.9414,
+      "step": 2912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007011831706858e-05,
+      "loss": 1.0078,
+      "step": 2913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006196459389366e-05,
+      "loss": 0.8867,
+      "step": 2914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005380769949793e-05,
+      "loss": 0.8711,
+      "step": 2915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004564763416856e-05,
+      "loss": 0.8594,
+      "step": 2916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003748439819293e-05,
+      "loss": 0.9141,
+      "step": 2917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002931799185847e-05,
+      "loss": 0.9336,
+      "step": 2918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900211484154527e-05,
+      "loss": 0.8242,
+      "step": 2919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900129756692633e-05,
+      "loss": 0.9688,
+      "step": 2920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000479975357807e-05,
+      "loss": 0.8867,
+      "step": 2921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999662066868484e-05,
+      "loss": 1.0859,
+      "step": 2922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998843841487164e-05,
+      "loss": 0.8672,
+      "step": 2923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899802529924266e-05,
+      "loss": 0.8516,
+      "step": 2924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997206440163793e-05,
+      "loss": 0.9453,
+      "step": 2925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996387264279393e-05,
+      "loss": 0.9336,
+      "step": 2926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995567771618312e-05,
+      "loss": 0.9805,
+      "step": 2927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994747962209396e-05,
+      "loss": 0.9375,
+      "step": 2928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993927836081514e-05,
+      "loss": 0.9688,
+      "step": 2929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993107393263552e-05,
+      "loss": 1.0234,
+      "step": 2930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992286633784393e-05,
+      "loss": 0.8711,
+      "step": 2931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991465557672937e-05,
+      "loss": 0.9219,
+      "step": 2932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990644164958092e-05,
+      "loss": 0.8711,
+      "step": 2933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989822455668786e-05,
+      "loss": 1.0547,
+      "step": 2934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989000429833954e-05,
+      "loss": 0.9141,
+      "step": 2935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988178087482537e-05,
+      "loss": 0.8594,
+      "step": 2936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898735542864349e-05,
+      "loss": 0.8672,
+      "step": 2937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986532453345785e-05,
+      "loss": 1.0234,
+      "step": 2938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985709161618394e-05,
+      "loss": 1.0078,
+      "step": 2939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984885553490312e-05,
+      "loss": 0.9219,
+      "step": 2940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984061628990536e-05,
+      "loss": 0.9531,
+      "step": 2941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983237388148077e-05,
+      "loss": 1.0234,
+      "step": 2942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982412830991962e-05,
+      "loss": 0.957,
+      "step": 2943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898158795755122e-05,
+      "loss": 1.0391,
+      "step": 2944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89807627678549e-05,
+      "loss": 0.8906,
+      "step": 2945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979937261932056e-05,
+      "loss": 0.9727,
+      "step": 2946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979111439811757e-05,
+      "loss": 0.8555,
+      "step": 2947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978285301523082e-05,
+      "loss": 0.8359,
+      "step": 2948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977458847095117e-05,
+      "loss": 0.9062,
+      "step": 2949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976632076556964e-05,
+      "loss": 0.9141,
+      "step": 2950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975804989937736e-05,
+      "loss": 0.8516,
+      "step": 2951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974977587266558e-05,
+      "loss": 0.9531,
+      "step": 2952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897414986857256e-05,
+      "loss": 0.9023,
+      "step": 2953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897332183388489e-05,
+      "loss": 0.9688,
+      "step": 2954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89724934832327e-05,
+      "loss": 0.9648,
+      "step": 2955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971664816645164e-05,
+      "loss": 0.8555,
+      "step": 2956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970835834151457e-05,
+      "loss": 0.9219,
+      "step": 2957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897000653578077e-05,
+      "loss": 0.9922,
+      "step": 2958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89691769215623e-05,
+      "loss": 0.8398,
+      "step": 2959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968346991525267e-05,
+      "loss": 0.8516,
+      "step": 2960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967516745698887e-05,
+      "loss": 1.0312,
+      "step": 2961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966686184112397e-05,
+      "loss": 1.0703,
+      "step": 2962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965855306795043e-05,
+      "loss": 1.0312,
+      "step": 2963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896502411377608e-05,
+      "loss": 0.7969,
+      "step": 2964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964192605084774e-05,
+      "loss": 0.9023,
+      "step": 2965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963360780750406e-05,
+      "loss": 0.9336,
+      "step": 2966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962528640802268e-05,
+      "loss": 0.9883,
+      "step": 2967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961696185269657e-05,
+      "loss": 1.0625,
+      "step": 2968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960863414181885e-05,
+      "loss": 0.9336,
+      "step": 2969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960030327568278e-05,
+      "loss": 0.7812,
+      "step": 2970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895919692545817e-05,
+      "loss": 0.9492,
+      "step": 2971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958363207880906e-05,
+      "loss": 0.9922,
+      "step": 2972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957529174865837e-05,
+      "loss": 0.9531,
+      "step": 2973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895669482644234e-05,
+      "loss": 0.9023,
+      "step": 2974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895586016263979e-05,
+      "loss": 0.9727,
+      "step": 2975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895502518348757e-05,
+      "loss": 0.8555,
+      "step": 2976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895418988901509e-05,
+      "loss": 1.0703,
+      "step": 2977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895335427925176e-05,
+      "loss": 0.9453,
+      "step": 2978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952518354227e-05,
+      "loss": 0.8906,
+      "step": 2979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895168211397025e-05,
+      "loss": 1.0078,
+      "step": 2980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950845558510945e-05,
+      "loss": 1.0312,
+      "step": 2981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895000868787855e-05,
+      "loss": 0.957,
+      "step": 2982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949171502102532e-05,
+      "loss": 1.0156,
+      "step": 2983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894833400121237e-05,
+      "loss": 0.8828,
+      "step": 2984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947496185237547e-05,
+      "loss": 0.8633,
+      "step": 2985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946658054207572e-05,
+      "loss": 0.8828,
+      "step": 2986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945819608151953e-05,
+      "loss": 0.8438,
+      "step": 2987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894498084710021e-05,
+      "loss": 1.0,
+      "step": 2988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894414177108188e-05,
+      "loss": 0.9922,
+      "step": 2989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894330238012651e-05,
+      "loss": 1.1172,
+      "step": 2990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942462674263657e-05,
+      "loss": 0.9258,
+      "step": 2991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941622653522885e-05,
+      "loss": 1.1094,
+      "step": 2992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940782317933773e-05,
+      "loss": 0.8711,
+      "step": 2993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893994166752591e-05,
+      "loss": 0.7891,
+      "step": 2994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89391007023289e-05,
+      "loss": 1.0469,
+      "step": 2995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893825942237235e-05,
+      "loss": 0.9375,
+      "step": 2996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893741782768589e-05,
+      "loss": 0.832,
+      "step": 2997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936575918299144e-05,
+      "loss": 0.8594,
+      "step": 2998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935733694241762e-05,
+      "loss": 0.8438,
+      "step": 2999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934891155543404e-05,
+      "loss": 0.9688,
+      "step": 3000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893404830223373e-05,
+      "loss": 0.9844,
+      "step": 3001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893320513434242e-05,
+      "loss": 0.832,
+      "step": 3002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932361651899164e-05,
+      "loss": 1.0156,
+      "step": 3003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931517854933664e-05,
+      "loss": 0.9219,
+      "step": 3004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930673743475632e-05,
+      "loss": 0.9062,
+      "step": 3005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929829317554785e-05,
+      "loss": 0.9375,
+      "step": 3006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928984577200865e-05,
+      "loss": 0.7578,
+      "step": 3007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928139522443606e-05,
+      "loss": 0.9414,
+      "step": 3008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927294153312775e-05,
+      "loss": 0.9414,
+      "step": 3009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926448469838127e-05,
+      "loss": 0.8984,
+      "step": 3010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892560247204945e-05,
+      "loss": 0.8633,
+      "step": 3011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892475615997653e-05,
+      "loss": 1.0078,
+      "step": 3012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923909533649163e-05,
+      "loss": 0.9609,
+      "step": 3013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923062593097167e-05,
+      "loss": 0.8398,
+      "step": 3014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922215338350358e-05,
+      "loss": 0.9219,
+      "step": 3015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921367769438567e-05,
+      "loss": 0.8555,
+      "step": 3016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920519886391642e-05,
+      "loss": 0.8164,
+      "step": 3017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919671689239446e-05,
+      "loss": 0.9297,
+      "step": 3018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918823178011833e-05,
+      "loss": 0.8789,
+      "step": 3019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8917974352738685e-05,
+      "loss": 0.9219,
+      "step": 3020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891712521344989e-05,
+      "loss": 0.9688,
+      "step": 3021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916275760175352e-05,
+      "loss": 0.8906,
+      "step": 3022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915425992944974e-05,
+      "loss": 0.9492,
+      "step": 3023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914575911788686e-05,
+      "loss": 0.9727,
+      "step": 3024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913725516736413e-05,
+      "loss": 0.918,
+      "step": 3025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89128748078181e-05,
+      "loss": 1.0078,
+      "step": 3026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912023785063706e-05,
+      "loss": 0.8984,
+      "step": 3027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911172448503193e-05,
+      "loss": 0.8477,
+      "step": 3028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910320798166538e-05,
+      "loss": 0.9375,
+      "step": 3029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909468834083736e-05,
+      "loss": 0.9883,
+      "step": 3030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908616556284776e-05,
+      "loss": 0.9336,
+      "step": 3031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907763964799673e-05,
+      "loss": 0.9414,
+      "step": 3032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906911059658445e-05,
+      "loss": 0.8906,
+      "step": 3033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890605784089113e-05,
+      "loss": 1.0547,
+      "step": 3034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905204308527767e-05,
+      "loss": 0.8477,
+      "step": 3035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904350462598406e-05,
+      "loss": 0.8906,
+      "step": 3036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890349630313312e-05,
+      "loss": 0.8945,
+      "step": 3037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902641830161985e-05,
+      "loss": 0.8945,
+      "step": 3038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901787043715083e-05,
+      "loss": 0.918,
+      "step": 3039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900931943822517e-05,
+      "loss": 0.9609,
+      "step": 3040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890007653051439e-05,
+      "loss": 0.8828,
+      "step": 3041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899220803820828e-05,
+      "loss": 0.9062,
+      "step": 3042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898364763771963e-05,
+      "loss": 0.8594,
+      "step": 3043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897508410397934e-05,
+      "loss": 1.0078,
+      "step": 3044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896651743728898e-05,
+      "loss": 0.9258,
+      "step": 3045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895794763795016e-05,
+      "loss": 0.8984,
+      "step": 3046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894937470626466e-05,
+      "loss": 1.0469,
+      "step": 3047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894079864253434e-05,
+      "loss": 0.9961,
+      "step": 3048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893221944706117e-05,
+      "loss": 0.8789,
+      "step": 3049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892363712014724e-05,
+      "loss": 0.9688,
+      "step": 3050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891505166209478e-05,
+      "loss": 0.9141,
+      "step": 3051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890646307320603e-05,
+      "loss": 0.9727,
+      "step": 3052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889787135378346e-05,
+      "loss": 0.8594,
+      "step": 3053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888927650412956e-05,
+      "loss": 0.875,
+      "step": 3054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888067852454703e-05,
+      "loss": 0.9414,
+      "step": 3055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887207741533852e-05,
+      "loss": 0.9062,
+      "step": 3056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88863473176807e-05,
+      "loss": 0.8516,
+      "step": 3057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885486580925535e-05,
+      "loss": 0.9844,
+      "step": 3058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888462553129867e-05,
+      "loss": 0.9531,
+      "step": 3059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883764168830423e-05,
+      "loss": 0.9883,
+      "step": 3060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882902493551123e-05,
+      "loss": 0.9492,
+      "step": 3061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888204050549111e-05,
+      "loss": 0.8516,
+      "step": 3062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881178204680736e-05,
+      "loss": 1.0078,
+      "step": 3063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880315591150366e-05,
+      "loss": 0.9297,
+      "step": 3064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879452664930374e-05,
+      "loss": 0.8984,
+      "step": 3065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887858942605114e-05,
+      "loss": 1.0,
+      "step": 3066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887772587454307e-05,
+      "loss": 0.8711,
+      "step": 3067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876862010436562e-05,
+      "loss": 0.8906,
+      "step": 3068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875997833762033e-05,
+      "loss": 0.9375,
+      "step": 3069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887513334454992e-05,
+      "loss": 0.9961,
+      "step": 3070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887426854283066e-05,
+      "loss": 1.0391,
+      "step": 3071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873403428634697e-05,
+      "loss": 0.8906,
+      "step": 3072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872538001992503e-05,
+      "loss": 0.9141,
+      "step": 3073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871672262934545e-05,
+      "loss": 0.9531,
+      "step": 3074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870806211491312e-05,
+      "loss": 0.8906,
+      "step": 3075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886993984769329e-05,
+      "loss": 0.9766,
+      "step": 3076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869073171570997e-05,
+      "loss": 0.957,
+      "step": 3077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868206183154944e-05,
+      "loss": 0.9531,
+      "step": 3078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867338882475654e-05,
+      "loss": 0.8438,
+      "step": 3079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866471269563672e-05,
+      "loss": 0.8945,
+      "step": 3080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886560334444955e-05,
+      "loss": 0.9453,
+      "step": 3081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864735107163842e-05,
+      "loss": 0.8555,
+      "step": 3082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863866557737126e-05,
+      "loss": 0.832,
+      "step": 3083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862997696199982e-05,
+      "loss": 0.8164,
+      "step": 3084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862128522583004e-05,
+      "loss": 1.0156,
+      "step": 3085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861259036916796e-05,
+      "loss": 0.875,
+      "step": 3086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886038923923198e-05,
+      "loss": 0.9492,
+      "step": 3087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859519129559175e-05,
+      "loss": 1.0156,
+      "step": 3088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858648707929027e-05,
+      "loss": 0.9961,
+      "step": 3089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857777974372172e-05,
+      "loss": 1.0156,
+      "step": 3090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856906928919284e-05,
+      "loss": 0.9258,
+      "step": 3091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856035571601027e-05,
+      "loss": 0.9805,
+      "step": 3092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855163902448086e-05,
+      "loss": 0.9922,
+      "step": 3093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854291921491147e-05,
+      "loss": 0.9023,
+      "step": 3094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853419628760922e-05,
+      "loss": 0.9492,
+      "step": 3095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852547024288122e-05,
+      "loss": 0.8281,
+      "step": 3096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885167410810347e-05,
+      "loss": 1.0234,
+      "step": 3097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885080088023771e-05,
+      "loss": 0.8789,
+      "step": 3098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849927340721583e-05,
+      "loss": 0.8633,
+      "step": 3099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884905348958585e-05,
+      "loss": 0.832,
+      "step": 3100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848179326861283e-05,
+      "loss": 0.9141,
+      "step": 3101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847304852578658e-05,
+      "loss": 0.8906,
+      "step": 3102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884643006676877e-05,
+      "loss": 1.0781,
+      "step": 3103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845554969462417e-05,
+      "loss": 0.8281,
+      "step": 3104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844679560690422e-05,
+      "loss": 0.8906,
+      "step": 3105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88438038404836e-05,
+      "loss": 0.9688,
+      "step": 3106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842927808872788e-05,
+      "loss": 0.9961,
+      "step": 3107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884205146588884e-05,
+      "loss": 0.8555,
+      "step": 3108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841174811562603e-05,
+      "loss": 0.9766,
+      "step": 3109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840297845924952e-05,
+      "loss": 0.8477,
+      "step": 3110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839420569006765e-05,
+      "loss": 0.9102,
+      "step": 3111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838542980838933e-05,
+      "loss": 0.9531,
+      "step": 3112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837665081452354e-05,
+      "loss": 0.9688,
+      "step": 3113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836786870877942e-05,
+      "loss": 0.8984,
+      "step": 3114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835908349146617e-05,
+      "loss": 0.8555,
+      "step": 3115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835029516289324e-05,
+      "loss": 0.8477,
+      "step": 3116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883415037233699e-05,
+      "loss": 0.9492,
+      "step": 3117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883327091732059e-05,
+      "loss": 0.9023,
+      "step": 3118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883239115127108e-05,
+      "loss": 1.0234,
+      "step": 3119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883151107421944e-05,
+      "loss": 0.8477,
+      "step": 3120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830630686196662e-05,
+      "loss": 0.9492,
+      "step": 3121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882974998723374e-05,
+      "loss": 0.8594,
+      "step": 3122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828868977361693e-05,
+      "loss": 0.9258,
+      "step": 3123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882798765661153e-05,
+      "loss": 0.9492,
+      "step": 3124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827106025014297e-05,
+      "loss": 0.9453,
+      "step": 3125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882622408260103e-05,
+      "loss": 1.0,
+      "step": 3126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825341829402783e-05,
+      "loss": 0.8633,
+      "step": 3127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824459265450626e-05,
+      "loss": 0.9453,
+      "step": 3128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823576390775633e-05,
+      "loss": 0.9062,
+      "step": 3129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882269320540889e-05,
+      "loss": 0.9375,
+      "step": 3130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88218097093815e-05,
+      "loss": 0.9453,
+      "step": 3131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882092590272457e-05,
+      "loss": 0.9609,
+      "step": 3132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820041785469215e-05,
+      "loss": 0.875,
+      "step": 3133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819157357646577e-05,
+      "loss": 0.7852,
+      "step": 3134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818272619287786e-05,
+      "loss": 0.9297,
+      "step": 3135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817387570424003e-05,
+      "loss": 1.0156,
+      "step": 3136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816502211086388e-05,
+      "loss": 0.8594,
+      "step": 3137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815616541306122e-05,
+      "loss": 0.8555,
+      "step": 3138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881473056111438e-05,
+      "loss": 0.9258,
+      "step": 3139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881384427054237e-05,
+      "loss": 0.9844,
+      "step": 3140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812957669621293e-05,
+      "loss": 0.875,
+      "step": 3141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812070758382367e-05,
+      "loss": 0.9375,
+      "step": 3142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811183536856828e-05,
+      "loss": 0.9062,
+      "step": 3143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881029600507591e-05,
+      "loss": 0.9727,
+      "step": 3144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809408163070866e-05,
+      "loss": 1.0,
+      "step": 3145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880852001087296e-05,
+      "loss": 0.9492,
+      "step": 3146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807631548513465e-05,
+      "loss": 0.8125,
+      "step": 3147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880674277602366e-05,
+      "loss": 0.8281,
+      "step": 3148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880585369343485e-05,
+      "loss": 0.793,
+      "step": 3149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880496430077833e-05,
+      "loss": 0.8633,
+      "step": 3150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804074598085425e-05,
+      "loss": 0.832,
+      "step": 3151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880318458538746e-05,
+      "loss": 1.0312,
+      "step": 3152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802294262715774e-05,
+      "loss": 0.9336,
+      "step": 3153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801403630101715e-05,
+      "loss": 1.0703,
+      "step": 3154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800512687576647e-05,
+      "loss": 0.9844,
+      "step": 3155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799621435171938e-05,
+      "loss": 0.8203,
+      "step": 3156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879872987291897e-05,
+      "loss": 1.0391,
+      "step": 3157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879783800084914e-05,
+      "loss": 0.8516,
+      "step": 3158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879694581899385e-05,
+      "loss": 0.9531,
+      "step": 3159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796053327384515e-05,
+      "loss": 1.0625,
+      "step": 3160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879516052605256e-05,
+      "loss": 0.9531,
+      "step": 3161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794267415029424e-05,
+      "loss": 0.9453,
+      "step": 3162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793373994346554e-05,
+      "loss": 0.8672,
+      "step": 3163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792480264035408e-05,
+      "loss": 0.9805,
+      "step": 3164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879158622412746e-05,
+      "loss": 0.8516,
+      "step": 3165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879069187465418e-05,
+      "loss": 0.8398,
+      "step": 3166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789797215647073e-05,
+      "loss": 0.8867,
+      "step": 3167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788902247137635e-05,
+      "loss": 0.9492,
+      "step": 3168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878800696915737e-05,
+      "loss": 0.8828,
+      "step": 3169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787111381737822e-05,
+      "loss": 0.8516,
+      "step": 3170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786215484910507e-05,
+      "loss": 0.9531,
+      "step": 3171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785319278706982e-05,
+      "loss": 0.9609,
+      "step": 3172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784422763158802e-05,
+      "loss": 1.0391,
+      "step": 3173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878352593829753e-05,
+      "loss": 0.8633,
+      "step": 3174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782628804154752e-05,
+      "loss": 0.9258,
+      "step": 3175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878173136076205e-05,
+      "loss": 0.9805,
+      "step": 3176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878083360815103e-05,
+      "loss": 0.9805,
+      "step": 3177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779935546353302e-05,
+      "loss": 0.8242,
+      "step": 3178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779037175400485e-05,
+      "loss": 1.0,
+      "step": 3179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778138495324213e-05,
+      "loss": 1.0312,
+      "step": 3180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777239506156136e-05,
+      "loss": 0.9219,
+      "step": 3181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776340207927897e-05,
+      "loss": 1.0078,
+      "step": 3182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775440600671174e-05,
+      "loss": 0.9531,
+      "step": 3183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774540684417638e-05,
+      "loss": 0.9766,
+      "step": 3184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773640459198975e-05,
+      "loss": 0.9414,
+      "step": 3185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772739925046885e-05,
+      "loss": 0.8711,
+      "step": 3186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877183908199308e-05,
+      "loss": 0.957,
+      "step": 3187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770937930069272e-05,
+      "loss": 0.9766,
+      "step": 3188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87700364693072e-05,
+      "loss": 0.8008,
+      "step": 3189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769134699738606e-05,
+      "loss": 0.9258,
+      "step": 3190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768232621395234e-05,
+      "loss": 1.0391,
+      "step": 3191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767330234308857e-05,
+      "loss": 0.9922,
+      "step": 3192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8766427538511246e-05,
+      "loss": 0.8906,
+      "step": 3193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765524534034182e-05,
+      "loss": 0.8906,
+      "step": 3194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876462122090947e-05,
+      "loss": 0.8672,
+      "step": 3195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763717599168912e-05,
+      "loss": 0.9531,
+      "step": 3196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762813668844325e-05,
+      "loss": 0.9023,
+      "step": 3197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761909429967536e-05,
+      "loss": 0.9297,
+      "step": 3198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876100488257039e-05,
+      "loss": 0.8711,
+      "step": 3199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760100026684736e-05,
+      "loss": 0.9297,
+      "step": 3200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759194862342433e-05,
+      "loss": 0.9609,
+      "step": 3201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758289389575355e-05,
+      "loss": 0.9219,
+      "step": 3202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757383608415385e-05,
+      "loss": 0.9609,
+      "step": 3203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875647751889442e-05,
+      "loss": 0.918,
+      "step": 3204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755571121044357e-05,
+      "loss": 0.9883,
+      "step": 3205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754664414897117e-05,
+      "loss": 0.9961,
+      "step": 3206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753757400484627e-05,
+      "loss": 0.9531,
+      "step": 3207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752850077838822e-05,
+      "loss": 0.9336,
+      "step": 3208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875194244699165e-05,
+      "loss": 0.8984,
+      "step": 3209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751034507975078e-05,
+      "loss": 0.9531,
+      "step": 3210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875012626082106e-05,
+      "loss": 0.8789,
+      "step": 3211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749217705561596e-05,
+      "loss": 0.8516,
+      "step": 3212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874830884222866e-05,
+      "loss": 0.8828,
+      "step": 3213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874739967085427e-05,
+      "loss": 0.9609,
+      "step": 3214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746490191470426e-05,
+      "loss": 0.9414,
+      "step": 3215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745580404109157e-05,
+      "loss": 0.7188,
+      "step": 3216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87446703088025e-05,
+      "loss": 0.8789,
+      "step": 3217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87437599055825e-05,
+      "loss": 0.8672,
+      "step": 3218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742849194481218e-05,
+      "loss": 0.875,
+      "step": 3219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874193817553071e-05,
+      "loss": 1.0,
+      "step": 3220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741026848763063e-05,
+      "loss": 1.0859,
+      "step": 3221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740115214210365e-05,
+      "loss": 1.0078,
+      "step": 3222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739203271904718e-05,
+      "loss": 1.0156,
+      "step": 3223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873829102187823e-05,
+      "loss": 0.8516,
+      "step": 3224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737378464163018e-05,
+      "loss": 0.9414,
+      "step": 3225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736465598791226e-05,
+      "loss": 0.9648,
+      "step": 3226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735552425794988e-05,
+      "loss": 0.8516,
+      "step": 3227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734638945206464e-05,
+      "loss": 0.8945,
+      "step": 3228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733725157057813e-05,
+      "loss": 0.9844,
+      "step": 3229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732811061381216e-05,
+      "loss": 0.9375,
+      "step": 3230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731896658208858e-05,
+      "loss": 0.9375,
+      "step": 3231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730981947572938e-05,
+      "loss": 0.9609,
+      "step": 3232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873006692950566e-05,
+      "loss": 0.9102,
+      "step": 3233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8729151604039252e-05,
+      "loss": 0.9062,
+      "step": 3234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728235971205935e-05,
+      "loss": 0.8672,
+      "step": 3235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872732003103795e-05,
+      "loss": 0.9375,
+      "step": 3236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726403783567556e-05,
+      "loss": 0.8945,
+      "step": 3237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725487228827008e-05,
+      "loss": 0.9766,
+      "step": 3238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724570366848585e-05,
+      "loss": 0.8516,
+      "step": 3239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723653197664567e-05,
+      "loss": 0.9023,
+      "step": 3240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722735721307253e-05,
+      "loss": 0.9336,
+      "step": 3241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721817937808944e-05,
+      "loss": 0.957,
+      "step": 3242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872089984720196e-05,
+      "loss": 0.9727,
+      "step": 3243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871998144951863e-05,
+      "loss": 1.0234,
+      "step": 3244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719062744791284e-05,
+      "loss": 0.9453,
+      "step": 3245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718143733052278e-05,
+      "loss": 0.8438,
+      "step": 3246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717224414333972e-05,
+      "loss": 0.9844,
+      "step": 3247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716304788668735e-05,
+      "loss": 0.9453,
+      "step": 3248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715384856088946e-05,
+      "loss": 1.0547,
+      "step": 3249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714464616627002e-05,
+      "loss": 0.875,
+      "step": 3250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713544070315305e-05,
+      "loss": 0.9258,
+      "step": 3251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871262321718627e-05,
+      "loss": 0.9648,
+      "step": 3252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871170205727231e-05,
+      "loss": 0.9844,
+      "step": 3253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710780590605878e-05,
+      "loss": 0.9805,
+      "step": 3254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709858817219407e-05,
+      "loss": 0.957,
+      "step": 3255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708936737145366e-05,
+      "loss": 1.0859,
+      "step": 3256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870801435041621e-05,
+      "loss": 0.9336,
+      "step": 3257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870709165706442e-05,
+      "loss": 0.8164,
+      "step": 3258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706168657122495e-05,
+      "loss": 1.0547,
+      "step": 3259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870524535062293e-05,
+      "loss": 0.8477,
+      "step": 3260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704321737598232e-05,
+      "loss": 0.8242,
+      "step": 3261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703397818080927e-05,
+      "loss": 0.8828,
+      "step": 3262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870247359210355e-05,
+      "loss": 0.9492,
+      "step": 3263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701549059698636e-05,
+      "loss": 0.9453,
+      "step": 3264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870062422089875e-05,
+      "loss": 0.8711,
+      "step": 3265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699699075736445e-05,
+      "loss": 1.0156,
+      "step": 3266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869877362424431e-05,
+      "loss": 0.9531,
+      "step": 3267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697847866454922e-05,
+      "loss": 1.0234,
+      "step": 3268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696921802400882e-05,
+      "loss": 1.0234,
+      "step": 3269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869599543211479e-05,
+      "loss": 0.9492,
+      "step": 3270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695068755629282e-05,
+      "loss": 0.9375,
+      "step": 3271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694141772976975e-05,
+      "loss": 0.9961,
+      "step": 3272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693214484190512e-05,
+      "loss": 0.8359,
+      "step": 3273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692286889302543e-05,
+      "loss": 0.8594,
+      "step": 3274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691358988345734e-05,
+      "loss": 0.9375,
+      "step": 3275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690430781352754e-05,
+      "loss": 0.8594,
+      "step": 3276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868950226835629e-05,
+      "loss": 0.8984,
+      "step": 3277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688573449389033e-05,
+      "loss": 0.9453,
+      "step": 3278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868764432448369e-05,
+      "loss": 0.875,
+      "step": 3279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686714893672977e-05,
+      "loss": 0.9844,
+      "step": 3280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868578515698962e-05,
+      "loss": 0.9102,
+      "step": 3281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684855114466357e-05,
+      "loss": 0.9375,
+      "step": 3282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683924766135935e-05,
+      "loss": 0.9141,
+      "step": 3283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682994112031116e-05,
+      "loss": 0.9766,
+      "step": 3284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868206315218467e-05,
+      "loss": 0.8906,
+      "step": 3285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868113188662937e-05,
+      "loss": 0.8164,
+      "step": 3286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680200315398015e-05,
+      "loss": 0.9805,
+      "step": 3287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8679268438523402e-05,
+      "loss": 0.8867,
+      "step": 3288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678336256038353e-05,
+      "loss": 0.8477,
+      "step": 3289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867740376797568e-05,
+      "loss": 1.0234,
+      "step": 3290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676470974368224e-05,
+      "loss": 0.8594,
+      "step": 3291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675537875248827e-05,
+      "loss": 0.9805,
+      "step": 3292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674604470650347e-05,
+      "loss": 0.9531,
+      "step": 3293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673670760605654e-05,
+      "loss": 0.9531,
+      "step": 3294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672736745147616e-05,
+      "loss": 0.8984,
+      "step": 3295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671802424309132e-05,
+      "loss": 0.918,
+      "step": 3296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867086779812309e-05,
+      "loss": 0.9844,
+      "step": 3297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669932866622407e-05,
+      "loss": 0.8594,
+      "step": 3298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668997629840004e-05,
+      "loss": 0.9297,
+      "step": 3299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668062087808807e-05,
+      "loss": 1.1016,
+      "step": 3300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866712624056176e-05,
+      "loss": 0.9023,
+      "step": 3301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666190088131813e-05,
+      "loss": 0.8828,
+      "step": 3302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866525363055194e-05,
+      "loss": 0.8945,
+      "step": 3303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664316867855105e-05,
+      "loss": 0.918,
+      "step": 3304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663379800074292e-05,
+      "loss": 0.8711,
+      "step": 3305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86624424272425e-05,
+      "loss": 0.8164,
+      "step": 3306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866150474939274e-05,
+      "loss": 1.0625,
+      "step": 3307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866056676655802e-05,
+      "loss": 0.8281,
+      "step": 3308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659628478771374e-05,
+      "loss": 0.8945,
+      "step": 3309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865868988606584e-05,
+      "loss": 0.9258,
+      "step": 3310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657750988474465e-05,
+      "loss": 1.0625,
+      "step": 3311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865681178603031e-05,
+      "loss": 0.8125,
+      "step": 3312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655872278766446e-05,
+      "loss": 0.9297,
+      "step": 3313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654932466715954e-05,
+      "loss": 0.9219,
+      "step": 3314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865399234991193e-05,
+      "loss": 0.9062,
+      "step": 3315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865305192838747e-05,
+      "loss": 1.0156,
+      "step": 3316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652111202175693e-05,
+      "loss": 1.1406,
+      "step": 3317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865117017130972e-05,
+      "loss": 0.8477,
+      "step": 3318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865022883582269e-05,
+      "loss": 0.9453,
+      "step": 3319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649287195747744e-05,
+      "loss": 0.8672,
+      "step": 3320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648345251118044e-05,
+      "loss": 0.918,
+      "step": 3321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864740300196675e-05,
+      "loss": 0.8438,
+      "step": 3322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864646044832705e-05,
+      "loss": 1.0312,
+      "step": 3323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645517590232127e-05,
+      "loss": 0.8555,
+      "step": 3324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644574427715182e-05,
+      "loss": 0.9062,
+      "step": 3325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864363096080942e-05,
+      "loss": 1.0469,
+      "step": 3326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864268718954807e-05,
+      "loss": 0.8906,
+      "step": 3327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641743113964355e-05,
+      "loss": 0.9102,
+      "step": 3328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640798734091526e-05,
+      "loss": 0.9766,
+      "step": 3329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639854049962833e-05,
+      "loss": 0.8945,
+      "step": 3330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638909061611537e-05,
+      "loss": 0.9492,
+      "step": 3331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637963769070916e-05,
+      "loss": 0.9727,
+      "step": 3332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863701817237425e-05,
+      "loss": 0.875,
+      "step": 3333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636072271554842e-05,
+      "loss": 1.0312,
+      "step": 3334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635126066645997e-05,
+      "loss": 0.8633,
+      "step": 3335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863417955768103e-05,
+      "loss": 0.8906,
+      "step": 3336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633232744693265e-05,
+      "loss": 0.9297,
+      "step": 3337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632285627716046e-05,
+      "loss": 0.8867,
+      "step": 3338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631338206782724e-05,
+      "loss": 0.9062,
+      "step": 3339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630390481926656e-05,
+      "loss": 0.8633,
+      "step": 3340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629442453181217e-05,
+      "loss": 0.9219,
+      "step": 3341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862849412057978e-05,
+      "loss": 0.957,
+      "step": 3342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862754548415575e-05,
+      "loss": 0.8438,
+      "step": 3343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626596543942517e-05,
+      "loss": 0.8945,
+      "step": 3344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86256472999735e-05,
+      "loss": 0.9922,
+      "step": 3345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624697752282127e-05,
+      "loss": 0.9375,
+      "step": 3346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623747900901828e-05,
+      "loss": 0.9648,
+      "step": 3347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622797745866052e-05,
+      "loss": 0.9609,
+      "step": 3348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8621847287208253e-05,
+      "loss": 0.9062,
+      "step": 3349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86208965249619e-05,
+      "loss": 0.875,
+      "step": 3350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619945459160465e-05,
+      "loss": 0.9805,
+      "step": 3351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618994089837442e-05,
+      "loss": 0.8125,
+      "step": 3352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618042417026333e-05,
+      "loss": 0.875,
+      "step": 3353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617090440760642e-05,
+      "loss": 1.0703,
+      "step": 3354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616138161073893e-05,
+      "loss": 0.9062,
+      "step": 3355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615185577999616e-05,
+      "loss": 0.9102,
+      "step": 3356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614232691571352e-05,
+      "loss": 0.9102,
+      "step": 3357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613279501822656e-05,
+      "loss": 0.9453,
+      "step": 3358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861232600878709e-05,
+      "loss": 0.8945,
+      "step": 3359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611372212498228e-05,
+      "loss": 0.8828,
+      "step": 3360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861041811298965e-05,
+      "loss": 1.0,
+      "step": 3361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860946371029496e-05,
+      "loss": 0.8047,
+      "step": 3362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608509004447758e-05,
+      "loss": 0.9453,
+      "step": 3363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8607553995481664e-05,
+      "loss": 1.0234,
+      "step": 3364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606598683430306e-05,
+      "loss": 0.8555,
+      "step": 3365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605643068327313e-05,
+      "loss": 0.918,
+      "step": 3366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604687150206348e-05,
+      "loss": 0.957,
+      "step": 3367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860373092910106e-05,
+      "loss": 0.9297,
+      "step": 3368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860277440504512e-05,
+      "loss": 0.8828,
+      "step": 3369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860181757807221e-05,
+      "loss": 0.9062,
+      "step": 3370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600860448216026e-05,
+      "loss": 0.9609,
+      "step": 3371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599903015510264e-05,
+      "loss": 0.9766,
+      "step": 3372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598945279988637e-05,
+      "loss": 0.8828,
+      "step": 3373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859798724168487e-05,
+      "loss": 0.9453,
+      "step": 3374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85970289006327e-05,
+      "loss": 0.9492,
+      "step": 3375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596070256865866e-05,
+      "loss": 0.9297,
+      "step": 3376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595111310418125e-05,
+      "loss": 0.8008,
+      "step": 3377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594152061323245e-05,
+      "loss": 0.8438,
+      "step": 3378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593192509615003e-05,
+      "loss": 0.8672,
+      "step": 3379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592232655327184e-05,
+      "loss": 0.9258,
+      "step": 3380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591272498493587e-05,
+      "loss": 0.7734,
+      "step": 3381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859031203914802e-05,
+      "loss": 1.0781,
+      "step": 3382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589351277324304e-05,
+      "loss": 0.918,
+      "step": 3383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588390213056266e-05,
+      "loss": 0.9844,
+      "step": 3384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858742884637775e-05,
+      "loss": 0.8984,
+      "step": 3385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586467177322602e-05,
+      "loss": 0.9531,
+      "step": 3386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858550520592469e-05,
+      "loss": 0.8555,
+      "step": 3387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584542932217885e-05,
+      "loss": 0.9414,
+      "step": 3388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583580356236065e-05,
+      "loss": 0.9141,
+      "step": 3389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858261747801313e-05,
+      "loss": 0.9727,
+      "step": 3390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858165429758298e-05,
+      "loss": 0.918,
+      "step": 3391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580690814979536e-05,
+      "loss": 1.0078,
+      "step": 3392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579727030236716e-05,
+      "loss": 1.0469,
+      "step": 3393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578762943388463e-05,
+      "loss": 0.8594,
+      "step": 3394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857779855446872e-05,
+      "loss": 0.9297,
+      "step": 3395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576833863511446e-05,
+      "loss": 0.8242,
+      "step": 3396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575868870550612e-05,
+      "loss": 0.8945,
+      "step": 3397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574903575620188e-05,
+      "loss": 1.0469,
+      "step": 3398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573937978754173e-05,
+      "loss": 0.8438,
+      "step": 3399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857297207998656e-05,
+      "loss": 0.8828,
+      "step": 3400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857200587935137e-05,
+      "loss": 0.9727,
+      "step": 3401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8571039376882613e-05,
+      "loss": 0.9922,
+      "step": 3402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570072572614327e-05,
+      "loss": 0.9609,
+      "step": 3403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569105466580555e-05,
+      "loss": 0.9453,
+      "step": 3404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856813805881535e-05,
+      "loss": 1.0156,
+      "step": 3405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567170349352768e-05,
+      "loss": 0.8672,
+      "step": 3406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566202338226894e-05,
+      "loss": 1.0625,
+      "step": 3407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856523402547181e-05,
+      "loss": 0.9609,
+      "step": 3408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564265411121612e-05,
+      "loss": 1.0234,
+      "step": 3409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563296495210408e-05,
+      "loss": 0.9727,
+      "step": 3410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856232727777231e-05,
+      "loss": 0.8906,
+      "step": 3411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561357758841446e-05,
+      "loss": 0.9297,
+      "step": 3412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856038793845196e-05,
+      "loss": 0.9023,
+      "step": 3413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559417816637997e-05,
+      "loss": 0.9531,
+      "step": 3414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558447393433713e-05,
+      "loss": 0.9375,
+      "step": 3415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8557476668873286e-05,
+      "loss": 0.9727,
+      "step": 3416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556505642990894e-05,
+      "loss": 0.8164,
+      "step": 3417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555534315820725e-05,
+      "loss": 0.8906,
+      "step": 3418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554562687396984e-05,
+      "loss": 0.875,
+      "step": 3419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855359075775388e-05,
+      "loss": 1.0078,
+      "step": 3420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552618526925642e-05,
+      "loss": 0.9219,
+      "step": 3421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85516459949465e-05,
+      "loss": 0.8516,
+      "step": 3422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550673161850698e-05,
+      "loss": 0.9727,
+      "step": 3423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854970002767249e-05,
+      "loss": 0.7812,
+      "step": 3424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548726592446145e-05,
+      "loss": 1.0547,
+      "step": 3425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547752856205936e-05,
+      "loss": 0.8086,
+      "step": 3426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854677881898615e-05,
+      "loss": 0.8125,
+      "step": 3427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545804480821094e-05,
+      "loss": 0.8984,
+      "step": 3428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854482984174506e-05,
+      "loss": 0.8984,
+      "step": 3429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8543854901792375e-05,
+      "loss": 0.8438,
+      "step": 3430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542879660997367e-05,
+      "loss": 0.8867,
+      "step": 3431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854190411939438e-05,
+      "loss": 0.8906,
+      "step": 3432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540928277017755e-05,
+      "loss": 0.8359,
+      "step": 3433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853995213390186e-05,
+      "loss": 0.9062,
+      "step": 3434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538975690081064e-05,
+      "loss": 0.8477,
+      "step": 3435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537998945589756e-05,
+      "loss": 1.0,
+      "step": 3436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537021900462315e-05,
+      "loss": 0.9219,
+      "step": 3437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536044554733153e-05,
+      "loss": 0.8984,
+      "step": 3438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535066908436685e-05,
+      "loss": 0.8242,
+      "step": 3439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534088961607333e-05,
+      "loss": 1.0156,
+      "step": 3440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853311071427953e-05,
+      "loss": 0.9844,
+      "step": 3441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532132166487727e-05,
+      "loss": 1.0156,
+      "step": 3442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531153318266373e-05,
+      "loss": 1.0312,
+      "step": 3443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853017416964994e-05,
+      "loss": 0.9219,
+      "step": 3444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529194720672903e-05,
+      "loss": 0.9414,
+      "step": 3445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528214971369752e-05,
+      "loss": 1.0078,
+      "step": 3446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527234921774986e-05,
+      "loss": 0.8672,
+      "step": 3447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852625457192311e-05,
+      "loss": 0.8398,
+      "step": 3448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525273921848647e-05,
+      "loss": 0.9688,
+      "step": 3449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524292971586123e-05,
+      "loss": 1.0156,
+      "step": 3450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523311721170088e-05,
+      "loss": 0.9492,
+      "step": 3451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852233017063508e-05,
+      "loss": 0.8867,
+      "step": 3452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521348320015674e-05,
+      "loss": 0.9375,
+      "step": 3453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520366169346437e-05,
+      "loss": 0.9648,
+      "step": 3454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851938371866195e-05,
+      "loss": 0.9648,
+      "step": 3455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518400967996806e-05,
+      "loss": 0.8633,
+      "step": 3456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517417917385616e-05,
+      "loss": 0.8633,
+      "step": 3457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516434566862987e-05,
+      "loss": 0.9453,
+      "step": 3458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851545091646355e-05,
+      "loss": 0.9062,
+      "step": 3459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514466966221936e-05,
+      "loss": 0.9062,
+      "step": 3460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513482716172793e-05,
+      "loss": 0.8945,
+      "step": 3461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851249816635078e-05,
+      "loss": 0.9141,
+      "step": 3462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511513316790566e-05,
+      "loss": 0.9961,
+      "step": 3463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510528167526825e-05,
+      "loss": 0.918,
+      "step": 3464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509542718594246e-05,
+      "loss": 1.0781,
+      "step": 3465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850855697002753e-05,
+      "loss": 1.0,
+      "step": 3466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507570921861383e-05,
+      "loss": 1.0078,
+      "step": 3467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506584574130532e-05,
+      "loss": 1.0156,
+      "step": 3468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505597926869702e-05,
+      "loss": 0.957,
+      "step": 3469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504610980113637e-05,
+      "loss": 0.957,
+      "step": 3470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850362373389709e-05,
+      "loss": 0.8516,
+      "step": 3471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502636188254817e-05,
+      "loss": 1.0859,
+      "step": 3472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85016483432216e-05,
+      "loss": 0.8945,
+      "step": 3473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500660198832217e-05,
+      "loss": 0.9336,
+      "step": 3474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499671755121466e-05,
+      "loss": 0.8906,
+      "step": 3475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498683012124145e-05,
+      "loss": 0.8711,
+      "step": 3476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497693969875074e-05,
+      "loss": 0.8867,
+      "step": 3477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849670462840908e-05,
+      "loss": 0.8828,
+      "step": 3478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495714987760995e-05,
+      "loss": 0.8984,
+      "step": 3479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849472504796567e-05,
+      "loss": 0.875,
+      "step": 3480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493734809057958e-05,
+      "loss": 0.9609,
+      "step": 3481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492744271072732e-05,
+      "loss": 0.8633,
+      "step": 3482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491753434044866e-05,
+      "loss": 0.8984,
+      "step": 3483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849076229800925e-05,
+      "loss": 1.0,
+      "step": 3484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489770863000784e-05,
+      "loss": 0.9766,
+      "step": 3485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488779129054374e-05,
+      "loss": 0.8242,
+      "step": 3486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848778709620495e-05,
+      "loss": 0.9141,
+      "step": 3487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486794764487436e-05,
+      "loss": 1.0547,
+      "step": 3488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485802133936774e-05,
+      "loss": 0.918,
+      "step": 3489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484809204587914e-05,
+      "loss": 0.875,
+      "step": 3490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483815976475826e-05,
+      "loss": 0.9141,
+      "step": 3491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482822449635475e-05,
+      "loss": 0.9922,
+      "step": 3492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481828624101848e-05,
+      "loss": 0.8555,
+      "step": 3493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480834499909942e-05,
+      "loss": 1.0234,
+      "step": 3494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847984007709476e-05,
+      "loss": 0.9414,
+      "step": 3495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847884535569131e-05,
+      "loss": 0.8047,
+      "step": 3496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847785033573463e-05,
+      "loss": 0.9922,
+      "step": 3497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476855017259746e-05,
+      "loss": 0.8828,
+      "step": 3498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475859400301708e-05,
+      "loss": 0.8984,
+      "step": 3499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474863484895578e-05,
+      "loss": 0.9531,
+      "step": 3500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473867271076414e-05,
+      "loss": 0.9844,
+      "step": 3501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472870758879303e-05,
+      "loss": 1.0234,
+      "step": 3502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471873948339332e-05,
+      "loss": 0.9375,
+      "step": 3503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470876839491597e-05,
+      "loss": 0.9258,
+      "step": 3504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846987943237121e-05,
+      "loss": 0.9023,
+      "step": 3505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846888172701329e-05,
+      "loss": 0.8359,
+      "step": 3506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846788372345297e-05,
+      "loss": 0.9727,
+      "step": 3507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846688542172539e-05,
+      "loss": 0.8516,
+      "step": 3508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465886821865698e-05,
+      "loss": 0.9141,
+      "step": 3509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464887923909062e-05,
+      "loss": 0.9492,
+      "step": 3510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463888727890653e-05,
+      "loss": 0.832,
+      "step": 3511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462889233845653e-05,
+      "loss": 0.8906,
+      "step": 3512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461889441809258e-05,
+      "loss": 0.9922,
+      "step": 3513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460889351816667e-05,
+      "loss": 0.9609,
+      "step": 3514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84598889639031e-05,
+      "loss": 1.0391,
+      "step": 3515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845888827810378e-05,
+      "loss": 0.9141,
+      "step": 3516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457887294453948e-05,
+      "loss": 1.0,
+      "step": 3517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456886012988838e-05,
+      "loss": 0.8516,
+      "step": 3518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845588443374372e-05,
+      "loss": 0.7695,
+      "step": 3519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454882556753854e-05,
+      "loss": 1.0391,
+      "step": 3520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453880382054517e-05,
+      "loss": 0.8555,
+      "step": 3521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452877909681e-05,
+      "loss": 0.9609,
+      "step": 3522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84518751396686e-05,
+      "loss": 1.1328,
+      "step": 3523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450872072052627e-05,
+      "loss": 0.9297,
+      "step": 3524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449868706868402e-05,
+      "loss": 0.8711,
+      "step": 3525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844886504415125e-05,
+      "loss": 0.9375,
+      "step": 3526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447861083936516e-05,
+      "loss": 0.9453,
+      "step": 3527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844685682625955e-05,
+      "loss": 1.0312,
+      "step": 3528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445852271155712e-05,
+      "loss": 0.9141,
+      "step": 3529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444847418660378e-05,
+      "loss": 0.8789,
+      "step": 3530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443842268808927e-05,
+      "loss": 0.9258,
+      "step": 3531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844283682163675e-05,
+      "loss": 1.0234,
+      "step": 3532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441831077179255e-05,
+      "loss": 0.7734,
+      "step": 3533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440825035471852e-05,
+      "loss": 0.8125,
+      "step": 3534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843981869654997e-05,
+      "loss": 0.9727,
+      "step": 3535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438812060449035e-05,
+      "loss": 0.8242,
+      "step": 3536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8437805127204503e-05,
+      "loss": 0.9688,
+      "step": 3537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8436797896851825e-05,
+      "loss": 1.0234,
+      "step": 3538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435790369426464e-05,
+      "loss": 0.9453,
+      "step": 3539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434782544963898e-05,
+      "loss": 0.9453,
+      "step": 3540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433774423499616e-05,
+      "loss": 1.0781,
+      "step": 3541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432766005069117e-05,
+      "loss": 0.7461,
+      "step": 3542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431757289707905e-05,
+      "loss": 0.9219,
+      "step": 3543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84307482774515e-05,
+      "loss": 0.9883,
+      "step": 3544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429738968335432e-05,
+      "loss": 0.8984,
+      "step": 3545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842872936239524e-05,
+      "loss": 0.9844,
+      "step": 3546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427719459666473e-05,
+      "loss": 0.8594,
+      "step": 3547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842670926018469e-05,
+      "loss": 0.8633,
+      "step": 3548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425698763985466e-05,
+      "loss": 0.9258,
+      "step": 3549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842468797110438e-05,
+      "loss": 0.9375,
+      "step": 3550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842367688157702e-05,
+      "loss": 0.8047,
+      "step": 3551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842266549543899e-05,
+      "loss": 1.0938,
+      "step": 3552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421653812725905e-05,
+      "loss": 0.9023,
+      "step": 3553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420641833473386e-05,
+      "loss": 0.8867,
+      "step": 3554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841962955771707e-05,
+      "loss": 0.793,
+      "step": 3555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418616985492592e-05,
+      "loss": 0.793,
+      "step": 3556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417604116835614e-05,
+      "loss": 0.9453,
+      "step": 3557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416590951781797e-05,
+      "loss": 0.9219,
+      "step": 3558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415577490366815e-05,
+      "loss": 1.0625,
+      "step": 3559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841456373262636e-05,
+      "loss": 0.9062,
+      "step": 3560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413549678596126e-05,
+      "loss": 0.9023,
+      "step": 3561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.9492,
+      "step": 3562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411520681809145e-05,
+      "loss": 0.9219,
+      "step": 3563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410505739123846e-05,
+      "loss": 0.9023,
+      "step": 3564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840949050029165e-05,
+      "loss": 0.9258,
+      "step": 3565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408474965348312e-05,
+      "loss": 0.8125,
+      "step": 3566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840745913432959e-05,
+      "loss": 0.8594,
+      "step": 3567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406443007271247e-05,
+      "loss": 0.8789,
+      "step": 3568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840542658420907e-05,
+      "loss": 0.957,
+      "step": 3569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404409865178843e-05,
+      "loss": 0.8594,
+      "step": 3570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403392850216367e-05,
+      "loss": 0.9375,
+      "step": 3571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840237553935746e-05,
+      "loss": 0.9961,
+      "step": 3572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840135793263793e-05,
+      "loss": 0.9805,
+      "step": 3573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840034003009362e-05,
+      "loss": 0.9648,
+      "step": 3574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399321831760363e-05,
+      "loss": 0.9961,
+      "step": 3575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839830333767402e-05,
+      "loss": 0.8945,
+      "step": 3576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397284547870445e-05,
+      "loss": 0.8789,
+      "step": 3577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839626546238552e-05,
+      "loss": 0.9297,
+      "step": 3578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8395246081255124e-05,
+      "loss": 0.9414,
+      "step": 3579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394226404515148e-05,
+      "loss": 0.9688,
+      "step": 3580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393206432201504e-05,
+      "loss": 1.0156,
+      "step": 3581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392186164350102e-05,
+      "loss": 0.8516,
+      "step": 3582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391165600996865e-05,
+      "loss": 0.8516,
+      "step": 3583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839014474217773e-05,
+      "loss": 0.9414,
+      "step": 3584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389123587928648e-05,
+      "loss": 0.9688,
+      "step": 3585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838810213828557e-05,
+      "loss": 0.9453,
+      "step": 3586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387080393284466e-05,
+      "loss": 0.8359,
+      "step": 3587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386058352961308e-05,
+      "loss": 1.0625,
+      "step": 3588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385036017352093e-05,
+      "loss": 0.8906,
+      "step": 3589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384013386492807e-05,
+      "loss": 1.0625,
+      "step": 3590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838299046041947e-05,
+      "loss": 0.9219,
+      "step": 3591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381967239168092e-05,
+      "loss": 0.8828,
+      "step": 3592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380943722774708e-05,
+      "loss": 0.9062,
+      "step": 3593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379919911275353e-05,
+      "loss": 0.9688,
+      "step": 3594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837889580470608e-05,
+      "loss": 0.9531,
+      "step": 3595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377871403102952e-05,
+      "loss": 0.9453,
+      "step": 3596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376846706502035e-05,
+      "loss": 0.9297,
+      "step": 3597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375821714939408e-05,
+      "loss": 0.9453,
+      "step": 3598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837479642845117e-05,
+      "loss": 0.9219,
+      "step": 3599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373770847073417e-05,
+      "loss": 0.8984,
+      "step": 3600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372744970842267e-05,
+      "loss": 0.8008,
+      "step": 3601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371718799793833e-05,
+      "loss": 0.8477,
+      "step": 3602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837069233396426e-05,
+      "loss": 0.9297,
+      "step": 3603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369665573389683e-05,
+      "loss": 0.9375,
+      "step": 3604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836863851810626e-05,
+      "loss": 0.9531,
+      "step": 3605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367611168150152e-05,
+      "loss": 1.0391,
+      "step": 3606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366583523557534e-05,
+      "loss": 0.8867,
+      "step": 3607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365555584364597e-05,
+      "loss": 0.9219,
+      "step": 3608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364527350607527e-05,
+      "loss": 1.0234,
+      "step": 3609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363498822322536e-05,
+      "loss": 0.9531,
+      "step": 3610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362469999545838e-05,
+      "loss": 0.8984,
+      "step": 3611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361440882313658e-05,
+      "loss": 0.8438,
+      "step": 3612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360411470662237e-05,
+      "loss": 0.9219,
+      "step": 3613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359381764627816e-05,
+      "loss": 0.8633,
+      "step": 3614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835835176424666e-05,
+      "loss": 1.0234,
+      "step": 3615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835732146955503e-05,
+      "loss": 1.0547,
+      "step": 3616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356290880589213e-05,
+      "loss": 1.1016,
+      "step": 3617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355259997385485e-05,
+      "loss": 0.9531,
+      "step": 3618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354228819980156e-05,
+      "loss": 0.918,
+      "step": 3619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835319734840953e-05,
+      "loss": 0.9414,
+      "step": 3620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835216558270993e-05,
+      "loss": 0.9688,
+      "step": 3621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351133522917682e-05,
+      "loss": 0.9688,
+      "step": 3622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350101169069133e-05,
+      "loss": 0.8359,
+      "step": 3623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349068521200622e-05,
+      "loss": 0.8672,
+      "step": 3624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348035579348523e-05,
+      "loss": 0.8906,
+      "step": 3625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83470023435492e-05,
+      "loss": 0.8945,
+      "step": 3626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345968813839037e-05,
+      "loss": 0.9844,
+      "step": 3627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344934990254425e-05,
+      "loss": 0.793,
+      "step": 3628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343900872831767e-05,
+      "loss": 0.832,
+      "step": 3629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8342866461607478e-05,
+      "loss": 0.9688,
+      "step": 3630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834183175661798e-05,
+      "loss": 0.9766,
+      "step": 3631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340796757899705e-05,
+      "loss": 0.9336,
+      "step": 3632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339761465489098e-05,
+      "loss": 0.8672,
+      "step": 3633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338725879422613e-05,
+      "loss": 0.8125,
+      "step": 3634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337689999736717e-05,
+      "loss": 0.9336,
+      "step": 3635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833665382646788e-05,
+      "loss": 0.9453,
+      "step": 3636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335617359652593e-05,
+      "loss": 0.9023,
+      "step": 3637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334580599327345e-05,
+      "loss": 0.9922,
+      "step": 3638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333543545528647e-05,
+      "loss": 0.8945,
+      "step": 3639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332506198293012e-05,
+      "loss": 0.9375,
+      "step": 3640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331468557656973e-05,
+      "loss": 0.9727,
+      "step": 3641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833043062365706e-05,
+      "loss": 1.0312,
+      "step": 3642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329392396329822e-05,
+      "loss": 0.8867,
+      "step": 3643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328353875711816e-05,
+      "loss": 0.9688,
+      "step": 3644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327315061839613e-05,
+      "loss": 0.8594,
+      "step": 3645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326275954749788e-05,
+      "loss": 0.9102,
+      "step": 3646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832523655447893e-05,
+      "loss": 0.9297,
+      "step": 3647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324196861063644e-05,
+      "loss": 0.9336,
+      "step": 3648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832315687454053e-05,
+      "loss": 0.9336,
+      "step": 3649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322116594946214e-05,
+      "loss": 1.0312,
+      "step": 3650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832107602231732e-05,
+      "loss": 0.9805,
+      "step": 3651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320035156690494e-05,
+      "loss": 0.9141,
+      "step": 3652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831899399810238e-05,
+      "loss": 0.9922,
+      "step": 3653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317952546589648e-05,
+      "loss": 0.8789,
+      "step": 3654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831691080218896e-05,
+      "loss": 0.8516,
+      "step": 3655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315868764937004e-05,
+      "loss": 0.9688,
+      "step": 3656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314826434870466e-05,
+      "loss": 0.9453,
+      "step": 3657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313783812026056e-05,
+      "loss": 0.8945,
+      "step": 3658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831274089644048e-05,
+      "loss": 0.9766,
+      "step": 3659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831169768815046e-05,
+      "loss": 0.8203,
+      "step": 3660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310654187192735e-05,
+      "loss": 1.0703,
+      "step": 3661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830961039360404e-05,
+      "loss": 0.8203,
+      "step": 3662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308566307421135e-05,
+      "loss": 0.8711,
+      "step": 3663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307521928680782e-05,
+      "loss": 1.0156,
+      "step": 3664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8306477257419756e-05,
+      "loss": 1.0312,
+      "step": 3665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305432293674837e-05,
+      "loss": 0.9727,
+      "step": 3666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304387037482826e-05,
+      "loss": 0.875,
+      "step": 3667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303341488880522e-05,
+      "loss": 0.8828,
+      "step": 3668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830229564790475e-05,
+      "loss": 0.8945,
+      "step": 3669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301249514592327e-05,
+      "loss": 0.8516,
+      "step": 3670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300203088980092e-05,
+      "loss": 1.0156,
+      "step": 3671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829915637110489e-05,
+      "loss": 1.0312,
+      "step": 3672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298109361003577e-05,
+      "loss": 0.8359,
+      "step": 3673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829706205871302e-05,
+      "loss": 1.0625,
+      "step": 3674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296014464270102e-05,
+      "loss": 0.9062,
+      "step": 3675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294966577711704e-05,
+      "loss": 1.0156,
+      "step": 3676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293918399074725e-05,
+      "loss": 0.9023,
+      "step": 3677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292869928396073e-05,
+      "loss": 0.9805,
+      "step": 3678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291821165712666e-05,
+      "loss": 0.9805,
+      "step": 3679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290772111061433e-05,
+      "loss": 0.918,
+      "step": 3680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289722764479316e-05,
+      "loss": 1.1875,
+      "step": 3681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828867312600326e-05,
+      "loss": 0.875,
+      "step": 3682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287623195670222e-05,
+      "loss": 0.8828,
+      "step": 3683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286572973517178e-05,
+      "loss": 0.9453,
+      "step": 3684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285522459581107e-05,
+      "loss": 0.918,
+      "step": 3685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284471653898995e-05,
+      "loss": 0.9062,
+      "step": 3686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283420556507846e-05,
+      "loss": 0.9062,
+      "step": 3687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828236916744467e-05,
+      "loss": 0.8828,
+      "step": 3688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281317486746487e-05,
+      "loss": 0.9453,
+      "step": 3689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828026551445033e-05,
+      "loss": 0.9375,
+      "step": 3690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279213250593235e-05,
+      "loss": 0.9297,
+      "step": 3691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278160695212264e-05,
+      "loss": 0.8594,
+      "step": 3692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827710784834447e-05,
+      "loss": 0.8125,
+      "step": 3693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827605471002693e-05,
+      "loss": 0.9844,
+      "step": 3694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275001280296724e-05,
+      "loss": 0.832,
+      "step": 3695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273947559190948e-05,
+      "loss": 0.9375,
+      "step": 3696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272893546746706e-05,
+      "loss": 0.8555,
+      "step": 3697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271839243001105e-05,
+      "loss": 1.0625,
+      "step": 3698
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270784647991275e-05,
+      "loss": 0.8789,
+      "step": 3699
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269729761754346e-05,
+      "loss": 0.9805,
+      "step": 3700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268674584327463e-05,
+      "loss": 0.9141,
+      "step": 3701
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267619115747784e-05,
+      "loss": 0.9648,
+      "step": 3702
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266563356052465e-05,
+      "loss": 0.9258,
+      "step": 3703
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265507305278693e-05,
+      "loss": 0.9336,
+      "step": 3704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264450963463644e-05,
+      "loss": 0.8672,
+      "step": 3705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263394330644517e-05,
+      "loss": 0.9297,
+      "step": 3706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262337406858518e-05,
+      "loss": 1.0312,
+      "step": 3707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261280192142857e-05,
+      "loss": 0.9727,
+      "step": 3708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826022268653477e-05,
+      "loss": 0.9297,
+      "step": 3709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259164890071485e-05,
+      "loss": 0.9688,
+      "step": 3710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258106802790255e-05,
+      "loss": 0.9375,
+      "step": 3711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257048424728333e-05,
+      "loss": 0.8477,
+      "step": 3712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825598975592299e-05,
+      "loss": 0.7812,
+      "step": 3713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254930796411496e-05,
+      "loss": 1.0312,
+      "step": 3714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253871546231147e-05,
+      "loss": 0.8906,
+      "step": 3715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825281200541923e-05,
+      "loss": 0.9688,
+      "step": 3716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251752174013065e-05,
+      "loss": 0.9219,
+      "step": 3717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250692052049967e-05,
+      "loss": 0.9453,
+      "step": 3718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824963163956726e-05,
+      "loss": 0.8984,
+      "step": 3719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824857093660229e-05,
+      "loss": 0.9297,
+      "step": 3720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247509943192395e-05,
+      "loss": 0.8672,
+      "step": 3721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246448659374946e-05,
+      "loss": 1.0312,
+      "step": 3722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245387085187304e-05,
+      "loss": 0.8242,
+      "step": 3723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824432522066685e-05,
+      "loss": 0.9844,
+      "step": 3724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824326306585098e-05,
+      "loss": 1.0391,
+      "step": 3725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242200620777087e-05,
+      "loss": 0.8828,
+      "step": 3726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241137885482587e-05,
+      "loss": 0.9727,
+      "step": 3727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82400748600049e-05,
+      "loss": 0.8906,
+      "step": 3728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239011544381447e-05,
+      "loss": 0.8711,
+      "step": 3729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823794793864968e-05,
+      "loss": 0.8516,
+      "step": 3730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236884042847048e-05,
+      "loss": 0.8594,
+      "step": 3731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823581985701101e-05,
+      "loss": 0.8438,
+      "step": 3732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823475538117904e-05,
+      "loss": 0.9688,
+      "step": 3733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233690615388617e-05,
+      "loss": 0.9336,
+      "step": 3734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232625559677234e-05,
+      "loss": 0.9453,
+      "step": 3735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231560214082398e-05,
+      "loss": 1.0078,
+      "step": 3736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230494578641615e-05,
+      "loss": 0.9648,
+      "step": 3737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822942865339241e-05,
+      "loss": 0.8828,
+      "step": 3738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228362438372313e-05,
+      "loss": 0.9609,
+      "step": 3739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822729593361887e-05,
+      "loss": 0.918,
+      "step": 3740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822622913916964e-05,
+      "loss": 0.8867,
+      "step": 3741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822516205506218e-05,
+      "loss": 0.9492,
+      "step": 3742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822409468133406e-05,
+      "loss": 0.8945,
+      "step": 3743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822302701802287e-05,
+      "loss": 0.8633,
+      "step": 3744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221959065166203e-05,
+      "loss": 0.918,
+      "step": 3745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822089082280166e-05,
+      "loss": 0.9805,
+      "step": 3746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821982229096686e-05,
+      "loss": 0.9102,
+      "step": 3747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218753469699428e-05,
+      "loss": 0.8789,
+      "step": 3748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217684359036995e-05,
+      "loss": 0.8984,
+      "step": 3749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821661495901721e-05,
+      "loss": 0.9688,
+      "step": 3750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821554526967772e-05,
+      "loss": 0.9922,
+      "step": 3751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82144752910562e-05,
+      "loss": 1.0,
+      "step": 3752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821340502319032e-05,
+      "loss": 1.0469,
+      "step": 3753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821233446611777e-05,
+      "loss": 0.9375,
+      "step": 3754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211263619876246e-05,
+      "loss": 0.8477,
+      "step": 3755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210192484503448e-05,
+      "loss": 0.9648,
+      "step": 3756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209121060037095e-05,
+      "loss": 0.875,
+      "step": 3757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208049346514917e-05,
+      "loss": 0.9922,
+      "step": 3758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820697734397465e-05,
+      "loss": 0.9453,
+      "step": 3759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205905052454035e-05,
+      "loss": 0.9375,
+      "step": 3760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204832471990834e-05,
+      "loss": 0.8516,
+      "step": 3761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820375960262281e-05,
+      "loss": 0.9336,
+      "step": 3762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202686444387754e-05,
+      "loss": 1.0391,
+      "step": 3763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201612997323438e-05,
+      "loss": 0.8867,
+      "step": 3764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200539261467663e-05,
+      "loss": 0.9062,
+      "step": 3765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819946523685824e-05,
+      "loss": 0.8711,
+      "step": 3766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198390923532986e-05,
+      "loss": 1.0234,
+      "step": 3767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197316321529732e-05,
+      "loss": 1.0547,
+      "step": 3768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819624143088631e-05,
+      "loss": 0.8125,
+      "step": 3769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195166251640575e-05,
+      "loss": 0.8711,
+      "step": 3770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194090783830383e-05,
+      "loss": 0.9062,
+      "step": 3771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81930150274936e-05,
+      "loss": 0.9219,
+      "step": 3772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819193898266811e-05,
+      "loss": 1.0234,
+      "step": 3773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81908626493918e-05,
+      "loss": 0.957,
+      "step": 3774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818978602770257e-05,
+      "loss": 0.8125,
+      "step": 3775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188709117638327e-05,
+      "loss": 0.8945,
+      "step": 3776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187631919236993e-05,
+      "loss": 0.9531,
+      "step": 3777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81865544325365e-05,
+      "loss": 0.9297,
+      "step": 3778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185476657574782e-05,
+      "loss": 1.0391,
+      "step": 3779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184398594389795e-05,
+      "loss": 0.9727,
+      "step": 3780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183320243019496e-05,
+      "loss": 0.9219,
+      "step": 3781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182241603501856e-05,
+      "loss": 1.1172,
+      "step": 3782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818116267587486e-05,
+      "loss": 1.0234,
+      "step": 3783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818008346017649e-05,
+      "loss": 1.0469,
+      "step": 3784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817900395644475e-05,
+      "loss": 0.8984,
+      "step": 3785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177924164717655e-05,
+      "loss": 0.832,
+      "step": 3786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176844085033223e-05,
+      "loss": 1.1094,
+      "step": 3787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8175763717429488e-05,
+      "loss": 0.8203,
+      "step": 3788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174683061944486e-05,
+      "loss": 0.832,
+      "step": 3789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173602118616272e-05,
+      "loss": 1.0078,
+      "step": 3790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172520887482908e-05,
+      "loss": 0.9219,
+      "step": 3791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171439368582464e-05,
+      "loss": 0.8477,
+      "step": 3792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170357561953025e-05,
+      "loss": 0.8828,
+      "step": 3793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169275467632678e-05,
+      "loss": 0.9531,
+      "step": 3794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168193085659535e-05,
+      "loss": 0.8125,
+      "step": 3795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167110416071696e-05,
+      "loss": 1.0078,
+      "step": 3796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816602745890729e-05,
+      "loss": 0.8438,
+      "step": 3797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816494421420445e-05,
+      "loss": 0.7461,
+      "step": 3798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163860682001318e-05,
+      "loss": 0.9414,
+      "step": 3799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162776862336047e-05,
+      "loss": 0.9258,
+      "step": 3800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161692755246794e-05,
+      "loss": 0.9805,
+      "step": 3801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160608360771744e-05,
+      "loss": 0.8359,
+      "step": 3802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159523678949075e-05,
+      "loss": 0.9531,
+      "step": 3803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158438709816974e-05,
+      "loss": 1.1016,
+      "step": 3804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815735345341365e-05,
+      "loss": 0.8398,
+      "step": 3805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815626790977732e-05,
+      "loss": 1.0156,
+      "step": 3806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155182078946197e-05,
+      "loss": 0.918,
+      "step": 3807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154095960958525e-05,
+      "loss": 0.9766,
+      "step": 3808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153009555852543e-05,
+      "loss": 1.0156,
+      "step": 3809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151922863666508e-05,
+      "loss": 1.0078,
+      "step": 3810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150835884438686e-05,
+      "loss": 0.8281,
+      "step": 3811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814974861820734e-05,
+      "loss": 0.9023,
+      "step": 3812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148661065010766e-05,
+      "loss": 0.918,
+      "step": 3813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8147573224887253e-05,
+      "loss": 0.8555,
+      "step": 3814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814648509787511e-05,
+      "loss": 0.9492,
+      "step": 3815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145396684012647e-05,
+      "loss": 0.9648,
+      "step": 3816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144307983338188e-05,
+      "loss": 0.8828,
+      "step": 3817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143218995890072e-05,
+      "loss": 0.8281,
+      "step": 3818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142129721706643e-05,
+      "loss": 0.9922,
+      "step": 3819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141040160826255e-05,
+      "loss": 0.957,
+      "step": 3820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813995031328727e-05,
+      "loss": 0.8945,
+      "step": 3821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138860179128072e-05,
+      "loss": 0.9922,
+      "step": 3822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137769758387036e-05,
+      "loss": 0.9688,
+      "step": 3823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136679051102563e-05,
+      "loss": 0.8672,
+      "step": 3824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135588057313058e-05,
+      "loss": 1.1016,
+      "step": 3825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134496777056936e-05,
+      "loss": 0.8438,
+      "step": 3826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813340521037262e-05,
+      "loss": 0.9023,
+      "step": 3827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132313357298553e-05,
+      "loss": 0.9844,
+      "step": 3828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131221217873175e-05,
+      "loss": 1.0625,
+      "step": 3829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130128792134947e-05,
+      "loss": 0.8398,
+      "step": 3830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129036080122328e-05,
+      "loss": 0.9336,
+      "step": 3831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127943081873795e-05,
+      "loss": 0.8867,
+      "step": 3832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812684979742784e-05,
+      "loss": 0.9609,
+      "step": 3833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125756226822956e-05,
+      "loss": 1.0,
+      "step": 3834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124662370097645e-05,
+      "loss": 0.9375,
+      "step": 3835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123568227290432e-05,
+      "loss": 0.9961,
+      "step": 3836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122473798439837e-05,
+      "loss": 0.8438,
+      "step": 3837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81213790835844e-05,
+      "loss": 0.8398,
+      "step": 3838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120284082762664e-05,
+      "loss": 0.918,
+      "step": 3839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119188796013186e-05,
+      "loss": 0.8906,
+      "step": 3840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811809322337454e-05,
+      "loss": 0.9062,
+      "step": 3841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116997364885295e-05,
+      "loss": 0.8828,
+      "step": 3842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811590122058404e-05,
+      "loss": 0.9141,
+      "step": 3843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114804790509367e-05,
+      "loss": 0.8008,
+      "step": 3844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8113708074699894e-05,
+      "loss": 0.9141,
+      "step": 3845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811261107319423e-05,
+      "loss": 0.9922,
+      "step": 3846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111513786031006e-05,
+      "loss": 0.8516,
+      "step": 3847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110416213248858e-05,
+      "loss": 1.0,
+      "step": 3848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810931835488643e-05,
+      "loss": 0.875,
+      "step": 3849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108220210982384e-05,
+      "loss": 0.918,
+      "step": 3850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107121781575385e-05,
+      "loss": 0.8945,
+      "step": 3851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810602306670411e-05,
+      "loss": 0.8945,
+      "step": 3852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104924066407247e-05,
+      "loss": 0.8672,
+      "step": 3853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103824780723495e-05,
+      "loss": 0.9922,
+      "step": 3854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102725209691558e-05,
+      "loss": 0.8789,
+      "step": 3855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810162535335016e-05,
+      "loss": 0.918,
+      "step": 3856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810052521173802e-05,
+      "loss": 0.9141,
+      "step": 3857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099424784893882e-05,
+      "loss": 0.9023,
+      "step": 3858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809832407285649e-05,
+      "loss": 0.9141,
+      "step": 3859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097223075664606e-05,
+      "loss": 1.0781,
+      "step": 3860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096121793356996e-05,
+      "loss": 0.8984,
+      "step": 3861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8095020225972436e-05,
+      "loss": 0.8984,
+      "step": 3862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093918373549714e-05,
+      "loss": 0.9258,
+      "step": 3863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809281623612763e-05,
+      "loss": 0.8867,
+      "step": 3864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809171381374499e-05,
+      "loss": 0.957,
+      "step": 3865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090611106440616e-05,
+      "loss": 0.9062,
+      "step": 3866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089508114253332e-05,
+      "loss": 0.8516,
+      "step": 3867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088404837221977e-05,
+      "loss": 0.8828,
+      "step": 3868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087301275385395e-05,
+      "loss": 0.8164,
+      "step": 3869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086197428782455e-05,
+      "loss": 0.918,
+      "step": 3870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085093297452013e-05,
+      "loss": 0.9141,
+      "step": 3871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083988881432953e-05,
+      "loss": 1.0781,
+      "step": 3872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808288418076417e-05,
+      "loss": 0.8281,
+      "step": 3873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081779195484548e-05,
+      "loss": 0.9336,
+      "step": 3874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080673925633004e-05,
+      "loss": 0.793,
+      "step": 3875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079568371248454e-05,
+      "loss": 1.0234,
+      "step": 3876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078462532369825e-05,
+      "loss": 0.9062,
+      "step": 3877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077356409036063e-05,
+      "loss": 0.8828,
+      "step": 3878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807625000128611e-05,
+      "loss": 0.9375,
+      "step": 3879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807514330915892e-05,
+      "loss": 0.8945,
+      "step": 3880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074036332693466e-05,
+      "loss": 0.8477,
+      "step": 3881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072929071928727e-05,
+      "loss": 0.9766,
+      "step": 3882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8071821526903695e-05,
+      "loss": 0.7812,
+      "step": 3883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807071369765736e-05,
+      "loss": 0.8203,
+      "step": 3884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806960558422874e-05,
+      "loss": 0.8555,
+      "step": 3885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068497186656842e-05,
+      "loss": 0.9961,
+      "step": 3886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80673885049807e-05,
+      "loss": 1.0234,
+      "step": 3887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066279539239358e-05,
+      "loss": 0.875,
+      "step": 3888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065170289471855e-05,
+      "loss": 0.9102,
+      "step": 3889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064060755717255e-05,
+      "loss": 0.9297,
+      "step": 3890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062950938014623e-05,
+      "loss": 0.9648,
+      "step": 3891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061840836403043e-05,
+      "loss": 0.957,
+      "step": 3892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060730450921595e-05,
+      "loss": 1.0156,
+      "step": 3893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059619781609384e-05,
+      "loss": 1.0156,
+      "step": 3894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805850882850552e-05,
+      "loss": 0.8984,
+      "step": 3895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057397591649115e-05,
+      "loss": 0.9336,
+      "step": 3896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80562860710793e-05,
+      "loss": 0.7461,
+      "step": 3897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055174266835214e-05,
+      "loss": 0.875,
+      "step": 3898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054062178956005e-05,
+      "loss": 0.8672,
+      "step": 3899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052949807480827e-05,
+      "loss": 0.8711,
+      "step": 3900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051837152448858e-05,
+      "loss": 0.9805,
+      "step": 3901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805072421389927e-05,
+      "loss": 0.9531,
+      "step": 3902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049610991871253e-05,
+      "loss": 0.9922,
+      "step": 3903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048497486404e-05,
+      "loss": 0.9844,
+      "step": 3904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047383697536727e-05,
+      "loss": 1.0312,
+      "step": 3905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804626962530865e-05,
+      "loss": 0.9883,
+      "step": 3906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045155269758992e-05,
+      "loss": 0.9102,
+      "step": 3907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044040630927e-05,
+      "loss": 0.9414,
+      "step": 3908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042925708851914e-05,
+      "loss": 0.9883,
+      "step": 3909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041810503572998e-05,
+      "loss": 0.9492,
+      "step": 3910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040695015129514e-05,
+      "loss": 0.8867,
+      "step": 3911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039579243560748e-05,
+      "loss": 0.8945,
+      "step": 3912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803846318890598e-05,
+      "loss": 0.8867,
+      "step": 3913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037346851204517e-05,
+      "loss": 0.8438,
+      "step": 3914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036230230495657e-05,
+      "loss": 0.9531,
+      "step": 3915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035113326818723e-05,
+      "loss": 0.9531,
+      "step": 3916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033996140213043e-05,
+      "loss": 1.0859,
+      "step": 3917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803287867071796e-05,
+      "loss": 0.9531,
+      "step": 3918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031760918372806e-05,
+      "loss": 1.0469,
+      "step": 3919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030642883216956e-05,
+      "loss": 0.9336,
+      "step": 3920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029524565289768e-05,
+      "loss": 0.8438,
+      "step": 3921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028405964630623e-05,
+      "loss": 0.9492,
+      "step": 3922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027287081278914e-05,
+      "loss": 0.9336,
+      "step": 3923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026167915274025e-05,
+      "loss": 0.8672,
+      "step": 3924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025048466655373e-05,
+      "loss": 1.0625,
+      "step": 3925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023928735462376e-05,
+      "loss": 0.918,
+      "step": 3926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022808721734458e-05,
+      "loss": 0.9297,
+      "step": 3927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021688425511058e-05,
+      "loss": 0.8789,
+      "step": 3928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802056784683162e-05,
+      "loss": 0.9023,
+      "step": 3929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801944698573561e-05,
+      "loss": 0.9844,
+      "step": 3930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018325842262486e-05,
+      "loss": 0.9258,
+      "step": 3931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801720441645173e-05,
+      "loss": 0.9375,
+      "step": 3932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016082708342824e-05,
+      "loss": 0.9258,
+      "step": 3933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014960717975277e-05,
+      "loss": 0.957,
+      "step": 3934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013838445388583e-05,
+      "loss": 0.9414,
+      "step": 3935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012715890622264e-05,
+      "loss": 0.9102,
+      "step": 3936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011593053715848e-05,
+      "loss": 0.918,
+      "step": 3937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801046993470887e-05,
+      "loss": 0.9883,
+      "step": 3938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009346533640877e-05,
+      "loss": 0.9141,
+      "step": 3939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800822285055143e-05,
+      "loss": 0.8789,
+      "step": 3940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007098885480085e-05,
+      "loss": 0.9023,
+      "step": 3941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800597463846643e-05,
+      "loss": 0.793,
+      "step": 3942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800485010955005e-05,
+      "loss": 0.9844,
+      "step": 3943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800372529877053e-05,
+      "loss": 0.8672,
+      "step": 3944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800260020616749e-05,
+      "loss": 0.7773,
+      "step": 3945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800147483178054e-05,
+      "loss": 0.9062,
+      "step": 3946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000349175649305e-05,
+      "loss": 0.9141,
+      "step": 3947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999223237813423e-05,
+      "loss": 1.0156,
+      "step": 3948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998097018312542e-05,
+      "loss": 1.0391,
+      "step": 3949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996970517186315e-05,
+      "loss": 0.9844,
+      "step": 3950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799584373447441e-05,
+      "loss": 1.1016,
+      "step": 3951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79947166702165e-05,
+      "loss": 0.9688,
+      "step": 3952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799358932445227e-05,
+      "loss": 0.957,
+      "step": 3953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992461697221422e-05,
+      "loss": 0.9922,
+      "step": 3954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991333788563657e-05,
+      "loss": 1.0938,
+      "step": 3955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799020559851869e-05,
+      "loss": 1.0,
+      "step": 3956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798907712712625e-05,
+      "loss": 0.9375,
+      "step": 3957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798794837442606e-05,
+      "loss": 0.9766,
+      "step": 3958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986819340457884e-05,
+      "loss": 0.8438,
+      "step": 3959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985690025261466e-05,
+      "loss": 0.9805,
+      "step": 3960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798456042887657e-05,
+      "loss": 0.8594,
+      "step": 3961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798343055134297e-05,
+      "loss": 0.957,
+      "step": 3962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798230039270046e-05,
+      "loss": 0.9375,
+      "step": 3963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981169952988826e-05,
+      "loss": 0.957,
+      "step": 3964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980039232247873e-05,
+      "loss": 0.8008,
+      "step": 3965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978908230517418e-05,
+      "loss": 0.8711,
+      "step": 3966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977776947837288e-05,
+      "loss": 0.8711,
+      "step": 3967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797664538424731e-05,
+      "loss": 0.9375,
+      "step": 3968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975513539787334e-05,
+      "loss": 1.0078,
+      "step": 3969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974381414497216e-05,
+      "loss": 0.9219,
+      "step": 3970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797324900841681e-05,
+      "loss": 0.7656,
+      "step": 3971
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972116321585997e-05,
+      "loss": 0.9648,
+      "step": 3972
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970983354044662e-05,
+      "loss": 0.8398,
+      "step": 3973
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969850105832693e-05,
+      "loss": 1.0469,
+      "step": 3974
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968716576989997e-05,
+      "loss": 0.8828,
+      "step": 3975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967582767556488e-05,
+      "loss": 0.8516,
+      "step": 3976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796644867757209e-05,
+      "loss": 1.0234,
+      "step": 3977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796531430707673e-05,
+      "loss": 0.8984,
+      "step": 3978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964179656110358e-05,
+      "loss": 0.8828,
+      "step": 3979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796304472471292e-05,
+      "loss": 0.8359,
+      "step": 3980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961909512924382e-05,
+      "loss": 0.8789,
+      "step": 3981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796077402078472e-05,
+      "loss": 0.9062,
+      "step": 3982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795963824833391e-05,
+      "loss": 0.8789,
+      "step": 3983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958502195611952e-05,
+      "loss": 0.9727,
+      "step": 3984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795736586265884e-05,
+      "loss": 0.8125,
+      "step": 3985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795622924951459e-05,
+      "loss": 1.0703,
+      "step": 3986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955092356219227e-05,
+      "loss": 0.8633,
+      "step": 3987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953955182812778e-05,
+      "loss": 1.1094,
+      "step": 3988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952817729335285e-05,
+      "loss": 1.0234,
+      "step": 3989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79516799958268e-05,
+      "loss": 0.9062,
+      "step": 3990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950541982327388e-05,
+      "loss": 0.8789,
+      "step": 3991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949403688877116e-05,
+      "loss": 0.9023,
+      "step": 3992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948265115516066e-05,
+      "loss": 0.9414,
+      "step": 3993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794712626228433e-05,
+      "loss": 1.0391,
+      "step": 3994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7945987129222006e-05,
+      "loss": 0.9023,
+      "step": 3995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944847716369208e-05,
+      "loss": 0.957,
+      "step": 3996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943708023766055e-05,
+      "loss": 0.8398,
+      "step": 3997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942568051452677e-05,
+      "loss": 1.0156,
+      "step": 3998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941427799469213e-05,
+      "loss": 1.0,
+      "step": 3999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940287267855815e-05,
+      "loss": 0.9297,
+      "step": 4000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939146456652645e-05,
+      "loss": 1.0703,
+      "step": 4001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793800536589987e-05,
+      "loss": 0.9297,
+      "step": 4002
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936863995637668e-05,
+      "loss": 1.0469,
+      "step": 4003
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793572234590623e-05,
+      "loss": 0.8984,
+      "step": 4004
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934580416745758e-05,
+      "loss": 0.918,
+      "step": 4005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7933438208196457e-05,
+      "loss": 0.9219,
+      "step": 4006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793229572029855e-05,
+      "loss": 0.8984,
+      "step": 4007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931152953092257e-05,
+      "loss": 0.9297,
+      "step": 4008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793000990661783e-05,
+      "loss": 0.8633,
+      "step": 4009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928866580915508e-05,
+      "loss": 0.9805,
+      "step": 4010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792772297602555e-05,
+      "loss": 0.8906,
+      "step": 4011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926579091988227e-05,
+      "loss": 0.9297,
+      "step": 4012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925434928843813e-05,
+      "loss": 0.7266,
+      "step": 4013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79242904866326e-05,
+      "loss": 1.0234,
+      "step": 4014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923145765394886e-05,
+      "loss": 0.9102,
+      "step": 4015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922000765170976e-05,
+      "loss": 0.9688,
+      "step": 4016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920855486001186e-05,
+      "loss": 0.875,
+      "step": 4017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919709927925842e-05,
+      "loss": 0.8203,
+      "step": 4018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918564090985287e-05,
+      "loss": 0.8594,
+      "step": 4019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917417975219863e-05,
+      "loss": 0.9258,
+      "step": 4020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916271580669926e-05,
+      "loss": 0.8828,
+      "step": 4021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915124907375845e-05,
+      "loss": 0.8633,
+      "step": 4022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7913977955377997e-05,
+      "loss": 1.0156,
+      "step": 4023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912830724716766e-05,
+      "loss": 1.0391,
+      "step": 4024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911683215432547e-05,
+      "loss": 0.8945,
+      "step": 4025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791053542756574e-05,
+      "loss": 0.9531,
+      "step": 4026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909387361156776e-05,
+      "loss": 1.0703,
+      "step": 4027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908239016246067e-05,
+      "loss": 0.9531,
+      "step": 4028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907090392874057e-05,
+      "loss": 0.9453,
+      "step": 4029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790594149108118e-05,
+      "loss": 0.9805,
+      "step": 4030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79047923109079e-05,
+      "loss": 0.9375,
+      "step": 4031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790364285239467e-05,
+      "loss": 0.9375,
+      "step": 4032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790249311558198e-05,
+      "loss": 0.8555,
+      "step": 4033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901343100510304e-05,
+      "loss": 0.9062,
+      "step": 4034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790019280722014e-05,
+      "loss": 0.8555,
+      "step": 4035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899042235751984e-05,
+      "loss": 0.8398,
+      "step": 4036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897891386146358e-05,
+      "loss": 0.9102,
+      "step": 4037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789674025844378e-05,
+      "loss": 0.8906,
+      "step": 4038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895588852684787e-05,
+      "loss": 0.9883,
+      "step": 4039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789443716890992e-05,
+      "loss": 0.8516,
+      "step": 4040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893285207159728e-05,
+      "loss": 0.8555,
+      "step": 4041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789213296747478e-05,
+      "loss": 0.9297,
+      "step": 4042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789098044989564e-05,
+      "loss": 0.8164,
+      "step": 4043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889827654462903e-05,
+      "loss": 0.9375,
+      "step": 4044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888674581217148e-05,
+      "loss": 0.8125,
+      "step": 4045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788752123019898e-05,
+      "loss": 0.9375,
+      "step": 4046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886367601449014e-05,
+      "loss": 0.8867,
+      "step": 4047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885213695007865e-05,
+      "loss": 0.7539,
+      "step": 4048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884059510916167e-05,
+      "loss": 1.0469,
+      "step": 4049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882905049214564e-05,
+      "loss": 0.9609,
+      "step": 4050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78817503099437e-05,
+      "loss": 0.9102,
+      "step": 4051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788059529314424e-05,
+      "loss": 0.9297,
+      "step": 4052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787943999885685e-05,
+      "loss": 0.9766,
+      "step": 4053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878284427122216e-05,
+      "loss": 0.9648,
+      "step": 4054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787712857798102e-05,
+      "loss": 0.9727,
+      "step": 4055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7875972451473965e-05,
+      "loss": 0.9336,
+      "step": 4056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787481604764176e-05,
+      "loss": 0.8125,
+      "step": 4057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787365936652512e-05,
+      "loss": 0.9062,
+      "step": 4058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787250240816478e-05,
+      "loss": 1.125,
+      "step": 4059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871345172601475e-05,
+      "loss": 0.9062,
+      "step": 4060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870187659875954e-05,
+      "loss": 0.9375,
+      "step": 4061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869029870028974e-05,
+      "loss": 0.8359,
+      "step": 4062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78678718031013e-05,
+      "loss": 0.8828,
+      "step": 4063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786671345913371e-05,
+      "loss": 0.9648,
+      "step": 4064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865554838166997e-05,
+      "loss": 0.9219,
+      "step": 4065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7864395940241955e-05,
+      "loss": 0.9453,
+      "step": 4066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786323676539939e-05,
+      "loss": 0.9375,
+      "step": 4067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862077313680112e-05,
+      "loss": 0.875,
+      "step": 4068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860917585124956e-05,
+      "loss": 0.8203,
+      "step": 4069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859757579774754e-05,
+      "loss": 0.8906,
+      "step": 4070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858597297670353e-05,
+      "loss": 0.9141,
+      "step": 4071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857436738852608e-05,
+      "loss": 0.7383,
+      "step": 4072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856275903362382e-05,
+      "loss": 0.9453,
+      "step": 4073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785511479124055e-05,
+      "loss": 1.1094,
+      "step": 4074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853953402528e-05,
+      "loss": 0.9609,
+      "step": 4075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852791737265624e-05,
+      "loss": 0.875,
+      "step": 4076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851629795494326e-05,
+      "loss": 0.7656,
+      "step": 4077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850467577255018e-05,
+      "loss": 0.9375,
+      "step": 4078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849305082588632e-05,
+      "loss": 0.9648,
+      "step": 4079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784814231153609e-05,
+      "loss": 0.9297,
+      "step": 4080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784697926413834e-05,
+      "loss": 1.0781,
+      "step": 4081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845815940436336e-05,
+      "loss": 0.8633,
+      "step": 4082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844652340471037e-05,
+      "loss": 0.8945,
+      "step": 4083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843488464283417e-05,
+      "loss": 0.9141,
+      "step": 4084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842324311914458e-05,
+      "loss": 0.8047,
+      "step": 4085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784115988340515e-05,
+      "loss": 0.9609,
+      "step": 4086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78399951787965e-05,
+      "loss": 0.9805,
+      "step": 4087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7838830198129512e-05,
+      "loss": 0.8398,
+      "step": 4088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783766494144521e-05,
+      "loss": 1.0703,
+      "step": 4089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836499408784624e-05,
+      "loss": 0.9102,
+      "step": 4090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835333600188795e-05,
+      "loss": 0.8867,
+      "step": 4091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834167515698774e-05,
+      "loss": 0.9922,
+      "step": 4092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833001155355613e-05,
+      "loss": 1.0156,
+      "step": 4093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831834519200392e-05,
+      "loss": 0.9961,
+      "step": 4094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830667607274185e-05,
+      "loss": 0.9297,
+      "step": 4095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782950041961808e-05,
+      "loss": 0.9688,
+      "step": 4096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828332956273175e-05,
+      "loss": 1.0156,
+      "step": 4097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827165217280582e-05,
+      "loss": 1.0,
+      "step": 4098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825997202681415e-05,
+      "loss": 0.8906,
+      "step": 4099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824828912516806e-05,
+      "loss": 0.8203,
+      "step": 4100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823660346827883e-05,
+      "loss": 0.9805,
+      "step": 4101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782249150565581e-05,
+      "loss": 0.8906,
+      "step": 4102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821322389041723e-05,
+      "loss": 0.9336,
+      "step": 4103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820152997026807e-05,
+      "loss": 0.9102,
+      "step": 4104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818983329652223e-05,
+      "loss": 0.8672,
+      "step": 4105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817813386959165e-05,
+      "loss": 0.8281,
+      "step": 4106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781664316898883e-05,
+      "loss": 0.7734,
+      "step": 4107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815472675782417e-05,
+      "loss": 0.9336,
+      "step": 4108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814301907381144e-05,
+      "loss": 0.8672,
+      "step": 4109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781313086382624e-05,
+      "loss": 0.9375,
+      "step": 4110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781195954515893e-05,
+      "loss": 0.8945,
+      "step": 4111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810787951420464e-05,
+      "loss": 0.9609,
+      "step": 4112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809616082652096e-05,
+      "loss": 0.8906,
+      "step": 4113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808443938895085e-05,
+      "loss": 0.9766,
+      "step": 4114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807271520190704e-05,
+      "loss": 0.8984,
+      "step": 4115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806098826580246e-05,
+      "loss": 0.8594,
+      "step": 4116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780492585810499e-05,
+      "loss": 0.9102,
+      "step": 4117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803752614806247e-05,
+      "loss": 0.9531,
+      "step": 4118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802579096725322e-05,
+      "loss": 0.918,
+      "step": 4119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801405303903545e-05,
+      "loss": 0.875,
+      "step": 4120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780023123638224e-05,
+      "loss": 0.9727,
+      "step": 4121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799056894202745e-05,
+      "loss": 0.8672,
+      "step": 4122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797882277406423e-05,
+      "loss": 0.8477,
+      "step": 4123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779670738603462e-05,
+      "loss": 0.9883,
+      "step": 4124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795532220128715e-05,
+      "loss": 1.0547,
+      "step": 4125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794356779730084e-05,
+      "loss": 1.0,
+      "step": 4126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793181064880114e-05,
+      "loss": 0.9844,
+      "step": 4127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779200507562021e-05,
+      "loss": 0.9961,
+      "step": 4128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790828811991773e-05,
+      "loss": 0.8281,
+      "step": 4129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789652274036225e-05,
+      "loss": 0.9844,
+      "step": 4130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788475461794993e-05,
+      "loss": 0.9258,
+      "step": 4131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787298375309517e-05,
+      "loss": 0.8594,
+      "step": 4132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786121014621242e-05,
+      "loss": 0.8594,
+      "step": 4133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784943379771617e-05,
+      "loss": 0.8516,
+      "step": 4134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7783765470802123e-05,
+      "loss": 0.7969,
+      "step": 4135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778258728775423e-05,
+      "loss": 0.9648,
+      "step": 4136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781408830669417e-05,
+      "loss": 0.8555,
+      "step": 4137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780230099589186e-05,
+      "loss": 1.0234,
+      "step": 4138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777905109455504e-05,
+      "loss": 1.0312,
+      "step": 4139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777871815608495e-05,
+      "loss": 1.0234,
+      "step": 4140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776692262791074e-05,
+      "loss": 0.9688,
+      "step": 4141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775512436144314e-05,
+      "loss": 0.9297,
+      "step": 4142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774332335709753e-05,
+      "loss": 0.9805,
+      "step": 4143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773151961528946e-05,
+      "loss": 0.9062,
+      "step": 4144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777197131364346e-05,
+      "loss": 1.0,
+      "step": 4145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770790392094864e-05,
+      "loss": 0.9648,
+      "step": 4146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769609196924736e-05,
+      "loss": 1.0234,
+      "step": 4147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7768427728174677e-05,
+      "loss": 0.9922,
+      "step": 4148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776724598588628e-05,
+      "loss": 0.7852,
+      "step": 4149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776606397010116e-05,
+      "loss": 0.9141,
+      "step": 4150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7764881680860937e-05,
+      "loss": 0.9648,
+      "step": 4151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763699118207244e-05,
+      "loss": 0.9492,
+      "step": 4152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762516282181713e-05,
+      "loss": 1.0156,
+      "step": 4153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761333172826e-05,
+      "loss": 0.9336,
+      "step": 4154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760149790181763e-05,
+      "loss": 1.0078,
+      "step": 4155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758966134290668e-05,
+      "loss": 1.0156,
+      "step": 4156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77577822051944e-05,
+      "loss": 0.9297,
+      "step": 4157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775659800293464e-05,
+      "loss": 0.8594,
+      "step": 4158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755413527553087e-05,
+      "loss": 0.8672,
+      "step": 4159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754228779091452e-05,
+      "loss": 0.8828,
+      "step": 4160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753043757591447e-05,
+      "loss": 1.0469,
+      "step": 4161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751858463094803e-05,
+      "loss": 0.8867,
+      "step": 4162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775067289564325e-05,
+      "loss": 0.8867,
+      "step": 4163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7749487055278544e-05,
+      "loss": 0.9062,
+      "step": 4164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748300942042428e-05,
+      "loss": 0.9883,
+      "step": 4165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774711455597667e-05,
+      "loss": 0.8945,
+      "step": 4166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745927897123053e-05,
+      "loss": 0.8477,
+      "step": 4167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744740965523352e-05,
+      "loss": 0.9102,
+      "step": 4168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743553761219364e-05,
+      "loss": 0.8984,
+      "step": 4169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742366284252895e-05,
+      "loss": 0.9414,
+      "step": 4170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741178534665752e-05,
+      "loss": 0.9844,
+      "step": 4171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773999051249976e-05,
+      "loss": 1.0547,
+      "step": 4172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738802217796754e-05,
+      "loss": 0.8633,
+      "step": 4173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773761365059857e-05,
+      "loss": 1.0312,
+      "step": 4174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736424810947068e-05,
+      "loss": 1.0156,
+      "step": 4175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735235698884098e-05,
+      "loss": 0.9023,
+      "step": 4176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734046314451535e-05,
+      "loss": 1.0,
+      "step": 4177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732856657691262e-05,
+      "loss": 0.8398,
+      "step": 4178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773166672864517e-05,
+      "loss": 1.0156,
+      "step": 4179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773047652735515e-05,
+      "loss": 0.9023,
+      "step": 4180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729286053863118e-05,
+      "loss": 0.8086,
+      "step": 4181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772809530821099e-05,
+      "loss": 0.9648,
+      "step": 4182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772690429044069e-05,
+      "loss": 0.8867,
+      "step": 4183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725713000594167e-05,
+      "loss": 0.9375,
+      "step": 4184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7724521438713356e-05,
+      "loss": 1.0156,
+      "step": 4185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772332960484022e-05,
+      "loss": 0.832,
+      "step": 4186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722137499016724e-05,
+      "loss": 0.8672,
+      "step": 4187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720945121284843e-05,
+      "loss": 0.8555,
+      "step": 4188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771975247168656e-05,
+      "loss": 0.8359,
+      "step": 4189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771855955026388e-05,
+      "loss": 0.9492,
+      "step": 4190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717366357058795e-05,
+      "loss": 0.9805,
+      "step": 4191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716172892113326e-05,
+      "loss": 0.9258,
+      "step": 4192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77149791554695e-05,
+      "loss": 0.9531,
+      "step": 4193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771378514716934e-05,
+      "loss": 0.8438,
+      "step": 4194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712590867254903e-05,
+      "loss": 1.0156,
+      "step": 4195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711396315768225e-05,
+      "loss": 0.9805,
+      "step": 4196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771020149275138e-05,
+      "loss": 0.9922,
+      "step": 4197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709006398246437e-05,
+      "loss": 0.8984,
+      "step": 4198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707811032295476e-05,
+      "loss": 0.9062,
+      "step": 4199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7706615394940586e-05,
+      "loss": 1.0391,
+      "step": 4200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705419486223868e-05,
+      "loss": 0.8438,
+      "step": 4201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704223306187433e-05,
+      "loss": 1.0078,
+      "step": 4202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703026854873404e-05,
+      "loss": 0.8906,
+      "step": 4203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77018301323239e-05,
+      "loss": 0.918,
+      "step": 4204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770063313858107e-05,
+      "loss": 1.0469,
+      "step": 4205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699435873687056e-05,
+      "loss": 0.9102,
+      "step": 4206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698238337684016e-05,
+      "loss": 0.9375,
+      "step": 4207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769704053061412e-05,
+      "loss": 0.9688,
+      "step": 4208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769584245251954e-05,
+      "loss": 0.8672,
+      "step": 4209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7694644103442468e-05,
+      "loss": 1.125,
+      "step": 4210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693445483425097e-05,
+      "loss": 0.9531,
+      "step": 4211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769224659250963e-05,
+      "loss": 0.9219,
+      "step": 4212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691047430738282e-05,
+      "loss": 0.9258,
+      "step": 4213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768984799815328e-05,
+      "loss": 0.9023,
+      "step": 4214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688648294796858e-05,
+      "loss": 0.8477,
+      "step": 4215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768744832071126e-05,
+      "loss": 0.8242,
+      "step": 4216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686248075938735e-05,
+      "loss": 0.8867,
+      "step": 4217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768504756052155e-05,
+      "loss": 0.8789,
+      "step": 4218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768384677450197e-05,
+      "loss": 0.9219,
+      "step": 4219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682645717922286e-05,
+      "loss": 0.9258,
+      "step": 4220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768144439082478e-05,
+      "loss": 0.8984,
+      "step": 4221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768024279325176e-05,
+      "loss": 0.8086,
+      "step": 4222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767904092524553e-05,
+      "loss": 0.8984,
+      "step": 4223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677838786848416e-05,
+      "loss": 0.832,
+      "step": 4224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676636378102742e-05,
+      "loss": 0.8398,
+      "step": 4225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675433699050854e-05,
+      "loss": 0.9219,
+      "step": 4226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7674230749735085e-05,
+      "loss": 0.8906,
+      "step": 4227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767302753019781e-05,
+      "loss": 0.8594,
+      "step": 4228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671824040481383e-05,
+      "loss": 0.9844,
+      "step": 4229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767062028062819e-05,
+      "loss": 0.8086,
+      "step": 4230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766941625068061e-05,
+      "loss": 0.8594,
+      "step": 4231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668211950681047e-05,
+      "loss": 0.9766,
+      "step": 4232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76670073806719e-05,
+      "loss": 0.9023,
+      "step": 4233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7665802540695582e-05,
+      "loss": 0.8398,
+      "step": 4234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664597430794525e-05,
+      "loss": 0.8789,
+      "step": 4235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663392051011156e-05,
+      "loss": 0.957,
+      "step": 4236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766218640138792e-05,
+      "loss": 0.8984,
+      "step": 4237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766098048196727e-05,
+      "loss": 0.8711,
+      "step": 4238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765977429279167e-05,
+      "loss": 1.0234,
+      "step": 4239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765856783390359e-05,
+      "loss": 0.8125,
+      "step": 4240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657361105345513e-05,
+      "loss": 0.8711,
+      "step": 4241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765615410715993e-05,
+      "loss": 0.918,
+      "step": 4242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7654946839389334e-05,
+      "loss": 0.9609,
+      "step": 4243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765373930207625e-05,
+      "loss": 0.8828,
+      "step": 4244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652531495263175e-05,
+      "loss": 0.9766,
+      "step": 4245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765132341899266e-05,
+      "loss": 0.9766,
+      "step": 4246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650115073307232e-05,
+      "loss": 0.7422,
+      "step": 4247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764890645824944e-05,
+      "loss": 0.7695,
+      "step": 4248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647697573861842e-05,
+      "loss": 0.8711,
+      "step": 4249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646488420187006e-05,
+      "loss": 0.8516,
+      "step": 4250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645278997267503e-05,
+      "loss": 0.8047,
+      "step": 4251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644069305145924e-05,
+      "loss": 1.0547,
+      "step": 4252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764285934386486e-05,
+      "loss": 0.8594,
+      "step": 4253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641649113466922e-05,
+      "loss": 1.0234,
+      "step": 4254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764043861399472e-05,
+      "loss": 0.6797,
+      "step": 4255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639227845490872e-05,
+      "loss": 0.9648,
+      "step": 4256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638016807998023e-05,
+      "loss": 0.9609,
+      "step": 4257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636805501558804e-05,
+      "loss": 0.7656,
+      "step": 4258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635593926215876e-05,
+      "loss": 0.9766,
+      "step": 4259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7634382082011898e-05,
+      "loss": 0.9414,
+      "step": 4260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7633169968989535e-05,
+      "loss": 0.9922,
+      "step": 4261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763195758719147e-05,
+      "loss": 1.1094,
+      "step": 4262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76307449366604e-05,
+      "loss": 0.8516,
+      "step": 4263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629532017439018e-05,
+      "loss": 1.0,
+      "step": 4264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628318829570033e-05,
+      "loss": 0.8398,
+      "step": 4265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627105373096163e-05,
+      "loss": 0.875,
+      "step": 4266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625891648060136e-05,
+      "loss": 0.918,
+      "step": 4267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624677654504694e-05,
+      "loss": 0.8164,
+      "step": 4268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623463392472574e-05,
+      "loss": 0.9375,
+      "step": 4269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762224886200654e-05,
+      "loss": 0.8789,
+      "step": 4270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621034063149355e-05,
+      "loss": 0.832,
+      "step": 4271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619818995943792e-05,
+      "loss": 0.8867,
+      "step": 4272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761860366043264e-05,
+      "loss": 0.8594,
+      "step": 4273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617388056658687e-05,
+      "loss": 0.8633,
+      "step": 4274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616172184664742e-05,
+      "loss": 0.9102,
+      "step": 4275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614956044493614e-05,
+      "loss": 0.8867,
+      "step": 4276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613739636188124e-05,
+      "loss": 0.8984,
+      "step": 4277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761252295979111e-05,
+      "loss": 0.8281,
+      "step": 4278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611306015345406e-05,
+      "loss": 0.957,
+      "step": 4279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610088802893867e-05,
+      "loss": 0.9414,
+      "step": 4280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760887132247935e-05,
+      "loss": 0.7891,
+      "step": 4281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607653574144724e-05,
+      "loss": 0.8594,
+      "step": 4282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760643555793287e-05,
+      "loss": 0.7695,
+      "step": 4283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760521727388668e-05,
+      "loss": 0.8594,
+      "step": 4284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603998722049045e-05,
+      "loss": 0.8828,
+      "step": 4285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760277990246287e-05,
+      "loss": 0.8828,
+      "step": 4286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601560815171083e-05,
+      "loss": 0.9961,
+      "step": 4287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76003414602166e-05,
+      "loss": 0.9141,
+      "step": 4288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759912183764236e-05,
+      "loss": 0.8047,
+      "step": 4289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597901947491305e-05,
+      "loss": 1.0469,
+      "step": 4290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7596681789806393e-05,
+      "loss": 1.0469,
+      "step": 4291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595461364630585e-05,
+      "loss": 0.9492,
+      "step": 4292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594240672006858e-05,
+      "loss": 0.8906,
+      "step": 4293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593019711978187e-05,
+      "loss": 0.9844,
+      "step": 4294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591798484587574e-05,
+      "loss": 0.832,
+      "step": 4295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590576989878008e-05,
+      "loss": 0.9453,
+      "step": 4296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758935522789251e-05,
+      "loss": 1.0234,
+      "step": 4297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588133198674094e-05,
+      "loss": 0.918,
+      "step": 4298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586910902265794e-05,
+      "loss": 0.9102,
+      "step": 4299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585688338710648e-05,
+      "loss": 0.9102,
+      "step": 4300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584465508051702e-05,
+      "loss": 0.9688,
+      "step": 4301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583242410332016e-05,
+      "loss": 0.8672,
+      "step": 4302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582019045594653e-05,
+      "loss": 0.8906,
+      "step": 4303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75807954138827e-05,
+      "loss": 0.832,
+      "step": 4304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757957151523923e-05,
+      "loss": 0.9414,
+      "step": 4305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7578347349707347e-05,
+      "loss": 0.9141,
+      "step": 4306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577122917330148e-05,
+      "loss": 0.9375,
+      "step": 4307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575898218150754e-05,
+      "loss": 0.8477,
+      "step": 4308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757467325221229e-05,
+      "loss": 0.8203,
+      "step": 4309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757344801955788e-05,
+      "loss": 0.8242,
+      "step": 4310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7572222520230676e-05,
+      "loss": 0.9492,
+      "step": 4311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7570996754273824e-05,
+      "loss": 0.875,
+      "step": 4312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756977072173049e-05,
+      "loss": 0.8672,
+      "step": 4313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568544422643837e-05,
+      "loss": 0.8516,
+      "step": 4314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756731785705705e-05,
+      "loss": 0.8789,
+      "step": 4315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756609102501332e-05,
+      "loss": 1.0156,
+      "step": 4316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756486392655584e-05,
+      "loss": 0.8125,
+      "step": 4317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563636561727823e-05,
+      "loss": 0.8477,
+      "step": 4318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562408930572485e-05,
+      "loss": 1.0312,
+      "step": 4319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561181033133048e-05,
+      "loss": 0.9688,
+      "step": 4320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559952869452757e-05,
+      "loss": 0.8789,
+      "step": 4321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558724439574847e-05,
+      "loss": 0.8945,
+      "step": 4322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.8281,
+      "step": 4323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556266781399227e-05,
+      "loss": 1.1172,
+      "step": 4324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555037553188048e-05,
+      "loss": 0.9922,
+      "step": 4325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553808058952334e-05,
+      "loss": 0.8594,
+      "step": 4326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7552578298735373e-05,
+      "loss": 0.8281,
+      "step": 4327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755134827258047e-05,
+      "loss": 0.9062,
+      "step": 4328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550117980530933e-05,
+      "loss": 0.8945,
+      "step": 4329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548887422630086e-05,
+      "loss": 0.9375,
+      "step": 4330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754765659892126e-05,
+      "loss": 0.918,
+      "step": 4331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754642550944779e-05,
+      "loss": 0.7891,
+      "step": 4332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545194154253024e-05,
+      "loss": 0.793,
+      "step": 4333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543962533380322e-05,
+      "loss": 0.9531,
+      "step": 4334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542730646873054e-05,
+      "loss": 1.0938,
+      "step": 4335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754149849477459e-05,
+      "loss": 0.8789,
+      "step": 4336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540266077128326e-05,
+      "loss": 0.8789,
+      "step": 4337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539033393977646e-05,
+      "loss": 0.8203,
+      "step": 4338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537800445365963e-05,
+      "loss": 0.9336,
+      "step": 4339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753656723133669e-05,
+      "loss": 0.8594,
+      "step": 4340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753533375193324e-05,
+      "loss": 0.8711,
+      "step": 4341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753410000719906e-05,
+      "loss": 0.8828,
+      "step": 4342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532865997177588e-05,
+      "loss": 0.9258,
+      "step": 4343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753163172191227e-05,
+      "loss": 0.8203,
+      "step": 4344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530397181446572e-05,
+      "loss": 0.9883,
+      "step": 4345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752916237582396e-05,
+      "loss": 0.9414,
+      "step": 4346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752792730508792e-05,
+      "loss": 0.8555,
+      "step": 4347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752669196928193e-05,
+      "loss": 0.9336,
+      "step": 4348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525456368449503e-05,
+      "loss": 0.8164,
+      "step": 4349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524220502634134e-05,
+      "loss": 0.8555,
+      "step": 4350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752298437187934e-05,
+      "loss": 1.0078,
+      "step": 4351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521747976228657e-05,
+      "loss": 0.8984,
+      "step": 4352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520511315725608e-05,
+      "loss": 0.8594,
+      "step": 4353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519274390413747e-05,
+      "loss": 0.9922,
+      "step": 4354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751803720033663e-05,
+      "loss": 0.9688,
+      "step": 4355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516799745537806e-05,
+      "loss": 0.8281,
+      "step": 4356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751556202606086e-05,
+      "loss": 0.9688,
+      "step": 4357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514324041949376e-05,
+      "loss": 0.8867,
+      "step": 4358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513085793246937e-05,
+      "loss": 0.8242,
+      "step": 4359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511847279997148e-05,
+      "loss": 0.8086,
+      "step": 4360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510608502243615e-05,
+      "loss": 0.9688,
+      "step": 4361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509369460029964e-05,
+      "loss": 0.9609,
+      "step": 4362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750813015339982e-05,
+      "loss": 0.9883,
+      "step": 4363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506890582396815e-05,
+      "loss": 0.918,
+      "step": 4364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505650747064605e-05,
+      "loss": 0.8867,
+      "step": 4365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504410647446847e-05,
+      "loss": 0.9805,
+      "step": 4366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75031702835872e-05,
+      "loss": 0.8672,
+      "step": 4367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501929655529344e-05,
+      "loss": 0.8594,
+      "step": 4368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750068876331696e-05,
+      "loss": 0.9102,
+      "step": 4369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499447606993746e-05,
+      "loss": 0.8633,
+      "step": 4370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498206186603402e-05,
+      "loss": 0.9805,
+      "step": 4371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496964502189643e-05,
+      "loss": 0.9258,
+      "step": 4372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749572255379618e-05,
+      "loss": 0.7969,
+      "step": 4373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494480341466762e-05,
+      "loss": 0.8828,
+      "step": 4374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493237865245116e-05,
+      "loss": 0.918,
+      "step": 4375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491995125174997e-05,
+      "loss": 0.9336,
+      "step": 4376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490752121300157e-05,
+      "loss": 0.9492,
+      "step": 4377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489508853664373e-05,
+      "loss": 1.0156,
+      "step": 4378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748826532231142e-05,
+      "loss": 0.8242,
+      "step": 4379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748702152728508e-05,
+      "loss": 1.0469,
+      "step": 4380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485777468629152e-05,
+      "loss": 1.0312,
+      "step": 4381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484533146387445e-05,
+      "loss": 0.8398,
+      "step": 4382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748328856060377e-05,
+      "loss": 0.9805,
+      "step": 4383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482043711321946e-05,
+      "loss": 0.8672,
+      "step": 4384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480798598585812e-05,
+      "loss": 1.0938,
+      "step": 4385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747955322243921e-05,
+      "loss": 0.8789,
+      "step": 4386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747830758292599e-05,
+      "loss": 0.8984,
+      "step": 4387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477061680090017e-05,
+      "loss": 0.918,
+      "step": 4388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747581551397515e-05,
+      "loss": 0.9062,
+      "step": 4389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7474569084625282e-05,
+      "loss": 0.8984,
+      "step": 4390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473322392084295e-05,
+      "loss": 1.0234,
+      "step": 4391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472075436396087e-05,
+      "loss": 0.8516,
+      "step": 4392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747082821760457e-05,
+      "loss": 0.9219,
+      "step": 4393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746958073575365e-05,
+      "loss": 0.918,
+      "step": 4394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746833299088726e-05,
+      "loss": 0.8906,
+      "step": 4395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746708498304934e-05,
+      "loss": 0.8828,
+      "step": 4396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465836712283822e-05,
+      "loss": 0.9336,
+      "step": 4397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746458817863467e-05,
+      "loss": 0.9141,
+      "step": 4398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463339382145842e-05,
+      "loss": 0.832,
+      "step": 4399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462090322861315e-05,
+      "loss": 0.9297,
+      "step": 4400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746084100082506e-05,
+      "loss": 1.0,
+      "step": 4401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459591416081076e-05,
+      "loss": 0.9375,
+      "step": 4402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745834156867336e-05,
+      "loss": 0.8945,
+      "step": 4403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457091458645926e-05,
+      "loss": 0.9219,
+      "step": 4404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455841086042784e-05,
+      "loss": 0.8906,
+      "step": 4405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454590450907967e-05,
+      "loss": 0.8984,
+      "step": 4406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745333955328551e-05,
+      "loss": 1.0391,
+      "step": 4407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452088393219456e-05,
+      "loss": 0.8906,
+      "step": 4408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450836970753867e-05,
+      "loss": 0.9141,
+      "step": 4409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449585285932806e-05,
+      "loss": 1.0078,
+      "step": 4410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448333338800342e-05,
+      "loss": 0.9727,
+      "step": 4411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447081129400562e-05,
+      "loss": 0.9492,
+      "step": 4412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744582865777756e-05,
+      "loss": 0.8672,
+      "step": 4413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744457592397543e-05,
+      "loss": 1.0156,
+      "step": 4414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744332292803829e-05,
+      "loss": 0.9102,
+      "step": 4415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442069670010253e-05,
+      "loss": 0.9766,
+      "step": 4416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440816149935455e-05,
+      "loss": 0.9062,
+      "step": 4417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439562367858034e-05,
+      "loss": 0.9492,
+      "step": 4418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7438308323822136e-05,
+      "loss": 0.9883,
+      "step": 4419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437054017871914e-05,
+      "loss": 0.9648,
+      "step": 4420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743579945005154e-05,
+      "loss": 0.832,
+      "step": 4421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434544620405183e-05,
+      "loss": 0.8555,
+      "step": 4422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433289528977034e-05,
+      "loss": 1.0703,
+      "step": 4423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743203417581128e-05,
+      "loss": 0.9844,
+      "step": 4424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430778560952132e-05,
+      "loss": 0.8828,
+      "step": 4425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429522684443797e-05,
+      "loss": 0.9375,
+      "step": 4426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428266546330495e-05,
+      "loss": 0.8516,
+      "step": 4427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742701014665646e-05,
+      "loss": 0.918,
+      "step": 4428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742575348546593e-05,
+      "loss": 0.7773,
+      "step": 4429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424496562803152e-05,
+      "loss": 1.0078,
+      "step": 4430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742323937871239e-05,
+      "loss": 1.0,
+      "step": 4431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421981933237908e-05,
+      "loss": 0.8633,
+      "step": 4432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420724226423982e-05,
+      "loss": 0.9062,
+      "step": 4433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74194662583149e-05,
+      "loss": 0.957,
+      "step": 4434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741820802895495e-05,
+      "loss": 0.9492,
+      "step": 4435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741694953838845e-05,
+      "loss": 0.9141,
+      "step": 4436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415690786659697e-05,
+      "loss": 0.9609,
+      "step": 4437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414431773813026e-05,
+      "loss": 1.0078,
+      "step": 4438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741317249989276e-05,
+      "loss": 1.0469,
+      "step": 4439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741191296494325e-05,
+      "loss": 0.8828,
+      "step": 4440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410653169008837e-05,
+      "loss": 0.8867,
+      "step": 4441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409393112133885e-05,
+      "loss": 0.9141,
+      "step": 4442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408132794362763e-05,
+      "loss": 1.0312,
+      "step": 4443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406872215739845e-05,
+      "loss": 0.8672,
+      "step": 4444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740561137630952e-05,
+      "loss": 0.9844,
+      "step": 4445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404350276116185e-05,
+      "loss": 0.8867,
+      "step": 4446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403088915204246e-05,
+      "loss": 0.918,
+      "step": 4447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401827293618114e-05,
+      "loss": 0.8477,
+      "step": 4448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7400565411402217e-05,
+      "loss": 0.8906,
+      "step": 4449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399303268600984e-05,
+      "loss": 0.9727,
+      "step": 4450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7398040865258855e-05,
+      "loss": 0.8984,
+      "step": 4451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739677820142029e-05,
+      "loss": 0.8945,
+      "step": 4452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739551527712974e-05,
+      "loss": 0.9883,
+      "step": 4453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739425209243168e-05,
+      "loss": 0.875,
+      "step": 4454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392988647370586e-05,
+      "loss": 0.7969,
+      "step": 4455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391724941990952e-05,
+      "loss": 0.9688,
+      "step": 4456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390460976337266e-05,
+      "loss": 0.9766,
+      "step": 4457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389196750454036e-05,
+      "loss": 0.8711,
+      "step": 4458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738793226438578e-05,
+      "loss": 1.0469,
+      "step": 4459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386667518177027e-05,
+      "loss": 0.9844,
+      "step": 4460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385402511872303e-05,
+      "loss": 0.9297,
+      "step": 4461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384137245516152e-05,
+      "loss": 0.8594,
+      "step": 4462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738287171915313e-05,
+      "loss": 0.8867,
+      "step": 4463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381605932827793e-05,
+      "loss": 0.9609,
+      "step": 4464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380339886584714e-05,
+      "loss": 0.9023,
+      "step": 4465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379073580468474e-05,
+      "loss": 0.9375,
+      "step": 4466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737780701452366e-05,
+      "loss": 0.7734,
+      "step": 4467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376540188794867e-05,
+      "loss": 0.8906,
+      "step": 4468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737527310332671e-05,
+      "loss": 0.9688,
+      "step": 4469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374005758163796e-05,
+      "loss": 0.8047,
+      "step": 4470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372738153350752e-05,
+      "loss": 0.9023,
+      "step": 4471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371470288932218e-05,
+      "loss": 1.0078,
+      "step": 4472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737020216495283e-05,
+      "loss": 0.9414,
+      "step": 4473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368933781457248e-05,
+      "loss": 1.0391,
+      "step": 4474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367665138490125e-05,
+      "loss": 0.832,
+      "step": 4475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366396236096142e-05,
+      "loss": 0.8906,
+      "step": 4476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365127074319973e-05,
+      "loss": 1.0703,
+      "step": 4477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7363857653206307e-05,
+      "loss": 0.9023,
+      "step": 4478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362587972799843e-05,
+      "loss": 0.9023,
+      "step": 4479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736131803314529e-05,
+      "loss": 0.8828,
+      "step": 4480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736004783428736e-05,
+      "loss": 0.9453,
+      "step": 4481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358777376270786e-05,
+      "loss": 0.8086,
+      "step": 4482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357506659140295e-05,
+      "loss": 0.8164,
+      "step": 4483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356235682940633e-05,
+      "loss": 0.9766,
+      "step": 4484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735496444771656e-05,
+      "loss": 0.8828,
+      "step": 4485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735369295351283e-05,
+      "loss": 0.9531,
+      "step": 4486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735242120037422e-05,
+      "loss": 0.8711,
+      "step": 4487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351149188345502e-05,
+      "loss": 0.8945,
+      "step": 4488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7349876917471474e-05,
+      "loss": 0.8516,
+      "step": 4489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734860438779693e-05,
+      "loss": 0.9102,
+      "step": 4490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347331599366684e-05,
+      "loss": 1.0078,
+      "step": 4491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346058552225544e-05,
+      "loss": 0.9766,
+      "step": 4492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734478524641834e-05,
+      "loss": 0.9141,
+      "step": 4493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734351168198991e-05,
+      "loss": 0.9492,
+      "step": 4494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342237858985093e-05,
+      "loss": 0.9648,
+      "step": 4495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340963777448744e-05,
+      "loss": 0.8438,
+      "step": 4496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339689437425726e-05,
+      "loss": 0.8164,
+      "step": 4497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338414838960915e-05,
+      "loss": 1.1094,
+      "step": 4498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733713998209918e-05,
+      "loss": 0.8555,
+      "step": 4499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335864866885424e-05,
+      "loss": 0.9453,
+      "step": 4500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334589493364537e-05,
+      "loss": 0.9531,
+      "step": 4501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333313861581427e-05,
+      "loss": 0.8555,
+      "step": 4502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332037971581015e-05,
+      "loss": 0.8711,
+      "step": 4503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330761823408226e-05,
+      "loss": 0.8789,
+      "step": 4504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329485417107995e-05,
+      "loss": 0.9375,
+      "step": 4505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328208752725264e-05,
+      "loss": 0.9688,
+      "step": 4506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732693183030499e-05,
+      "loss": 0.8086,
+      "step": 4507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732565464989213e-05,
+      "loss": 0.8672,
+      "step": 4508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7324377211531662e-05,
+      "loss": 0.8906,
+      "step": 4509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323099515268564e-05,
+      "loss": 0.9219,
+      "step": 4510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321821561147823e-05,
+      "loss": 0.8359,
+      "step": 4511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320543349214444e-05,
+      "loss": 0.9219,
+      "step": 4512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319264879513425e-05,
+      "loss": 0.8555,
+      "step": 4513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317986152089795e-05,
+      "loss": 0.918,
+      "step": 4514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731670716698857e-05,
+      "loss": 0.7578,
+      "step": 4515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315427924254787e-05,
+      "loss": 0.9062,
+      "step": 4516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314148423933494e-05,
+      "loss": 0.9297,
+      "step": 4517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312868666069746e-05,
+      "loss": 1.0,
+      "step": 4518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311588650708595e-05,
+      "loss": 0.7852,
+      "step": 4519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731030837789512e-05,
+      "loss": 0.8906,
+      "step": 4520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73090278476744e-05,
+      "loss": 0.7891,
+      "step": 4521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7307747060091525e-05,
+      "loss": 0.9219,
+      "step": 4522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306466015191595e-05,
+      "loss": 0.8555,
+      "step": 4523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305184713019715e-05,
+      "loss": 0.9062,
+      "step": 4524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303903153621e-05,
+      "loss": 0.9492,
+      "step": 4525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7302621337040576e-05,
+      "loss": 1.0625,
+      "step": 4526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730133926332358e-05,
+      "loss": 0.9414,
+      "step": 4527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730005693251516e-05,
+      "loss": 0.9844,
+      "step": 4528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729877434466046e-05,
+      "loss": 0.9023,
+      "step": 4529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297491499804645e-05,
+      "loss": 0.9648,
+      "step": 4530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729620839799289e-05,
+      "loss": 1.0469,
+      "step": 4531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294925039270373e-05,
+      "loss": 1.0,
+      "step": 4532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729364142368228e-05,
+      "loss": 0.9766,
+      "step": 4533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729235755127381e-05,
+      "loss": 0.8242,
+      "step": 4534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291073422090176e-05,
+      "loss": 0.8516,
+      "step": 4535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7289789036176582e-05,
+      "loss": 0.9219,
+      "step": 4536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288504393578267e-05,
+      "loss": 0.8438,
+      "step": 4537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287219494340454e-05,
+      "loss": 0.8594,
+      "step": 4538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285934338508396e-05,
+      "loss": 0.9609,
+      "step": 4539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284648926127337e-05,
+      "loss": 0.9688,
+      "step": 4540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283363257242544e-05,
+      "loss": 0.9805,
+      "step": 4541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728207733189928e-05,
+      "loss": 0.8828,
+      "step": 4542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728079115014284e-05,
+      "loss": 0.8711,
+      "step": 4543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279504712018495e-05,
+      "loss": 0.9453,
+      "step": 4544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727821801757155e-05,
+      "loss": 0.8789,
+      "step": 4545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727693106684731e-05,
+      "loss": 0.7695,
+      "step": 4546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275643859891092e-05,
+      "loss": 0.7031,
+      "step": 4547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274356396748225e-05,
+      "loss": 0.8594,
+      "step": 4548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273068677464034e-05,
+      "loss": 0.9766,
+      "step": 4549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271780702083863e-05,
+      "loss": 0.9883,
+      "step": 4550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727049247065307e-05,
+      "loss": 0.8867,
+      "step": 4551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726920398321701e-05,
+      "loss": 0.9062,
+      "step": 4552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726791523982105e-05,
+      "loss": 0.9805,
+      "step": 4553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266626240510577e-05,
+      "loss": 0.9375,
+      "step": 4554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7265336985330975e-05,
+      "loss": 0.918,
+      "step": 4555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726404747432764e-05,
+      "loss": 0.8477,
+      "step": 4556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262757707545976e-05,
+      "loss": 0.8711,
+      "step": 4557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7261467685031398e-05,
+      "loss": 0.8789,
+      "step": 4558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260177406829333e-05,
+      "loss": 0.8125,
+      "step": 4559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258886872985214e-05,
+      "loss": 0.9492,
+      "step": 4560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725759608354448e-05,
+      "loss": 0.9023,
+      "step": 4561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256305038552582e-05,
+      "loss": 0.8828,
+      "step": 4562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255013738054978e-05,
+      "loss": 0.9141,
+      "step": 4563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725372218209714e-05,
+      "loss": 0.8594,
+      "step": 4564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252430370724542e-05,
+      "loss": 1.0391,
+      "step": 4565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251138303982675e-05,
+      "loss": 0.9375,
+      "step": 4566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724984598191703e-05,
+      "loss": 0.7734,
+      "step": 4567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7248553404573118e-05,
+      "loss": 0.9336,
+      "step": 4568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247260571996447e-05,
+      "loss": 0.9727,
+      "step": 4569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724596748423254e-05,
+      "loss": 0.9219,
+      "step": 4570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244674141326932e-05,
+      "loss": 0.8633,
+      "step": 4571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243380543325164e-05,
+      "loss": 0.875,
+      "step": 4572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724208669027278e-05,
+      "loss": 0.8281,
+      "step": 4573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240792582215344e-05,
+      "loss": 0.875,
+      "step": 4574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723949821919842e-05,
+      "loss": 0.9531,
+      "step": 4575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238203601267586e-05,
+      "loss": 0.918,
+      "step": 4576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723690872846843e-05,
+      "loss": 0.9414,
+      "step": 4577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235613600846538e-05,
+      "loss": 0.9062,
+      "step": 4578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234318218447525e-05,
+      "loss": 0.8672,
+      "step": 4579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233022581316997e-05,
+      "loss": 0.9023,
+      "step": 4580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7231726689500575e-05,
+      "loss": 0.918,
+      "step": 4581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230430543043887e-05,
+      "loss": 0.9844,
+      "step": 4582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229134141992582e-05,
+      "loss": 0.8711,
+      "step": 4583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72278374863923e-05,
+      "loss": 0.8555,
+      "step": 4584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226540576288696e-05,
+      "loss": 1.0,
+      "step": 4585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722524341172745e-05,
+      "loss": 0.9922,
+      "step": 4586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223945992754218e-05,
+      "loss": 0.8945,
+      "step": 4587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72226483194147e-05,
+      "loss": 0.9609,
+      "step": 4588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722135039175458e-05,
+      "loss": 0.8828,
+      "step": 4589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722005220981956e-05,
+      "loss": 1.0234,
+      "step": 4590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721875377365536e-05,
+      "loss": 0.8945,
+      "step": 4591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7217455083307692e-05,
+      "loss": 0.8047,
+      "step": 4592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216156138822286e-05,
+      "loss": 0.8867,
+      "step": 4593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721485694024488e-05,
+      "loss": 0.8047,
+      "step": 4594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213557487621224e-05,
+      "loss": 0.9805,
+      "step": 4595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721225778099707e-05,
+      "loss": 0.9805,
+      "step": 4596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210957820418183e-05,
+      "loss": 0.7383,
+      "step": 4597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720965760593034e-05,
+      "loss": 0.9297,
+      "step": 4598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208357137579318e-05,
+      "loss": 0.9297,
+      "step": 4599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207056415410913e-05,
+      "loss": 0.9297,
+      "step": 4600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205755439470923e-05,
+      "loss": 0.7695,
+      "step": 4601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7204454209805157e-05,
+      "loss": 0.8125,
+      "step": 4602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203152726459437e-05,
+      "loss": 0.9688,
+      "step": 4603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201850989479588e-05,
+      "loss": 0.8906,
+      "step": 4604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200548998911444e-05,
+      "loss": 0.8438,
+      "step": 4605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199246754800853e-05,
+      "loss": 0.9766,
+      "step": 4606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197944257193667e-05,
+      "loss": 0.9414,
+      "step": 4607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7196641506135755e-05,
+      "loss": 0.9492,
+      "step": 4608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195338501672978e-05,
+      "loss": 0.8672,
+      "step": 4609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194035243851225e-05,
+      "loss": 0.8906,
+      "step": 4610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7192731732716385e-05,
+      "loss": 0.9531,
+      "step": 4611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191427968314353e-05,
+      "loss": 0.9766,
+      "step": 4612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719012395069104e-05,
+      "loss": 0.8594,
+      "step": 4613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718881967989236e-05,
+      "loss": 0.8516,
+      "step": 4614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718751515596424e-05,
+      "loss": 0.8516,
+      "step": 4615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186210378952616e-05,
+      "loss": 0.9297,
+      "step": 4616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184905348903427e-05,
+      "loss": 1.0078,
+      "step": 4617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718360006586263e-05,
+      "loss": 0.7656,
+      "step": 4618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182294529876178e-05,
+      "loss": 0.9453,
+      "step": 4619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718098874099005e-05,
+      "loss": 0.9219,
+      "step": 4620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717968269925022e-05,
+      "loss": 0.9883,
+      "step": 4621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178376404702678e-05,
+      "loss": 0.9141,
+      "step": 4622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177069857393414e-05,
+      "loss": 0.8945,
+      "step": 4623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175763057368445e-05,
+      "loss": 0.9922,
+      "step": 4624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174456004673778e-05,
+      "loss": 0.957,
+      "step": 4625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173148699355438e-05,
+      "loss": 0.8867,
+      "step": 4626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171841141459455e-05,
+      "loss": 1.0078,
+      "step": 4627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170533331031875e-05,
+      "loss": 0.8203,
+      "step": 4628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7169225268118742e-05,
+      "loss": 0.9453,
+      "step": 4629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716791695276612e-05,
+      "loss": 1.1484,
+      "step": 4630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166608385020074e-05,
+      "loss": 0.9102,
+      "step": 4631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165299564926683e-05,
+      "loss": 0.8789,
+      "step": 4632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716399049253203e-05,
+      "loss": 0.957,
+      "step": 4633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716268116788221e-05,
+      "loss": 0.7773,
+      "step": 4634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161371591023326e-05,
+      "loss": 1.1328,
+      "step": 4635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160061762001494e-05,
+      "loss": 1.0,
+      "step": 4636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158751680862832e-05,
+      "loss": 0.9062,
+      "step": 4637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715744134765347e-05,
+      "loss": 0.8516,
+      "step": 4638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156130762419545e-05,
+      "loss": 0.7734,
+      "step": 4639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715481992520721e-05,
+      "loss": 0.9648,
+      "step": 4640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153508836062617e-05,
+      "loss": 0.8945,
+      "step": 4641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715219749503193e-05,
+      "loss": 0.9805,
+      "step": 4642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715088590216133e-05,
+      "loss": 0.9336,
+      "step": 4643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149574057496994e-05,
+      "loss": 0.9141,
+      "step": 4644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148261961085117e-05,
+      "loss": 0.9141,
+      "step": 4645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71469496129719e-05,
+      "loss": 0.8594,
+      "step": 4646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145637013203555e-05,
+      "loss": 0.957,
+      "step": 4647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144324161826292e-05,
+      "loss": 0.7461,
+      "step": 4648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714301105888635e-05,
+      "loss": 0.8906,
+      "step": 4649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141697704429957e-05,
+      "loss": 1.0391,
+      "step": 4650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714038409850336e-05,
+      "loss": 0.8789,
+      "step": 4651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713907024115282e-05,
+      "loss": 0.8711,
+      "step": 4652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713775613242459e-05,
+      "loss": 0.8047,
+      "step": 4653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136441772364943e-05,
+      "loss": 0.8594,
+      "step": 4654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135127161020166e-05,
+      "loss": 0.8633,
+      "step": 4655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133812298436543e-05,
+      "loss": 0.9336,
+      "step": 4656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132497184660376e-05,
+      "loss": 0.7852,
+      "step": 4657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131181819737972e-05,
+      "loss": 0.8477,
+      "step": 4658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129866203715642e-05,
+      "loss": 0.8672,
+      "step": 4659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128550336639717e-05,
+      "loss": 0.9414,
+      "step": 4660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127234218556527e-05,
+      "loss": 0.9062,
+      "step": 4661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125917849512417e-05,
+      "loss": 0.8477,
+      "step": 4662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7124601229553736e-05,
+      "loss": 0.8867,
+      "step": 4663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7123284358726848e-05,
+      "loss": 0.8477,
+      "step": 4664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121967237078118e-05,
+      "loss": 0.9258,
+      "step": 4665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120649864653923e-05,
+      "loss": 0.9609,
+      "step": 4666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119332241500655e-05,
+      "loss": 0.9766,
+      "step": 4667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118014367664706e-05,
+      "loss": 0.8047,
+      "step": 4668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116696243192477e-05,
+      "loss": 1.0547,
+      "step": 4669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711537786813039e-05,
+      "loss": 0.9492,
+      "step": 4670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711405924252486e-05,
+      "loss": 0.9766,
+      "step": 4671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711274036642232e-05,
+      "loss": 0.7969,
+      "step": 4672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711142123986921e-05,
+      "loss": 1.0781,
+      "step": 4673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711010186291198e-05,
+      "loss": 0.875,
+      "step": 4674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108782235597083e-05,
+      "loss": 0.8359,
+      "step": 4675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710746235797099e-05,
+      "loss": 0.9922,
+      "step": 4676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710614223008017e-05,
+      "loss": 0.9883,
+      "step": 4677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710482185197111e-05,
+      "loss": 0.875,
+      "step": 4678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103501223690308e-05,
+      "loss": 0.9336,
+      "step": 4679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102180345284255e-05,
+      "loss": 0.8242,
+      "step": 4680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100859216799467e-05,
+      "loss": 0.9688,
+      "step": 4681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099537838282466e-05,
+      "loss": 0.9102,
+      "step": 4682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709821620977977e-05,
+      "loss": 0.9453,
+      "step": 4683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7096894331337926e-05,
+      "loss": 0.9062,
+      "step": 4684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095572203003473e-05,
+      "loss": 0.8359,
+      "step": 4685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094249824822968e-05,
+      "loss": 0.8164,
+      "step": 4686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7092927196842977e-05,
+      "loss": 0.9492,
+      "step": 4687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091604319110063e-05,
+      "loss": 0.9727,
+      "step": 4688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090281191670813e-05,
+      "loss": 0.8086,
+      "step": 4689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088957814571813e-05,
+      "loss": 0.875,
+      "step": 4690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087634187859668e-05,
+      "loss": 0.9023,
+      "step": 4691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086310311580975e-05,
+      "loss": 0.8633,
+      "step": 4692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084986185782354e-05,
+      "loss": 0.8906,
+      "step": 4693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083661810510435e-05,
+      "loss": 0.9922,
+      "step": 4694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082337185811844e-05,
+      "loss": 0.8984,
+      "step": 4695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081012311733225e-05,
+      "loss": 0.9258,
+      "step": 4696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707968718832123e-05,
+      "loss": 0.8945,
+      "step": 4697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078361815622518e-05,
+      "loss": 1.0,
+      "step": 4698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077036193683758e-05,
+      "loss": 1.0156,
+      "step": 4699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075710322551627e-05,
+      "loss": 0.9805,
+      "step": 4700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074384202272813e-05,
+      "loss": 0.8984,
+      "step": 4701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073057832894006e-05,
+      "loss": 0.9883,
+      "step": 4702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707173121446191e-05,
+      "loss": 0.8945,
+      "step": 4703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7070404347023238e-05,
+      "loss": 0.957,
+      "step": 4704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069077230624716e-05,
+      "loss": 0.9062,
+      "step": 4705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706774986531307e-05,
+      "loss": 0.8906,
+      "step": 4706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706642225113504e-05,
+      "loss": 1.0078,
+      "step": 4707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065094388137365e-05,
+      "loss": 0.9961,
+      "step": 4708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063766276366814e-05,
+      "loss": 0.9336,
+      "step": 4709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7062437915870145e-05,
+      "loss": 1.0,
+      "step": 4710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061109306694133e-05,
+      "loss": 0.9062,
+      "step": 4711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059780448885562e-05,
+      "loss": 0.9102,
+      "step": 4712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058451342491218e-05,
+      "loss": 1.0547,
+      "step": 4713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057121987557902e-05,
+      "loss": 0.8867,
+      "step": 4714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055792384132428e-05,
+      "loss": 0.8203,
+      "step": 4715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705446253226161e-05,
+      "loss": 0.8047,
+      "step": 4716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053132431992274e-05,
+      "loss": 0.8828,
+      "step": 4717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051802083371255e-05,
+      "loss": 0.9375,
+      "step": 4718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050471486445393e-05,
+      "loss": 0.9062,
+      "step": 4719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704914064126155e-05,
+      "loss": 0.9453,
+      "step": 4720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704780954786658e-05,
+      "loss": 0.9141,
+      "step": 4721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704647820630735e-05,
+      "loss": 1.0078,
+      "step": 4722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704514661663075e-05,
+      "loss": 0.8438,
+      "step": 4723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704381477888365e-05,
+      "loss": 0.9414,
+      "step": 4724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704248269311296e-05,
+      "loss": 1.0,
+      "step": 4725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041150359365585e-05,
+      "loss": 1.0156,
+      "step": 4726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703981777768843e-05,
+      "loss": 0.8438,
+      "step": 4727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7038484948128423e-05,
+      "loss": 0.8906,
+      "step": 4728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037151870732495e-05,
+      "loss": 0.8516,
+      "step": 4729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703581854554758e-05,
+      "loss": 0.9766,
+      "step": 4730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034484972620636e-05,
+      "loss": 0.8633,
+      "step": 4731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033151151998613e-05,
+      "loss": 0.9766,
+      "step": 4732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031817083728477e-05,
+      "loss": 1.0,
+      "step": 4733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7030482767857206e-05,
+      "loss": 1.0391,
+      "step": 4734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702914820443178e-05,
+      "loss": 0.9141,
+      "step": 4735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027813393499197e-05,
+      "loss": 0.9141,
+      "step": 4736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026478335106448e-05,
+      "loss": 0.9258,
+      "step": 4737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025143029300552e-05,
+      "loss": 0.8125,
+      "step": 4738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702380747612852e-05,
+      "loss": 0.8945,
+      "step": 4739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022471675637385e-05,
+      "loss": 0.8828,
+      "step": 4740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021135627874178e-05,
+      "loss": 0.8945,
+      "step": 4741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019799332885944e-05,
+      "loss": 0.8945,
+      "step": 4742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701846279071974e-05,
+      "loss": 0.9297,
+      "step": 4743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701712600142262e-05,
+      "loss": 0.8047,
+      "step": 4744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015788965041663e-05,
+      "loss": 0.8555,
+      "step": 4745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7014451681623944e-05,
+      "loss": 0.8984,
+      "step": 4746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701311415121655e-05,
+      "loss": 1.1016,
+      "step": 4747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011776373866577e-05,
+      "loss": 0.8203,
+      "step": 4748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010438349621136e-05,
+      "loss": 0.9258,
+      "step": 4749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009100078527333e-05,
+      "loss": 0.8906,
+      "step": 4750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007761560632294e-05,
+      "loss": 0.9219,
+      "step": 4751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700642279598315e-05,
+      "loss": 0.9531,
+      "step": 4752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005083784627047e-05,
+      "loss": 0.9297,
+      "step": 4753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7003744526611122e-05,
+      "loss": 0.9805,
+      "step": 4754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002405021982543e-05,
+      "loss": 0.8555,
+      "step": 4755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001065270788468e-05,
+      "loss": 0.8672,
+      "step": 4756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999725273076074e-05,
+      "loss": 0.9258,
+      "step": 4757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699838502889255e-05,
+      "loss": 0.9297,
+      "step": 4758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699704453828508e-05,
+      "loss": 1.1172,
+      "step": 4759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699570380130087e-05,
+      "loss": 0.8906,
+      "step": 4760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994362817987128e-05,
+      "loss": 1.0,
+      "step": 4761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993021588391073e-05,
+      "loss": 0.9375,
+      "step": 4762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699168011255993e-05,
+      "loss": 0.9141,
+      "step": 4763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990338390540935e-05,
+      "loss": 0.9258,
+      "step": 4764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698899642238133e-05,
+      "loss": 0.9219,
+      "step": 4765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987654208128373e-05,
+      "loss": 0.9609,
+      "step": 4766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698631174782932e-05,
+      "loss": 0.8672,
+      "step": 4767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984969041531444e-05,
+      "loss": 1.0312,
+      "step": 4768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983626089282025e-05,
+      "loss": 0.9141,
+      "step": 4769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982282891128347e-05,
+      "loss": 0.7891,
+      "step": 4770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980939447117707e-05,
+      "loss": 0.9258,
+      "step": 4771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979595757297415e-05,
+      "loss": 0.9766,
+      "step": 4772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978251821714776e-05,
+      "loss": 0.9531,
+      "step": 4773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976907640417114e-05,
+      "loss": 0.8438,
+      "step": 4774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975563213451767e-05,
+      "loss": 1.0234,
+      "step": 4775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6974218540866068e-05,
+      "loss": 0.9375,
+      "step": 4776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697287362270736e-05,
+      "loss": 0.8984,
+      "step": 4777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971528459023012e-05,
+      "loss": 0.8242,
+      "step": 4778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970183049860383e-05,
+      "loss": 0.9922,
+      "step": 4779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6968837395266844e-05,
+      "loss": 0.957,
+      "step": 4780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967491495289782e-05,
+      "loss": 0.8047,
+      "step": 4781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966145349976585e-05,
+      "loss": 0.8867,
+      "step": 4782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964798959374658e-05,
+      "loss": 0.9414,
+      "step": 4783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963452323531403e-05,
+      "loss": 0.7617,
+      "step": 4784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962105442494242e-05,
+      "loss": 0.7812,
+      "step": 4785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960758316310597e-05,
+      "loss": 0.9141,
+      "step": 4786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6959410945027903e-05,
+      "loss": 1.0078,
+      "step": 4787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958063328693607e-05,
+      "loss": 0.918,
+      "step": 4788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695671546735516e-05,
+      "loss": 0.8711,
+      "step": 4789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955367361060015e-05,
+      "loss": 1.0,
+      "step": 4790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954019009855648e-05,
+      "loss": 0.9297,
+      "step": 4791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952670413789536e-05,
+      "loss": 1.0,
+      "step": 4792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951321572909164e-05,
+      "loss": 0.9453,
+      "step": 4793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949972487262024e-05,
+      "loss": 1.0,
+      "step": 4794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948623156895627e-05,
+      "loss": 0.9453,
+      "step": 4795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947273581857474e-05,
+      "loss": 0.8867,
+      "step": 4796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6945923762195096e-05,
+      "loss": 0.9219,
+      "step": 4797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944573697956016e-05,
+      "loss": 0.8789,
+      "step": 4798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943223389187774e-05,
+      "loss": 1.0312,
+      "step": 4799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941872835937915e-05,
+      "loss": 0.8516,
+      "step": 4800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940522038254e-05,
+      "loss": 0.8398,
+      "step": 4801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693917099618358e-05,
+      "loss": 0.9961,
+      "step": 4802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937819709774244e-05,
+      "loss": 0.9219,
+      "step": 4803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6936468179073558e-05,
+      "loss": 0.9336,
+      "step": 4804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935116404129116e-05,
+      "loss": 0.8867,
+      "step": 4805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693376438498852e-05,
+      "loss": 0.9531,
+      "step": 4806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932412121699377e-05,
+      "loss": 0.8203,
+      "step": 4807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69310596143093e-05,
+      "loss": 0.8203,
+      "step": 4808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692970686286591e-05,
+      "loss": 0.7773,
+      "step": 4809
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692835386741684e-05,
+      "loss": 0.8125,
+      "step": 4810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927000628009735e-05,
+      "loss": 0.9141,
+      "step": 4811
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6925647144692243e-05,
+      "loss": 0.9336,
+      "step": 4812
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924293417512024e-05,
+      "loss": 0.8672,
+      "step": 4813
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692293944651674e-05,
+      "loss": 0.9297,
+      "step": 4814
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692158523175407e-05,
+      "loss": 1.0078,
+      "step": 4815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692023077327169e-05,
+      "loss": 0.8594,
+      "step": 4816
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691887607111731e-05,
+      "loss": 0.8828,
+      "step": 4817
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6917521125338615e-05,
+      "loss": 0.9961,
+      "step": 4818
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916165935983323e-05,
+      "loss": 0.8633,
+      "step": 4819
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691481050309915e-05,
+      "loss": 1.0391,
+      "step": 4820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691345482673382e-05,
+      "loss": 0.8438,
+      "step": 4821
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691209890693507e-05,
+      "loss": 0.9648,
+      "step": 4822
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910742743750654e-05,
+      "loss": 0.8594,
+      "step": 4823
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690938633722831e-05,
+      "loss": 0.8945,
+      "step": 4824
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908029687415806e-05,
+      "loss": 1.0078,
+      "step": 4825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906672794360907e-05,
+      "loss": 1.0078,
+      "step": 4826
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69053156581114e-05,
+      "loss": 0.8711,
+      "step": 4827
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690395827871506e-05,
+      "loss": 0.8281,
+      "step": 4828
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902600656219697e-05,
+      "loss": 1.0391,
+      "step": 4829
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901242790673108e-05,
+      "loss": 0.9453,
+      "step": 4830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68998846821231e-05,
+      "loss": 0.8281,
+      "step": 4831
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68985263306175e-05,
+      "loss": 0.8633,
+      "step": 4832
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897167736204135e-05,
+      "loss": 0.8867,
+      "step": 4833
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689580889893085e-05,
+      "loss": 0.9453,
+      "step": 4834
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689444981884548e-05,
+      "loss": 0.8633,
+      "step": 4835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6893090495995892e-05,
+      "loss": 0.8594,
+      "step": 4836
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6891730930429943e-05,
+      "loss": 0.9258,
+      "step": 4837
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890371122195505e-05,
+      "loss": 0.7812,
+      "step": 4838
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6889011071340466e-05,
+      "loss": 0.8672,
+      "step": 4839
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6887650777912705e-05,
+      "loss": 0.9141,
+      "step": 4840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688629024196013e-05,
+      "loss": 0.9141,
+      "step": 4841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6884929463530643e-05,
+      "loss": 0.8242,
+      "step": 4842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6883568442672155e-05,
+      "loss": 0.8789,
+      "step": 4843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882207179432598e-05,
+      "loss": 0.793,
+      "step": 4844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68808456738599e-05,
+      "loss": 0.9727,
+      "step": 4845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6879483926002002e-05,
+      "loss": 0.9844,
+      "step": 4846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6878121935906847e-05,
+      "loss": 0.9531,
+      "step": 4847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876759703622404e-05,
+      "loss": 1.0781,
+      "step": 4848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6875397229196636e-05,
+      "loss": 1.0625,
+      "step": 4849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874034512677514e-05,
+      "loss": 0.9336,
+      "step": 4850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687267155411302e-05,
+      "loss": 0.918,
+      "step": 4851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871308353551153e-05,
+      "loss": 0.8516,
+      "step": 4852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686994491103991e-05,
+      "loss": 0.8516,
+      "step": 4853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868581226627298e-05,
+      "loss": 1.1172,
+      "step": 4854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867217300361335e-05,
+      "loss": 0.8945,
+      "step": 4855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686585313229005e-05,
+      "loss": 0.9219,
+      "step": 4856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864488722461473e-05,
+      "loss": 0.8672,
+      "step": 4857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6863124070923655e-05,
+      "loss": 0.9609,
+      "step": 4858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861759177724636e-05,
+      "loss": 0.9844,
+      "step": 4859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6860394042912484e-05,
+      "loss": 0.8516,
+      "step": 4860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859028666535263e-05,
+      "loss": 0.8086,
+      "step": 4861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6857663048641055e-05,
+      "loss": 0.8242,
+      "step": 4862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685629718927794e-05,
+      "loss": 0.8594,
+      "step": 4863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854931088494016e-05,
+      "loss": 0.8711,
+      "step": 4864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853564746337388e-05,
+      "loss": 0.918,
+      "step": 4865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685219816285616e-05,
+      "loss": 0.8633,
+      "step": 4866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850831338098455e-05,
+      "loss": 0.7773,
+      "step": 4867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849464272112404e-05,
+      "loss": 0.8672,
+      "step": 4868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684809696494614e-05,
+      "loss": 0.918,
+      "step": 4869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846729416647804e-05,
+      "loss": 0.9414,
+      "step": 4870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6845361627265556e-05,
+      "loss": 0.9648,
+      "step": 4871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6843993596847554e-05,
+      "loss": 0.8867,
+      "step": 4872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842625325441973e-05,
+      "loss": 0.8711,
+      "step": 4873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684125681309699e-05,
+      "loss": 0.8438,
+      "step": 4874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683988805986079e-05,
+      "loss": 0.9922,
+      "step": 4875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683851906578157e-05,
+      "loss": 0.8867,
+      "step": 4876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6837149830907536e-05,
+      "loss": 0.8359,
+      "step": 4877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68357803552869e-05,
+      "loss": 0.9258,
+      "step": 4878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6834410638967882e-05,
+      "loss": 0.8555,
+      "step": 4879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833040681998717e-05,
+      "loss": 0.9297,
+      "step": 4880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831670484427638e-05,
+      "loss": 0.8984,
+      "step": 4881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683030004630289e-05,
+      "loss": 0.8945,
+      "step": 4882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828929367672733e-05,
+      "loss": 0.8125,
+      "step": 4883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827558448585433e-05,
+      "loss": 0.9531,
+      "step": 4884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826187289089254e-05,
+      "loss": 0.9414,
+      "step": 4885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682481588923248e-05,
+      "loss": 0.9297,
+      "step": 4886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823444249063407e-05,
+      "loss": 0.8945,
+      "step": 4887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682207236863032e-05,
+      "loss": 0.875,
+      "step": 4888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820700247981533e-05,
+      "loss": 0.8711,
+      "step": 4889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819327887165362e-05,
+      "loss": 0.9141,
+      "step": 4890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817955286230127e-05,
+      "loss": 0.9219,
+      "step": 4891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6816582445224158e-05,
+      "loss": 0.9062,
+      "step": 4892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681520936419579e-05,
+      "loss": 0.9375,
+      "step": 4893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681383604319339e-05,
+      "loss": 0.9414,
+      "step": 4894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812462482265295e-05,
+      "loss": 0.9297,
+      "step": 4895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811088681459876e-05,
+      "loss": 1.0,
+      "step": 4896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680971464082551e-05,
+      "loss": 0.8438,
+      "step": 4897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808340360410583e-05,
+      "loss": 0.9805,
+      "step": 4898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806965840263475e-05,
+      "loss": 0.7812,
+      "step": 4899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680559108043259e-05,
+      "loss": 0.8164,
+      "step": 4900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6804216080966336e-05,
+      "loss": 0.9297,
+      "step": 4901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802840841913125e-05,
+      "loss": 0.8867,
+      "step": 4902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680146536332139e-05,
+      "loss": 0.8438,
+      "step": 4903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800089645239554e-05,
+      "loss": 0.8438,
+      "step": 4904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798713687716063e-05,
+      "loss": 0.8789,
+      "step": 4905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797337490799367e-05,
+      "loss": 0.9102,
+      "step": 4906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679596105453792e-05,
+      "loss": 0.9219,
+      "step": 4907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794584378980196e-05,
+      "loss": 0.7969,
+      "step": 4908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679320746417466e-05,
+      "loss": 0.8359,
+      "step": 4909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6791830310169804e-05,
+      "loss": 0.9102,
+      "step": 4910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790452917014115e-05,
+      "loss": 0.8867,
+      "step": 4911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789075284756092e-05,
+      "loss": 0.9102,
+      "step": 4912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678769741344425e-05,
+      "loss": 0.8555,
+      "step": 4913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786319303127102e-05,
+      "loss": 1.0312,
+      "step": 4914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784940953853172e-05,
+      "loss": 0.9922,
+      "step": 4915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6783562365670994e-05,
+      "loss": 0.8477,
+      "step": 4916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782183538629112e-05,
+      "loss": 0.7031,
+      "step": 4917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678080447277608e-05,
+      "loss": 0.8203,
+      "step": 4918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6779425168160447e-05,
+      "loss": 0.8867,
+      "step": 4919
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778045624830788e-05,
+      "loss": 0.9492,
+      "step": 4920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776665842835678e-05,
+      "loss": 0.9648,
+      "step": 4921
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6775285822223702e-05,
+      "loss": 0.8086,
+      "step": 4922
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677390556304345e-05,
+      "loss": 1.0156,
+      "step": 4923
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677252506534352e-05,
+      "loss": 0.9297,
+      "step": 4924
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677114432917253e-05,
+      "loss": 0.9492,
+      "step": 4925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769763354579093e-05,
+      "loss": 0.7773,
+      "step": 4926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768382141611838e-05,
+      "loss": 0.832,
+      "step": 4927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676700069031939e-05,
+      "loss": 0.8828,
+      "step": 4928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676561900075041e-05,
+      "loss": 1.0859,
+      "step": 4929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764237072953534e-05,
+      "loss": 0.793,
+      "step": 4930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762854906977427e-05,
+      "loss": 0.8594,
+      "step": 4931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761472502870754e-05,
+      "loss": 1.0,
+      "step": 4932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760089860682198e-05,
+      "loss": 1.0312,
+      "step": 4933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758706980460436e-05,
+      "loss": 0.7617,
+      "step": 4934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6757323862254173e-05,
+      "loss": 1.0469,
+      "step": 4935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755940506112096e-05,
+      "loss": 0.8555,
+      "step": 4936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6754556912082926e-05,
+      "loss": 0.9062,
+      "step": 4937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675317308021538e-05,
+      "loss": 1.0,
+      "step": 4938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675178901055818e-05,
+      "loss": 0.8359,
+      "step": 4939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750404703160063e-05,
+      "loss": 0.9453,
+      "step": 4940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749020158069778e-05,
+      "loss": 0.875,
+      "step": 4941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6747635375336072e-05,
+      "loss": 0.832,
+      "step": 4942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746250355007706e-05,
+      "loss": 0.8594,
+      "step": 4943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744865097133446e-05,
+      "loss": 0.9805,
+      "step": 4944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743479601762075e-05,
+      "loss": 0.9258,
+      "step": 4945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742093868942374e-05,
+      "loss": 0.9492,
+      "step": 4946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674070789872314e-05,
+      "loss": 0.9844,
+      "step": 4947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6739321691153167e-05,
+      "loss": 0.8633,
+      "step": 4948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737935246281277e-05,
+      "loss": 0.918,
+      "step": 4949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736548564156282e-05,
+      "loss": 1.0234,
+      "step": 4950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735161644827015e-05,
+      "loss": 0.8711,
+      "step": 4951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67337744883423e-05,
+      "loss": 0.9688,
+      "step": 4952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732387094750996e-05,
+      "loss": 0.918,
+      "step": 4953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730999464101942e-05,
+      "loss": 0.918,
+      "step": 4954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672961159644401e-05,
+      "loss": 0.8906,
+      "step": 4955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728223491826055e-05,
+      "loss": 0.8516,
+      "step": 4956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672683515029697e-05,
+      "loss": 0.8555,
+      "step": 4957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672544657190563e-05,
+      "loss": 0.8203,
+      "step": 4958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6724057756700933e-05,
+      "loss": 0.9336,
+      "step": 4959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672266870473178e-05,
+      "loss": 1.0391,
+      "step": 4960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672127941604708e-05,
+      "loss": 0.8711,
+      "step": 4961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719889890695756e-05,
+      "loss": 0.918,
+      "step": 4962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718500128726734e-05,
+      "loss": 0.9648,
+      "step": 4963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671711013018895e-05,
+      "loss": 0.8984,
+      "step": 4964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715719895131342e-05,
+      "loss": 0.7383,
+      "step": 4965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6714329423602874e-05,
+      "loss": 0.9531,
+      "step": 4966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6712938715652497e-05,
+      "loss": 1.0781,
+      "step": 4967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711547771329184e-05,
+      "loss": 0.9375,
+      "step": 4968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671015659068191e-05,
+      "loss": 0.8828,
+      "step": 4969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708765173759664e-05,
+      "loss": 0.9258,
+      "step": 4970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707373520611435e-05,
+      "loss": 0.8945,
+      "step": 4971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705981631286233e-05,
+      "loss": 0.8789,
+      "step": 4972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670458950583306e-05,
+      "loss": 1.0312,
+      "step": 4973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703197144300937e-05,
+      "loss": 0.8867,
+      "step": 4974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701804546738894e-05,
+      "loss": 0.8789,
+      "step": 4975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670041171319597e-05,
+      "loss": 0.8242,
+      "step": 4976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699018643721197e-05,
+      "loss": 0.8398,
+      "step": 4977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669762533836364e-05,
+      "loss": 0.8516,
+      "step": 4978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6696231797172352e-05,
+      "loss": 0.9531,
+      "step": 4979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66948380201964e-05,
+      "loss": 0.8359,
+      "step": 4980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669344400748487e-05,
+      "loss": 0.8711,
+      "step": 4981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669204975908684e-05,
+      "loss": 0.8125,
+      "step": 4982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6690655275051405e-05,
+      "loss": 1.0312,
+      "step": 4983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689260555427666e-05,
+      "loss": 1.0078,
+      "step": 4984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687865600264736e-05,
+      "loss": 0.9141,
+      "step": 4985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668647040961173e-05,
+      "loss": 1.0312,
+      "step": 4986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668507498351778e-05,
+      "loss": 0.7812,
+      "step": 4987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668367932203202e-05,
+      "loss": 1.1484,
+      "step": 4988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668228342520359e-05,
+      "loss": 0.8477,
+      "step": 4989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668088729308164e-05,
+      "loss": 0.8672,
+      "step": 4990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679490925715333e-05,
+      "loss": 0.8555,
+      "step": 4991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6678094323153843e-05,
+      "loss": 0.8672,
+      "step": 4992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6676697485446336e-05,
+      "loss": 0.9453,
+      "step": 4993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6675300412642003e-05,
+      "loss": 0.8594,
+      "step": 4994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6673903104790035e-05,
+      "loss": 0.8555,
+      "step": 4995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672505561939635e-05,
+      "loss": 0.9453,
+      "step": 4996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671107784140015e-05,
+      "loss": 0.9688,
+      "step": 4997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669709771440385e-05,
+      "loss": 0.8789,
+      "step": 4998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6668311523889977e-05,
+      "loss": 0.8867,
+      "step": 4999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6666913041538025e-05,
+      "loss": 0.9805,
+      "step": 5000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665514324433775e-05,
+      "loss": 0.9453,
+      "step": 5001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666411537262647e-05,
+      "loss": 1.0,
+      "step": 5002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6662716186165373e-05,
+      "loss": 0.8438,
+      "step": 5003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666131676509975e-05,
+      "loss": 0.9648,
+      "step": 5004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659917109478882e-05,
+      "loss": 0.7891,
+      "step": 5005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665851721935205e-05,
+      "loss": 0.9102,
+      "step": 5006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657117094768543e-05,
+      "loss": 0.9805,
+      "step": 5007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665571673577767e-05,
+      "loss": 0.8594,
+      "step": 5008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6654316142428727e-05,
+      "loss": 0.9102,
+      "step": 5009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652915314771045e-05,
+      "loss": 0.9766,
+      "step": 5010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665151425285394e-05,
+      "loss": 0.875,
+      "step": 5011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650112956726748e-05,
+      "loss": 1.0234,
+      "step": 5012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664871142643881e-05,
+      "loss": 0.957,
+      "step": 5013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647309662039478e-05,
+      "loss": 0.8828,
+      "step": 5014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664590766357811e-05,
+      "loss": 0.8672,
+      "step": 5015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644505431104073e-05,
+      "loss": 0.7695,
+      "step": 5016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664310296466674e-05,
+      "loss": 0.9375,
+      "step": 5017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6641700264315493e-05,
+      "loss": 0.9531,
+      "step": 5018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640297330099726e-05,
+      "loss": 0.9102,
+      "step": 5019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638894162068836e-05,
+      "loss": 0.9297,
+      "step": 5020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637490760272234e-05,
+      "loss": 0.9531,
+      "step": 5021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6636087124759336e-05,
+      "loss": 1.0,
+      "step": 5022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663468325557956e-05,
+      "loss": 0.8789,
+      "step": 5023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633279152782346e-05,
+      "loss": 0.9219,
+      "step": 5024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631874816417126e-05,
+      "loss": 0.875,
+      "step": 5025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630470246533358e-05,
+      "loss": 0.8633,
+      "step": 5026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629065443180497e-05,
+      "loss": 0.8438,
+      "step": 5027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627660406408e-05,
+      "loss": 0.9062,
+      "step": 5028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626255136265355e-05,
+      "loss": 1.0547,
+      "step": 5029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624849632802033e-05,
+      "loss": 0.8828,
+      "step": 5030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623443896067527e-05,
+      "loss": 0.9102,
+      "step": 5031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622037926111332e-05,
+      "loss": 0.7773,
+      "step": 5032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662063172298296e-05,
+      "loss": 0.9219,
+      "step": 5033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661922528673192e-05,
+      "loss": 0.8203,
+      "step": 5034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661781861740774e-05,
+      "loss": 0.8945,
+      "step": 5035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616411715059948e-05,
+      "loss": 1.0156,
+      "step": 5036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615004579738084e-05,
+      "loss": 0.9805,
+      "step": 5037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6613597211491694e-05,
+      "loss": 0.9688,
+      "step": 5038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612189610370336e-05,
+      "loss": 0.832,
+      "step": 5039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661078177642357e-05,
+      "loss": 0.9922,
+      "step": 5040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6609373709700975e-05,
+      "loss": 0.7734,
+      "step": 5041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607965410252126e-05,
+      "loss": 0.9297,
+      "step": 5042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660655687812661e-05,
+      "loss": 0.8398,
+      "step": 5043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605148113374028e-05,
+      "loss": 0.8789,
+      "step": 5044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603739116043983e-05,
+      "loss": 0.9414,
+      "step": 5045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6602329886186084e-05,
+      "loss": 0.8242,
+      "step": 5046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600920423849958e-05,
+      "loss": 0.8828,
+      "step": 5047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599510729085232e-05,
+      "loss": 0.8828,
+      "step": 5048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6598100801941544e-05,
+      "loss": 0.8047,
+      "step": 5049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596690642468537e-05,
+      "loss": 0.9219,
+      "step": 5050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659528025071587e-05,
+      "loss": 0.8164,
+      "step": 5051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593869626733202e-05,
+      "loss": 0.7812,
+      "step": 5052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592458770570205e-05,
+      "loss": 0.832,
+      "step": 5053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659104768227655e-05,
+      "loss": 0.8789,
+      "step": 5054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589636361901935e-05,
+      "loss": 0.7734,
+      "step": 5055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588224809496046e-05,
+      "loss": 0.8398,
+      "step": 5056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658681302510859e-05,
+      "loss": 0.8555,
+      "step": 5057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658540100878928e-05,
+      "loss": 0.9141,
+      "step": 5058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583988760587834e-05,
+      "loss": 0.9805,
+      "step": 5059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582576280553972e-05,
+      "loss": 0.9414,
+      "step": 5060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581163568737442e-05,
+      "loss": 0.8828,
+      "step": 5061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579750625187974e-05,
+      "loss": 0.9219,
+      "step": 5062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6578337449955335e-05,
+      "loss": 0.9219,
+      "step": 5063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576924043089274e-05,
+      "loss": 0.9023,
+      "step": 5064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657551040463956e-05,
+      "loss": 0.8516,
+      "step": 5065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574096534655975e-05,
+      "loss": 0.875,
+      "step": 5066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6572682433188305e-05,
+      "loss": 0.8281,
+      "step": 5067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657126810028633e-05,
+      "loss": 0.875,
+      "step": 5068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569853535999864e-05,
+      "loss": 0.8164,
+      "step": 5069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568438740378712e-05,
+      "loss": 0.9375,
+      "step": 5070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567023713472686e-05,
+      "loss": 0.9023,
+      "step": 5071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565608455331622e-05,
+      "loss": 0.8555,
+      "step": 5072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564192966005343e-05,
+      "loss": 0.9922,
+      "step": 5073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656277724554369e-05,
+      "loss": 0.9453,
+      "step": 5074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561361293996524e-05,
+      "loss": 0.8828,
+      "step": 5075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559945111413696e-05,
+      "loss": 0.8281,
+      "step": 5076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6558528697845065e-05,
+      "loss": 0.9297,
+      "step": 5077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655711205334052e-05,
+      "loss": 0.9883,
+      "step": 5078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555695177949934e-05,
+      "loss": 0.8906,
+      "step": 5079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6554278071723197e-05,
+      "loss": 0.8477,
+      "step": 5080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655286073471021e-05,
+      "loss": 0.9297,
+      "step": 5081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655144316696088e-05,
+      "loss": 0.9922,
+      "step": 5082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655002536852512e-05,
+      "loss": 0.9414,
+      "step": 5083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.875,
+      "step": 5084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654718907979401e-05,
+      "loss": 0.8633,
+      "step": 5085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6545770589598535e-05,
+      "loss": 1.0,
+      "step": 5086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544351868916365e-05,
+      "loss": 0.9219,
+      "step": 5087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542932917797462e-05,
+      "loss": 0.8047,
+      "step": 5088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6541513736291788e-05,
+      "loss": 0.9531,
+      "step": 5089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654009432444932e-05,
+      "loss": 0.9414,
+      "step": 5090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6538674682320032e-05,
+      "loss": 0.8555,
+      "step": 5091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653725480995391e-05,
+      "loss": 0.8789,
+      "step": 5092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6535834707400953e-05,
+      "loss": 0.8672,
+      "step": 5093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534414374711167e-05,
+      "loss": 0.9062,
+      "step": 5094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532993811934558e-05,
+      "loss": 0.8281,
+      "step": 5095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653157301912115e-05,
+      "loss": 1.0625,
+      "step": 5096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530151996320975e-05,
+      "loss": 0.8789,
+      "step": 5097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528730743584067e-05,
+      "loss": 0.8086,
+      "step": 5098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527309260960468e-05,
+      "loss": 0.8594,
+      "step": 5099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652588754850023e-05,
+      "loss": 0.8906,
+      "step": 5100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524465606253412e-05,
+      "loss": 1.0078,
+      "step": 5101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652304343427009e-05,
+      "loss": 0.9062,
+      "step": 5102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652162103260034e-05,
+      "loss": 0.8516,
+      "step": 5103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520198401294242e-05,
+      "loss": 0.8906,
+      "step": 5104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6518775540401887e-05,
+      "loss": 0.9453,
+      "step": 5105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517352449973386e-05,
+      "loss": 0.9453,
+      "step": 5106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515929130058838e-05,
+      "loss": 1.0078,
+      "step": 5107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651450558070837e-05,
+      "loss": 0.9648,
+      "step": 5108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513081801972096e-05,
+      "loss": 0.8672,
+      "step": 5109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651165779390016e-05,
+      "loss": 0.957,
+      "step": 5110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510233556542697e-05,
+      "loss": 0.8867,
+      "step": 5111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650880908994986e-05,
+      "loss": 0.8828,
+      "step": 5112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65073843941718e-05,
+      "loss": 0.9453,
+      "step": 5113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6505959469258694e-05,
+      "loss": 0.8711,
+      "step": 5114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650453431526071e-05,
+      "loss": 0.8906,
+      "step": 5115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650310893222803e-05,
+      "loss": 0.875,
+      "step": 5116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650168332021084e-05,
+      "loss": 0.8906,
+      "step": 5117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500257479259344e-05,
+      "loss": 0.9648,
+      "step": 5118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498831409423746e-05,
+      "loss": 0.9141,
+      "step": 5119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649740511075426e-05,
+      "loss": 1.0312,
+      "step": 5120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495978583301107e-05,
+      "loss": 0.8164,
+      "step": 5121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649455182711452e-05,
+      "loss": 0.9258,
+      "step": 5122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493124842244733e-05,
+      "loss": 0.7891,
+      "step": 5123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491697628741998e-05,
+      "loss": 0.8477,
+      "step": 5124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490270186656566e-05,
+      "loss": 0.8047,
+      "step": 5125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6488842516038698e-05,
+      "loss": 0.8789,
+      "step": 5126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487414616938665e-05,
+      "loss": 0.9258,
+      "step": 5127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648598648940675e-05,
+      "loss": 0.9531,
+      "step": 5128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484558133493233e-05,
+      "loss": 0.8164,
+      "step": 5129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6483129549248415e-05,
+      "loss": 0.8867,
+      "step": 5130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481700736722595e-05,
+      "loss": 0.8398,
+      "step": 5131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480271695966084e-05,
+      "loss": 0.8164,
+      "step": 5132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64788424270292e-05,
+      "loss": 0.9336,
+      "step": 5133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477412929962272e-05,
+      "loss": 0.8672,
+      "step": 5134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475983204815636e-05,
+      "loss": 0.9062,
+      "step": 5135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647455325163963e-05,
+      "loss": 0.9219,
+      "step": 5136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6473123070484606e-05,
+      "loss": 1.0234,
+      "step": 5137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6471692661400925e-05,
+      "loss": 0.9219,
+      "step": 5138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470262024438954e-05,
+      "loss": 0.8477,
+      "step": 5139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468831159649067e-05,
+      "loss": 0.9453,
+      "step": 5140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467400067081646e-05,
+      "loss": 0.8867,
+      "step": 5141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465968746787083e-05,
+      "loss": 0.8672,
+      "step": 5142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646453719881578e-05,
+      "loss": 0.9102,
+      "step": 5143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6463105423218137e-05,
+      "loss": 0.8711,
+      "step": 5144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461673420044575e-05,
+      "loss": 0.8359,
+      "step": 5145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460241189345516e-05,
+      "loss": 0.8711,
+      "step": 5146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458808731171387e-05,
+      "loss": 0.9844,
+      "step": 5147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457376045572633e-05,
+      "loss": 0.8633,
+      "step": 5148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6455943132599698e-05,
+      "loss": 1.0938,
+      "step": 5149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645450999230304e-05,
+      "loss": 0.9141,
+      "step": 5150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453076624733112e-05,
+      "loss": 1.0391,
+      "step": 5151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6451643029940397e-05,
+      "loss": 0.9219,
+      "step": 5152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450209207975373e-05,
+      "loss": 0.9688,
+      "step": 5153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448775158888516e-05,
+      "loss": 0.8594,
+      "step": 5154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447340882730335e-05,
+      "loss": 0.8672,
+      "step": 5155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445906379551324e-05,
+      "loss": 0.9258,
+      "step": 5156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444471649402e-05,
+      "loss": 0.9648,
+      "step": 5157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443036692332876e-05,
+      "loss": 0.8984,
+      "step": 5158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644160150839448e-05,
+      "loss": 0.8867,
+      "step": 5159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440166097637355e-05,
+      "loss": 0.9336,
+      "step": 5160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6438730460112032e-05,
+      "loss": 0.8047,
+      "step": 5161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643729459586907e-05,
+      "loss": 0.9219,
+      "step": 5162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435858504959032e-05,
+      "loss": 0.8945,
+      "step": 5163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643442218743247e-05,
+      "loss": 0.8438,
+      "step": 5164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432985643339974e-05,
+      "loss": 0.9375,
+      "step": 5165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6431548872732118e-05,
+      "loss": 0.8789,
+      "step": 5166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430111875659497e-05,
+      "loss": 1.0469,
+      "step": 5167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642867465217271e-05,
+      "loss": 0.918,
+      "step": 5168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427237202322363e-05,
+      "loss": 0.9102,
+      "step": 5169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425799526159067e-05,
+      "loss": 0.9141,
+      "step": 5170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642436162373345e-05,
+      "loss": 0.8633,
+      "step": 5171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422923495096143e-05,
+      "loss": 0.8203,
+      "step": 5172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642148514029778e-05,
+      "loss": 0.8125,
+      "step": 5173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420046559389013e-05,
+      "loss": 0.8828,
+      "step": 5174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418607752420493e-05,
+      "loss": 0.8398,
+      "step": 5175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417168719442882e-05,
+      "loss": 0.9531,
+      "step": 5176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415729460506855e-05,
+      "loss": 0.9297,
+      "step": 5177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6414289975663086e-05,
+      "loss": 0.8672,
+      "step": 5178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6412850264962263e-05,
+      "loss": 0.8633,
+      "step": 5179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641141032845508e-05,
+      "loss": 0.957,
+      "step": 5180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409970166192243e-05,
+      "loss": 1.0312,
+      "step": 5181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640852977822446e-05,
+      "loss": 0.8828,
+      "step": 5182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407089164602446e-05,
+      "loss": 0.8711,
+      "step": 5183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640564832537693e-05,
+      "loss": 1.0,
+      "step": 5184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640420726059865e-05,
+      "loss": 0.875,
+      "step": 5185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640276597031834e-05,
+      "loss": 0.9258,
+      "step": 5186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401324454586757e-05,
+      "loss": 0.8477,
+      "step": 5187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399882713454655e-05,
+      "loss": 0.9102,
+      "step": 5188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63984407469728e-05,
+      "loss": 0.9531,
+      "step": 5189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639699855519197e-05,
+      "loss": 0.9375,
+      "step": 5190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639555613816294e-05,
+      "loss": 0.7812,
+      "step": 5191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639411349593651e-05,
+      "loss": 0.8906,
+      "step": 5192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392670628563466e-05,
+      "loss": 0.8906,
+      "step": 5193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639122753609462e-05,
+      "loss": 0.9531,
+      "step": 5194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6389784218580785e-05,
+      "loss": 0.8477,
+      "step": 5195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388340676072784e-05,
+      "loss": 0.9453,
+      "step": 5196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386896908621438e-05,
+      "loss": 0.832,
+      "step": 5197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385452916277596e-05,
+      "loss": 0.8359,
+      "step": 5198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384008699092097e-05,
+      "loss": 0.9336,
+      "step": 5199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382564257115794e-05,
+      "loss": 0.9727,
+      "step": 5200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638111959039955e-05,
+      "loss": 0.8711,
+      "step": 5201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637967469899423e-05,
+      "loss": 0.793,
+      "step": 5202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637822958295071e-05,
+      "loss": 0.8047,
+      "step": 5203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637678424231989e-05,
+      "loss": 0.7891,
+      "step": 5204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375338677152642e-05,
+      "loss": 0.9922,
+      "step": 5205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637389288749988e-05,
+      "loss": 0.8516,
+      "step": 5206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372446873412506e-05,
+      "loss": 0.9531,
+      "step": 5207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637100063494144e-05,
+      "loss": 0.9766,
+      "step": 5208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369554172137604e-05,
+      "loss": 0.9727,
+      "step": 5209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6368107485051932e-05,
+      "loss": 0.8789,
+      "step": 5210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366660573735362e-05,
+      "loss": 0.8203,
+      "step": 5211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365213438238844e-05,
+      "loss": 0.957,
+      "step": 5212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636376607861333e-05,
+      "loss": 0.875,
+      "step": 5213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636231849490979e-05,
+      "loss": 0.9609,
+      "step": 5214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360870687179188e-05,
+      "loss": 0.9844,
+      "step": 5215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6359422655472514e-05,
+      "loss": 0.707,
+      "step": 5216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357974399840748e-05,
+      "loss": 0.9414,
+      "step": 5217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356525920334884e-05,
+      "loss": 0.9375,
+      "step": 5218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355077217005927e-05,
+      "loss": 0.8555,
+      "step": 5219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353628289904895e-05,
+      "loss": 1.0078,
+      "step": 5220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6352179139082795e-05,
+      "loss": 0.8359,
+      "step": 5221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350729764590663e-05,
+      "loss": 0.8203,
+      "step": 5222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634928016647953e-05,
+      "loss": 0.8594,
+      "step": 5223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634783034480044e-05,
+      "loss": 0.8398,
+      "step": 5224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634638029960444e-05,
+      "loss": 0.9531,
+      "step": 5225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344930030942593e-05,
+      "loss": 0.957,
+      "step": 5226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343479538865962e-05,
+      "loss": 0.8477,
+      "step": 5227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634202882342562e-05,
+      "loss": 1.0078,
+      "step": 5228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6340577884672653e-05,
+      "loss": 0.8867,
+      "step": 5229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339126722658148e-05,
+      "loss": 0.8945,
+      "step": 5230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337675337433206e-05,
+      "loss": 0.8828,
+      "step": 5231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633622372904893e-05,
+      "loss": 1.0391,
+      "step": 5232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633477189755643e-05,
+      "loss": 0.8984,
+      "step": 5233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333319843006834e-05,
+      "loss": 0.9492,
+      "step": 5234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331867565451266e-05,
+      "loss": 0.9219,
+      "step": 5235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330415064940864e-05,
+      "loss": 0.8086,
+      "step": 5236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328962341526776e-05,
+      "loss": 1.0234,
+      "step": 5237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632750939526015e-05,
+      "loss": 0.957,
+      "step": 5238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6326056226192148e-05,
+      "loss": 0.9883,
+      "step": 5239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632460283437394e-05,
+      "loss": 0.918,
+      "step": 5240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6323149219856702e-05,
+      "loss": 0.8164,
+      "step": 5241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6321695382691616e-05,
+      "loss": 0.918,
+      "step": 5242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6320241322929873e-05,
+      "loss": 0.875,
+      "step": 5243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318787040622675e-05,
+      "loss": 0.8711,
+      "step": 5244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6317332535821236e-05,
+      "loss": 0.9492,
+      "step": 5245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315877808576758e-05,
+      "loss": 0.8711,
+      "step": 5246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631442285894047e-05,
+      "loss": 0.9102,
+      "step": 5247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6312967686963606e-05,
+      "loss": 0.8828,
+      "step": 5248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311512292697403e-05,
+      "loss": 0.9141,
+      "step": 5249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6310056676193105e-05,
+      "loss": 0.9648,
+      "step": 5250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6308600837501968e-05,
+      "loss": 0.8555,
+      "step": 5251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630714477667526e-05,
+      "loss": 0.8047,
+      "step": 5252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305688493764242e-05,
+      "loss": 0.8242,
+      "step": 5253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304231988820198e-05,
+      "loss": 0.8711,
+      "step": 5254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630277526189441e-05,
+      "loss": 0.9102,
+      "step": 5255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6301318313038175e-05,
+      "loss": 0.875,
+      "step": 5256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629986114230279e-05,
+      "loss": 1.0391,
+      "step": 5257
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298403749739573e-05,
+      "loss": 0.8477,
+      "step": 5258
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6296946135399835e-05,
+      "loss": 0.9531,
+      "step": 5259
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295488299334895e-05,
+      "loss": 0.8477,
+      "step": 5260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6294030241596096e-05,
+      "loss": 0.8164,
+      "step": 5261
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6292571962234772e-05,
+      "loss": 0.8359,
+      "step": 5262
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291113461302277e-05,
+      "loss": 0.8867,
+      "step": 5263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6289654738849962e-05,
+      "loss": 0.8867,
+      "step": 5264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628819579492919e-05,
+      "loss": 0.9297,
+      "step": 5265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628673662959134e-05,
+      "loss": 0.9219,
+      "step": 5266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285277242887784e-05,
+      "loss": 0.875,
+      "step": 5267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628381763486991e-05,
+      "loss": 0.793,
+      "step": 5268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282357805589117e-05,
+      "loss": 0.8438,
+      "step": 5269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280897755096805e-05,
+      "loss": 0.9961,
+      "step": 5270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279437483444383e-05,
+      "loss": 1.0391,
+      "step": 5271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6277976990683274e-05,
+      "loss": 0.875,
+      "step": 5272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276516276864905e-05,
+      "loss": 1.0078,
+      "step": 5273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275055342040703e-05,
+      "loss": 0.8594,
+      "step": 5274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273594186262115e-05,
+      "loss": 0.9414,
+      "step": 5275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627213280958059e-05,
+      "loss": 0.8438,
+      "step": 5276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6270671212047587e-05,
+      "loss": 0.9727,
+      "step": 5277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269209393714563e-05,
+      "loss": 0.9922,
+      "step": 5278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6267747354633e-05,
+      "loss": 0.8672,
+      "step": 5279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266285094854374e-05,
+      "loss": 0.9844,
+      "step": 5280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6264822614430176e-05,
+      "loss": 0.9375,
+      "step": 5281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62633599134119e-05,
+      "loss": 0.8828,
+      "step": 5282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626189699185105e-05,
+      "loss": 0.9453,
+      "step": 5283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626043384979914e-05,
+      "loss": 0.8711,
+      "step": 5284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625897048730769e-05,
+      "loss": 0.9453,
+      "step": 5285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625750690442822e-05,
+      "loss": 0.9531,
+      "step": 5286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256043101212274e-05,
+      "loss": 0.8984,
+      "step": 5287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254579077711388e-05,
+      "loss": 0.8906,
+      "step": 5288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625311483397712e-05,
+      "loss": 0.9062,
+      "step": 5289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625165037006102e-05,
+      "loss": 0.8633,
+      "step": 5290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250185686014662e-05,
+      "loss": 0.9062,
+      "step": 5291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6248720781889612e-05,
+      "loss": 0.8945,
+      "step": 5292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247255657737458e-05,
+      "loss": 0.7344,
+      "step": 5293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245790313609787e-05,
+      "loss": 0.8398,
+      "step": 5294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244324749558193e-05,
+      "loss": 0.9062,
+      "step": 5295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242858965634285e-05,
+      "loss": 0.9922,
+      "step": 5296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624139296188967e-05,
+      "loss": 0.8672,
+      "step": 5297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623992673837598e-05,
+      "loss": 0.8086,
+      "step": 5298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623846029514483e-05,
+      "loss": 0.8984,
+      "step": 5299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623699363224786e-05,
+      "loss": 0.8867,
+      "step": 5300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6235526749736714e-05,
+      "loss": 0.8672,
+      "step": 5301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234059647663047e-05,
+      "loss": 0.9258,
+      "step": 5302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232592326078512e-05,
+      "loss": 0.9492,
+      "step": 5303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231124785034777e-05,
+      "loss": 0.9531,
+      "step": 5304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622965702458352e-05,
+      "loss": 1.0,
+      "step": 5305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622818904477642e-05,
+      "loss": 1.0391,
+      "step": 5306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622672084566517e-05,
+      "loss": 0.875,
+      "step": 5307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225252427301458e-05,
+      "loss": 0.8281,
+      "step": 5308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6223783789736998e-05,
+      "loss": 0.875,
+      "step": 5309
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622231493302351e-05,
+      "loss": 0.957,
+      "step": 5310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220845857212696e-05,
+      "loss": 0.8164,
+      "step": 5311
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62193765623563e-05,
+      "loss": 0.8906,
+      "step": 5312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621790704850605e-05,
+      "loss": 0.8984,
+      "step": 5313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6216437315713693e-05,
+      "loss": 0.8828,
+      "step": 5314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621496736403098e-05,
+      "loss": 1.0469,
+      "step": 5315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621349719350967e-05,
+      "loss": 0.9023,
+      "step": 5316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621202680420153e-05,
+      "loss": 0.9414,
+      "step": 5317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210556196158336e-05,
+      "loss": 0.8438,
+      "step": 5318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6209085369431868e-05,
+      "loss": 0.8633,
+      "step": 5319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207614324073917e-05,
+      "loss": 0.9492,
+      "step": 5320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620614306013628e-05,
+      "loss": 1.0156,
+      "step": 5321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620467157767077e-05,
+      "loss": 0.8633,
+      "step": 5322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620319987672919e-05,
+      "loss": 0.8398,
+      "step": 5323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620172795736336e-05,
+      "loss": 0.8711,
+      "step": 5324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200255819625118e-05,
+      "loss": 0.9453,
+      "step": 5325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198783463566297e-05,
+      "loss": 0.9102,
+      "step": 5326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197310889238734e-05,
+      "loss": 0.9219,
+      "step": 5327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619583809669429e-05,
+      "loss": 0.9648,
+      "step": 5328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619436508598482e-05,
+      "loss": 0.7578,
+      "step": 5329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619289185716219e-05,
+      "loss": 0.8945,
+      "step": 5330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6191418410278275e-05,
+      "loss": 0.9141,
+      "step": 5331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618994474538496e-05,
+      "loss": 0.8984,
+      "step": 5332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188470862534135e-05,
+      "loss": 0.8711,
+      "step": 5333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186996761777694e-05,
+      "loss": 0.7266,
+      "step": 5334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185522443167543e-05,
+      "loss": 0.9844,
+      "step": 5335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61840479067556e-05,
+      "loss": 0.8945,
+      "step": 5336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182573152593778e-05,
+      "loss": 0.9219,
+      "step": 5337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618109818073401e-05,
+      "loss": 0.9688,
+      "step": 5338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617962299122823e-05,
+      "loss": 0.8789,
+      "step": 5339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178147584128386e-05,
+      "loss": 0.8672,
+      "step": 5340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6176671959486428e-05,
+      "loss": 0.957,
+      "step": 5341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617519611735431e-05,
+      "loss": 0.875,
+      "step": 5342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6173720057784007e-05,
+      "loss": 0.8281,
+      "step": 5343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6172243780827485e-05,
+      "loss": 0.8242,
+      "step": 5344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617076728653673e-05,
+      "loss": 0.8398,
+      "step": 5345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616929057496373e-05,
+      "loss": 0.8086,
+      "step": 5346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167813646160487e-05,
+      "loss": 0.9062,
+      "step": 5347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166336500179e-05,
+      "loss": 0.8203,
+      "step": 5348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6164859137071282e-05,
+      "loss": 0.8828,
+      "step": 5349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163381556889356e-05,
+      "loss": 0.9141,
+      "step": 5350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161903759685253e-05,
+      "loss": 0.9805,
+      "step": 5351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6160425745511e-05,
+      "loss": 0.9297,
+      "step": 5352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6158947514418647e-05,
+      "loss": 0.9102,
+      "step": 5353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157469066460242e-05,
+      "loss": 0.875,
+      "step": 5354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615599040168785e-05,
+      "loss": 0.8828,
+      "step": 5355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154511520153523e-05,
+      "loss": 0.8516,
+      "step": 5356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153032421909344e-05,
+      "loss": 0.8398,
+      "step": 5357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151553107007395e-05,
+      "loss": 0.8242,
+      "step": 5358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6150073575499764e-05,
+      "loss": 0.8867,
+      "step": 5359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148593827438545e-05,
+      "loss": 0.9961,
+      "step": 5360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6147113862875844e-05,
+      "loss": 0.9219,
+      "step": 5361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145633681863776e-05,
+      "loss": 0.9062,
+      "step": 5362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144153284454455e-05,
+      "loss": 1.0547,
+      "step": 5363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6142672670700007e-05,
+      "loss": 0.9062,
+      "step": 5364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141191840652576e-05,
+      "loss": 0.8945,
+      "step": 5365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139710794364294e-05,
+      "loss": 1.0938,
+      "step": 5366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6138229531887316e-05,
+      "loss": 0.8867,
+      "step": 5367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61367480532738e-05,
+      "loss": 0.8672,
+      "step": 5368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613526635857591e-05,
+      "loss": 0.9375,
+      "step": 5369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613378444784582e-05,
+      "loss": 0.9688,
+      "step": 5370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6132302321135703e-05,
+      "loss": 0.8359,
+      "step": 5371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613081997849776e-05,
+      "loss": 0.8125,
+      "step": 5372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129337419984176e-05,
+      "loss": 0.9766,
+      "step": 5373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612785464564716e-05,
+      "loss": 0.957,
+      "step": 5374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612637165553892e-05,
+      "loss": 0.9062,
+      "step": 5375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612488844971167e-05,
+      "loss": 0.8281,
+      "step": 5376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6123405028217648e-05,
+      "loss": 1.0469,
+      "step": 5377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6121921391109074e-05,
+      "loss": 0.9297,
+      "step": 5378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61204375384382e-05,
+      "loss": 0.8047,
+      "step": 5379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118953470257267e-05,
+      "loss": 0.957,
+      "step": 5380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611746918661854e-05,
+      "loss": 0.8633,
+      "step": 5381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115984687574274e-05,
+      "loss": 0.8438,
+      "step": 5382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114499973176743e-05,
+      "loss": 0.9297,
+      "step": 5383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611301504347823e-05,
+      "loss": 0.8164,
+      "step": 5384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111529898531018e-05,
+      "loss": 0.8594,
+      "step": 5385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61100445383874e-05,
+      "loss": 0.9844,
+      "step": 5386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6108558963099682e-05,
+      "loss": 0.918,
+      "step": 5387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610707317272017e-05,
+      "loss": 0.9141,
+      "step": 5388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105587167301186e-05,
+      "loss": 0.8359,
+      "step": 5389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104100946895046e-05,
+      "loss": 0.8672,
+      "step": 5390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610261451155409e-05,
+      "loss": 0.9609,
+      "step": 5391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610112786133065e-05,
+      "loss": 0.9688,
+      "step": 5392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099640996277083e-05,
+      "loss": 0.9883,
+      "step": 5393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098153916445735e-05,
+      "loss": 0.8281,
+      "step": 5394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096666621888976e-05,
+      "loss": 0.9492,
+      "step": 5395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609517911265917e-05,
+      "loss": 0.9609,
+      "step": 5396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093691388808696e-05,
+      "loss": 0.9102,
+      "step": 5397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092203450389942e-05,
+      "loss": 1.0078,
+      "step": 5398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090715297455297e-05,
+      "loss": 0.9219,
+      "step": 5399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608922693005716e-05,
+      "loss": 0.9297,
+      "step": 5400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087738348247946e-05,
+      "loss": 0.8828,
+      "step": 5401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608624955208006e-05,
+      "loss": 0.9297,
+      "step": 5402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084760541605934e-05,
+      "loss": 0.9453,
+      "step": 5403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608327131687799e-05,
+      "loss": 0.832,
+      "step": 5404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081781877948674e-05,
+      "loss": 0.8672,
+      "step": 5405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608029222487043e-05,
+      "loss": 0.9609,
+      "step": 5406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078802357695703e-05,
+      "loss": 0.8789,
+      "step": 5407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077312276476964e-05,
+      "loss": 0.9961,
+      "step": 5408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075821981266673e-05,
+      "loss": 0.9336,
+      "step": 5409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074331472117314e-05,
+      "loss": 0.8828,
+      "step": 5410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607284074908136e-05,
+      "loss": 1.0234,
+      "step": 5411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071349812211308e-05,
+      "loss": 0.8828,
+      "step": 5412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069858661559656e-05,
+      "loss": 0.8477,
+      "step": 5413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068367297178907e-05,
+      "loss": 0.9062,
+      "step": 5414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606687571912158e-05,
+      "loss": 0.9141,
+      "step": 5415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065383927440186e-05,
+      "loss": 0.9336,
+      "step": 5416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606389192218726e-05,
+      "loss": 0.8711,
+      "step": 5417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6062399703415338e-05,
+      "loss": 1.0,
+      "step": 5418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6060907271176964e-05,
+      "loss": 0.8828,
+      "step": 5419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059414625524686e-05,
+      "loss": 0.8672,
+      "step": 5420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057921766511063e-05,
+      "loss": 0.9414,
+      "step": 5421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605642869418866e-05,
+      "loss": 0.8047,
+      "step": 5422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6054935408610056e-05,
+      "loss": 0.7773,
+      "step": 5423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053441909827823e-05,
+      "loss": 0.8672,
+      "step": 5424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6051948197894553e-05,
+      "loss": 0.8789,
+      "step": 5425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050454272862847e-05,
+      "loss": 0.9805,
+      "step": 5426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6048960134785302e-05,
+      "loss": 0.9219,
+      "step": 5427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604746578371453e-05,
+      "loss": 0.875,
+      "step": 5428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604597121970315e-05,
+      "loss": 0.7148,
+      "step": 5429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6044476442803793e-05,
+      "loss": 0.8867,
+      "step": 5430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042981453069084e-05,
+      "loss": 0.9531,
+      "step": 5431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604148625055167e-05,
+      "loss": 0.8711,
+      "step": 5432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039990835304192e-05,
+      "loss": 0.9414,
+      "step": 5433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6038495207379316e-05,
+      "loss": 0.9961,
+      "step": 5434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6036999366829698e-05,
+      "loss": 0.7969,
+      "step": 5435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035503313708014e-05,
+      "loss": 1.0078,
+      "step": 5436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6034007048066936e-05,
+      "loss": 0.8359,
+      "step": 5437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032510569959153e-05,
+      "loss": 0.8359,
+      "step": 5438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031013879437363e-05,
+      "loss": 0.8906,
+      "step": 5439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602951697655426e-05,
+      "loss": 0.8438,
+      "step": 5440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028019861362554e-05,
+      "loss": 0.9609,
+      "step": 5441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6026522533914963e-05,
+      "loss": 0.8711,
+      "step": 5442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025024994264206e-05,
+      "loss": 0.8594,
+      "step": 5443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602352724246302e-05,
+      "loss": 0.9219,
+      "step": 5444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022029278564138e-05,
+      "loss": 0.832,
+      "step": 5445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020531102620303e-05,
+      "loss": 0.8516,
+      "step": 5446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601903271468428e-05,
+      "loss": 0.9375,
+      "step": 5447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017534114808818e-05,
+      "loss": 0.8906,
+      "step": 5448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6016035303046687e-05,
+      "loss": 1.0703,
+      "step": 5449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014536279450666e-05,
+      "loss": 0.8164,
+      "step": 5450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013037044073535e-05,
+      "loss": 0.8477,
+      "step": 5451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011537596968088e-05,
+      "loss": 0.9141,
+      "step": 5452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6010037938187118e-05,
+      "loss": 0.9219,
+      "step": 5453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6008538067783434e-05,
+      "loss": 0.8672,
+      "step": 5454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600703798580985e-05,
+      "loss": 0.8789,
+      "step": 5455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600553769231918e-05,
+      "loss": 1.1094,
+      "step": 5456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004037187364253e-05,
+      "loss": 0.8477,
+      "step": 5457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600253647099791e-05,
+      "loss": 0.8906,
+      "step": 5458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600103554327299e-05,
+      "loss": 0.8984,
+      "step": 5459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5999534404242342e-05,
+      "loss": 0.8945,
+      "step": 5460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998033053958826e-05,
+      "loss": 0.8789,
+      "step": 5461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996531492475306e-05,
+      "loss": 0.918,
+      "step": 5462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599502971984465e-05,
+      "loss": 0.8398,
+      "step": 5463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993527736119742e-05,
+      "loss": 0.9297,
+      "step": 5464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599202554135347e-05,
+      "loss": 0.8359,
+      "step": 5465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990523135598725e-05,
+      "loss": 0.9375,
+      "step": 5466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989020518908413e-05,
+      "loss": 0.9336,
+      "step": 5467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987517691335443e-05,
+      "loss": 0.9258,
+      "step": 5468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598601465293273e-05,
+      "loss": 0.9102,
+      "step": 5469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984511403753192e-05,
+      "loss": 1.0703,
+      "step": 5470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983007943849774e-05,
+      "loss": 0.957,
+      "step": 5471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981504273275404e-05,
+      "loss": 0.918,
+      "step": 5472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598000039208304e-05,
+      "loss": 0.957,
+      "step": 5473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978496300325623e-05,
+      "loss": 0.9336,
+      "step": 5474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976991998056125e-05,
+      "loss": 0.8594,
+      "step": 5475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5975487485327508e-05,
+      "loss": 0.8359,
+      "step": 5476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597398276219275e-05,
+      "loss": 0.9219,
+      "step": 5477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972477828704835e-05,
+      "loss": 0.9336,
+      "step": 5478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970972684916754e-05,
+      "loss": 1.0469,
+      "step": 5479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969467330881508e-05,
+      "loss": 0.957,
+      "step": 5480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59679617666521e-05,
+      "loss": 0.8789,
+      "step": 5481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596645599228154e-05,
+      "loss": 0.8438,
+      "step": 5482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5964950007822854e-05,
+      "loss": 1.0078,
+      "step": 5483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596344381332907e-05,
+      "loss": 1.0,
+      "step": 5484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596193740885322e-05,
+      "loss": 0.8633,
+      "step": 5485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596043079444835e-05,
+      "loss": 0.9961,
+      "step": 5486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595892397016751e-05,
+      "loss": 0.8359,
+      "step": 5487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5957416936063755e-05,
+      "loss": 0.9062,
+      "step": 5488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595590969219015e-05,
+      "loss": 0.832,
+      "step": 5489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595440223859977e-05,
+      "loss": 0.8672,
+      "step": 5490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952894575345694e-05,
+      "loss": 0.8438,
+      "step": 5491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951386702481008e-05,
+      "loss": 0.9141,
+      "step": 5492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949878620058808e-05,
+      "loss": 0.9219,
+      "step": 5493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948370328132194e-05,
+      "loss": 0.918,
+      "step": 5494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594686182675428e-05,
+      "loss": 1.1719,
+      "step": 5495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945353115978173e-05,
+      "loss": 0.793,
+      "step": 5496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943844195857008e-05,
+      "loss": 0.8555,
+      "step": 5497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594233506644391e-05,
+      "loss": 0.8438,
+      "step": 5498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940825727792023e-05,
+      "loss": 0.9258,
+      "step": 5499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939316179954483e-05,
+      "loss": 0.7969,
+      "step": 5500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5937806422984454e-05,
+      "loss": 0.9531,
+      "step": 5501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936296456935092e-05,
+      "loss": 0.9336,
+      "step": 5502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934786281859568e-05,
+      "loss": 0.9727,
+      "step": 5503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5933275897811055e-05,
+      "loss": 0.8711,
+      "step": 5504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931765304842736e-05,
+      "loss": 0.8867,
+      "step": 5505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5930254503007798e-05,
+      "loss": 1.0078,
+      "step": 5506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592874349235945e-05,
+      "loss": 0.9219,
+      "step": 5507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927232272950885e-05,
+      "loss": 0.8711,
+      "step": 5508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592572084483532e-05,
+      "loss": 0.8789,
+      "step": 5509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924209208065977e-05,
+      "loss": 0.7695,
+      "step": 5510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592269736269608e-05,
+      "loss": 0.8555,
+      "step": 5511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5921185308778863e-05,
+      "loss": 0.9531,
+      "step": 5512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5919673046367572e-05,
+      "loss": 0.9805,
+      "step": 5513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591816057551545e-05,
+      "loss": 0.7344,
+      "step": 5514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916647896275758e-05,
+      "loss": 0.9141,
+      "step": 5515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915135008701758e-05,
+      "loss": 0.7969,
+      "step": 5516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591362191284672e-05,
+      "loss": 0.875,
+      "step": 5517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912108608763926e-05,
+      "loss": 0.832,
+      "step": 5518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910595096506658e-05,
+      "loss": 0.8594,
+      "step": 5519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5909081376128214e-05,
+      "loss": 0.9531,
+      "step": 5520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907567447681886e-05,
+      "loss": 0.8555,
+      "step": 5521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590605331122099e-05,
+      "loss": 0.8359,
+      "step": 5522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5904538966798836e-05,
+      "loss": 0.9844,
+      "step": 5523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590302441446875e-05,
+      "loss": 0.8672,
+      "step": 5524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901509654284062e-05,
+      "loss": 0.8516,
+      "step": 5525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899994686298102e-05,
+      "loss": 0.8984,
+      "step": 5526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898479510564223e-05,
+      "loss": 0.9414,
+      "step": 5527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5896964127135778e-05,
+      "loss": 1.0,
+      "step": 5528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895448536066114e-05,
+      "loss": 0.8359,
+      "step": 5529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5893932737408606e-05,
+      "loss": 0.8906,
+      "step": 5530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892416731216626e-05,
+      "loss": 0.8594,
+      "step": 5531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589090051754356e-05,
+      "loss": 0.8516,
+      "step": 5532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5889384096442782e-05,
+      "loss": 0.957,
+      "step": 5533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887867467967703e-05,
+      "loss": 0.8828,
+      "step": 5534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5886350632171718e-05,
+      "loss": 0.918,
+      "step": 5535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884833589108238e-05,
+      "loss": 0.832,
+      "step": 5536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883316338830687e-05,
+      "loss": 0.7266,
+      "step": 5537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881798881392475e-05,
+      "loss": 0.9219,
+      "step": 5538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880281216847047e-05,
+      "loss": 0.8867,
+      "step": 5539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587876334524784e-05,
+      "loss": 0.9102,
+      "step": 5540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877245266648297e-05,
+      "loss": 0.9922,
+      "step": 5541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5875726981101873e-05,
+      "loss": 0.8477,
+      "step": 5542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587420848866203e-05,
+      "loss": 0.9375,
+      "step": 5543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587268978938224e-05,
+      "loss": 0.9922,
+      "step": 5544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587117088331597e-05,
+      "loss": 0.8906,
+      "step": 5545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586965177051671e-05,
+      "loss": 0.8945,
+      "step": 5546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586813245103795e-05,
+      "loss": 0.8906,
+      "step": 5547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866612924933186e-05,
+      "loss": 0.918,
+      "step": 5548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586509319225592e-05,
+      "loss": 0.8906,
+      "step": 5549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586357325305967e-05,
+      "loss": 0.957,
+      "step": 5550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586205310739795e-05,
+      "loss": 1.0312,
+      "step": 5551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860532755324293e-05,
+      "loss": 0.7812,
+      "step": 5552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859012196892224e-05,
+      "loss": 0.8008,
+      "step": 5553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5857491432155292e-05,
+      "loss": 0.9844,
+      "step": 5554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585597046116704e-05,
+      "loss": 0.9531,
+      "step": 5555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585444928398103e-05,
+      "loss": 0.8867,
+      "step": 5556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585292790065082e-05,
+      "loss": 0.875,
+      "step": 5557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851406311229983e-05,
+      "loss": 1.0078,
+      "step": 5558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849884515772092e-05,
+      "loss": 0.8828,
+      "step": 5559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848362514330737e-05,
+      "loss": 0.8398,
+      "step": 5560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584684030695951e-05,
+      "loss": 0.9648,
+      "step": 5561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845317893712007e-05,
+      "loss": 0.9531,
+      "step": 5562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843795274641833e-05,
+      "loss": 0.957,
+      "step": 5563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584227244980261e-05,
+      "loss": 0.8281,
+      "step": 5564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584074941924795e-05,
+      "loss": 0.9062,
+      "step": 5565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839226183031484e-05,
+      "loss": 0.8867,
+      "step": 5566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5837702741206848e-05,
+      "loss": 1.0625,
+      "step": 5567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583617909382769e-05,
+      "loss": 0.8555,
+      "step": 5568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583465524094765e-05,
+      "loss": 0.8164,
+      "step": 5569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833131182620392e-05,
+      "loss": 0.8984,
+      "step": 5570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583160691889958e-05,
+      "loss": 1.0,
+      "step": 5571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830082449838883e-05,
+      "loss": 0.8945,
+      "step": 5572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582855777549198e-05,
+      "loss": 0.9766,
+      "step": 5573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5827032895912563e-05,
+      "loss": 0.9336,
+      "step": 5574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825507811154318e-05,
+      "loss": 0.9336,
+      "step": 5575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823982521270947e-05,
+      "loss": 0.8633,
+      "step": 5576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582245702631616e-05,
+      "loss": 0.9531,
+      "step": 5577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5820931326343673e-05,
+      "loss": 0.7969,
+      "step": 5578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819405421407202e-05,
+      "loss": 0.8398,
+      "step": 5579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817879311560486e-05,
+      "loss": 0.875,
+      "step": 5580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816352996857255e-05,
+      "loss": 0.9844,
+      "step": 5581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814826477351256e-05,
+      "loss": 0.9609,
+      "step": 5582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5813299753096235e-05,
+      "loss": 0.9609,
+      "step": 5583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811772824145958e-05,
+      "loss": 0.8047,
+      "step": 5584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5810245690554184e-05,
+      "loss": 0.8594,
+      "step": 5585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5808718352374687e-05,
+      "loss": 0.9336,
+      "step": 5586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807190809661253e-05,
+      "loss": 0.8945,
+      "step": 5587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580566306246766e-05,
+      "loss": 0.9375,
+      "step": 5588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5804135110847708e-05,
+      "loss": 1.0078,
+      "step": 5589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802606954855196e-05,
+      "loss": 0.9375,
+      "step": 5590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5801078594543935e-05,
+      "loss": 0.9805,
+      "step": 5591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579955002996774e-05,
+      "loss": 0.8945,
+      "step": 5592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798021261180435e-05,
+      "loss": 0.8906,
+      "step": 5593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796492288235847e-05,
+      "loss": 0.8945,
+      "step": 5594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579496311118782e-05,
+      "loss": 0.957,
+      "step": 5595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579343373009019e-05,
+      "loss": 0.9023,
+      "step": 5596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791904144996816e-05,
+      "loss": 0.8594,
+      "step": 5597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579037435596155e-05,
+      "loss": 0.7852,
+      "step": 5598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5788844363038268e-05,
+      "loss": 0.8555,
+      "step": 5599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5787314166280833e-05,
+      "loss": 0.9648,
+      "step": 5600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785783765743135e-05,
+      "loss": 0.8438,
+      "step": 5601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5784253161479054e-05,
+      "loss": 0.8945,
+      "step": 5602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782722353542493e-05,
+      "loss": 0.9258,
+      "step": 5603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781191341987346e-05,
+      "loss": 1.0938,
+      "step": 5604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5779660126867526e-05,
+      "loss": 1.0234,
+      "step": 5605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577812870823695e-05,
+      "loss": 0.9414,
+      "step": 5606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5776597086149538e-05,
+      "loss": 0.9766,
+      "step": 5607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5775065260659226e-05,
+      "loss": 0.8242,
+      "step": 5608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577353323181995e-05,
+      "loss": 0.875,
+      "step": 5609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772000999685653e-05,
+      "loss": 0.8398,
+      "step": 5610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577046856431029e-05,
+      "loss": 1.1172,
+      "step": 5611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576893592574782e-05,
+      "loss": 0.8906,
+      "step": 5612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5767403084052207e-05,
+      "loss": 0.9336,
+      "step": 5613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5765870039277428e-05,
+      "loss": 0.8633,
+      "step": 5614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764336791477458e-05,
+      "loss": 0.9648,
+      "step": 5615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762803340706297e-05,
+      "loss": 0.832,
+      "step": 5616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5761269687017927e-05,
+      "loss": 0.9062,
+      "step": 5617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575973583046636e-05,
+      "loss": 0.8711,
+      "step": 5618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5758201771105595e-05,
+      "loss": 0.8672,
+      "step": 5619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575666750898966e-05,
+      "loss": 0.9258,
+      "step": 5620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755133044172573e-05,
+      "loss": 0.8672,
+      "step": 5621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753598376708365e-05,
+      "loss": 0.8711,
+      "step": 5622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752063506651077e-05,
+      "loss": 0.8203,
+      "step": 5623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5750528434054747e-05,
+      "loss": 0.8633,
+      "step": 5624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574899315897344e-05,
+      "loss": 0.9922,
+      "step": 5625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747457681461198e-05,
+      "loss": 0.9805,
+      "step": 5626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745922001572105e-05,
+      "loss": 0.8438,
+      "step": 5627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744386119360223e-05,
+      "loss": 0.8789,
+      "step": 5628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5742850034879636e-05,
+      "loss": 0.918,
+      "step": 5629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574131374818443e-05,
+      "loss": 0.9609,
+      "step": 5630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5739777259328707e-05,
+      "loss": 0.918,
+      "step": 5631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738240568366562e-05,
+      "loss": 0.8516,
+      "step": 5632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736703675352108e-05,
+      "loss": 0.9609,
+      "step": 5633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573516658033946e-05,
+      "loss": 0.9453,
+      "step": 5634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573362928338274e-05,
+      "loss": 0.9727,
+      "step": 5635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573209178453608e-05,
+      "loss": 0.9414,
+      "step": 5636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573055408385362e-05,
+      "loss": 0.8711,
+      "step": 5637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729016181389502e-05,
+      "loss": 0.9297,
+      "step": 5638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572747807719788e-05,
+      "loss": 0.9336,
+      "step": 5639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5725939771332913e-05,
+      "loss": 0.9531,
+      "step": 5640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572440126384876e-05,
+      "loss": 0.8203,
+      "step": 5641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722862554799605e-05,
+      "loss": 0.9805,
+      "step": 5642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572132364423962e-05,
+      "loss": 0.8945,
+      "step": 5643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5719784532223e-05,
+      "loss": 0.9727,
+      "step": 5644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5718245218803933e-05,
+      "loss": 0.9102,
+      "step": 5645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571670570403662e-05,
+      "loss": 0.9023,
+      "step": 5646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715165987975275e-05,
+      "loss": 0.875,
+      "step": 5647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5713626070674114e-05,
+      "loss": 0.8125,
+      "step": 5648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571208595218735e-05,
+      "loss": 0.875,
+      "step": 5649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571054563256923e-05,
+      "loss": 0.9102,
+      "step": 5650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5709005111873972e-05,
+      "loss": 0.9023,
+      "step": 5651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707464390155837e-05,
+      "loss": 1.0,
+      "step": 5652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705923467469065e-05,
+      "loss": 0.9453,
+      "step": 5653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5704382343867918e-05,
+      "loss": 1.0625,
+      "step": 5654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570284101940666e-05,
+      "loss": 0.8633,
+      "step": 5655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5701299494139566e-05,
+      "loss": 0.8945,
+      "step": 5656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5699757768120912e-05,
+      "loss": 0.8633,
+      "step": 5657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569821584140499e-05,
+      "loss": 0.918,
+      "step": 5658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569667371404609e-05,
+      "loss": 0.9531,
+      "step": 5659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569513138609851e-05,
+      "loss": 0.9727,
+      "step": 5660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569358885761656e-05,
+      "loss": 0.9258,
+      "step": 5661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5692046128654558e-05,
+      "loss": 0.9141,
+      "step": 5662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5690503199266825e-05,
+      "loss": 0.957,
+      "step": 5663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688960069507685e-05,
+      "loss": 0.8398,
+      "step": 5664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687416739431478e-05,
+      "loss": 0.9844,
+      "step": 5665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5685873209092547e-05,
+      "loss": 0.9102,
+      "step": 5666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568432947854524e-05,
+      "loss": 0.8359,
+      "step": 5667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682785547843916e-05,
+      "loss": 1.0312,
+      "step": 5668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568124141704294e-05,
+      "loss": 1.0859,
+      "step": 5669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567969708619668e-05,
+      "loss": 0.9492,
+      "step": 5670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5678152555359517e-05,
+      "loss": 0.8984,
+      "step": 5671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676607824585833e-05,
+      "loss": 0.8906,
+      "step": 5672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675062893930025e-05,
+      "loss": 0.9414,
+      "step": 5673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5673517763446494e-05,
+      "loss": 0.9609,
+      "step": 5674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671972433189636e-05,
+      "loss": 1.1328,
+      "step": 5675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5670426903213874e-05,
+      "loss": 0.9805,
+      "step": 5676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668881173573625e-05,
+      "loss": 0.8086,
+      "step": 5677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566733524432332e-05,
+      "loss": 0.8086,
+      "step": 5678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566578911551739e-05,
+      "loss": 0.9219,
+      "step": 5679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664242787210276e-05,
+      "loss": 0.8047,
+      "step": 5680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5662696259456428e-05,
+      "loss": 0.8984,
+      "step": 5681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56611495323103e-05,
+      "loss": 0.875,
+      "step": 5682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659602605826357e-05,
+      "loss": 0.8672,
+      "step": 5683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658055480059072e-05,
+      "loss": 0.9141,
+      "step": 5684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5656508155062913e-05,
+      "loss": 0.8516,
+      "step": 5685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654960630892368e-05,
+      "loss": 0.8594,
+      "step": 5686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565341290760193e-05,
+      "loss": 0.8398,
+      "step": 5687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5651864985246095e-05,
+      "loss": 1.0547,
+      "step": 5688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5650316863879367e-05,
+      "loss": 0.8438,
+      "step": 5689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648768543556256e-05,
+      "loss": 0.8711,
+      "step": 5690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647220024331285e-05,
+      "loss": 0.8555,
+      "step": 5691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5645671306258978e-05,
+      "loss": 0.8086,
+      "step": 5692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644122389393865e-05,
+      "loss": 1.0391,
+      "step": 5693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564257327379049e-05,
+      "loss": 1.0703,
+      "step": 5694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56410239595034e-05,
+      "loss": 0.8789,
+      "step": 5695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563947444658714e-05,
+      "loss": 1.0625,
+      "step": 5696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637924735096283e-05,
+      "loss": 0.8906,
+      "step": 5697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563637482508539e-05,
+      "loss": 0.9492,
+      "step": 5698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5634824716609037e-05,
+      "loss": 0.8906,
+      "step": 5699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5633274409721804e-05,
+      "loss": 0.957,
+      "step": 5700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5631723904478282e-05,
+      "loss": 0.9375,
+      "step": 5701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563017320093307e-05,
+      "loss": 0.9766,
+      "step": 5702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562862229914076e-05,
+      "loss": 0.8828,
+      "step": 5703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562707119915597e-05,
+      "loss": 0.8594,
+      "step": 5704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562551990103332e-05,
+      "loss": 0.9219,
+      "step": 5705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623968404827424e-05,
+      "loss": 0.7656,
+      "step": 5706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622416710592916e-05,
+      "loss": 0.8477,
+      "step": 5707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562086481838444e-05,
+      "loss": 0.75,
+      "step": 5708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619312728256628e-05,
+      "loss": 0.9766,
+      "step": 5709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617760440264143e-05,
+      "loss": 0.8008,
+      "step": 5710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561620795446164e-05,
+      "loss": 1.0625,
+      "step": 5711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614655270903778e-05,
+      "loss": 0.8477,
+      "step": 5712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561310238964524e-05,
+      "loss": 0.8555,
+      "step": 5713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56115493107407e-05,
+      "loss": 0.832,
+      "step": 5714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5609996034244844e-05,
+      "loss": 0.9648,
+      "step": 5715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608442560212364e-05,
+      "loss": 0.9297,
+      "step": 5716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5606888888697964e-05,
+      "loss": 0.8711,
+      "step": 5717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560533501975635e-05,
+      "loss": 0.9219,
+      "step": 5718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603780953442234e-05,
+      "loss": 0.9453,
+      "step": 5719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560222668981034e-05,
+      "loss": 0.8984,
+      "step": 5720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560067222891539e-05,
+      "loss": 0.9883,
+      "step": 5721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599117570812126e-05,
+      "loss": 0.9141,
+      "step": 5722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597562715555288e-05,
+      "loss": 1.0625,
+      "step": 5723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596007663199625e-05,
+      "loss": 0.918,
+      "step": 5724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559445241379989e-05,
+      "loss": 1.0469,
+      "step": 5725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592896967410847e-05,
+      "loss": 1.0156,
+      "step": 5726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591341324087267e-05,
+      "loss": 0.8359,
+      "step": 5727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558978548388393e-05,
+      "loss": 0.8086,
+      "step": 5728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5588229446855613e-05,
+      "loss": 1.0703,
+      "step": 5729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586673213057105e-05,
+      "loss": 0.8555,
+      "step": 5730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558511678254321e-05,
+      "loss": 0.9609,
+      "step": 5731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558356015536873e-05,
+      "loss": 0.9727,
+      "step": 5732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5582003331588478e-05,
+      "loss": 0.9141,
+      "step": 5733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580446311257266e-05,
+      "loss": 0.9766,
+      "step": 5734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578889094429927e-05,
+      "loss": 0.9141,
+      "step": 5735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5577331681161286e-05,
+      "loss": 0.9414,
+      "step": 5736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557577407150619e-05,
+      "loss": 1.1172,
+      "step": 5737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5574216265519477e-05,
+      "loss": 0.9453,
+      "step": 5738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572658263256003e-05,
+      "loss": 0.8672,
+      "step": 5739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571100064770625e-05,
+      "loss": 0.9648,
+      "step": 5740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569541670118213e-05,
+      "loss": 0.8125,
+      "step": 5741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5567983079353642e-05,
+      "loss": 0.9609,
+      "step": 5742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566424292531786e-05,
+      "loss": 0.9336,
+      "step": 5743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564865309707543e-05,
+      "loss": 0.8672,
+      "step": 5744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556330613093579e-05,
+      "loss": 0.9023,
+      "step": 5745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556174675627145e-05,
+      "loss": 0.9609,
+      "step": 5746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556018718576941e-05,
+      "loss": 0.8555,
+      "step": 5747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558627419484598e-05,
+      "loss": 1.0703,
+      "step": 5748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555706745747193e-05,
+      "loss": 0.9336,
+      "step": 5749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555550729978634e-05,
+      "loss": 0.9258,
+      "step": 5750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553946946482756e-05,
+      "loss": 0.7734,
+      "step": 5751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555238639761613e-05,
+      "loss": 1.0625,
+      "step": 5752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550825653241403e-05,
+      "loss": 0.9258,
+      "step": 5753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5549264713413536e-05,
+      "loss": 0.9492,
+      "step": 5754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554770357818749e-05,
+      "loss": 0.8594,
+      "step": 5755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546142247618236e-05,
+      "loss": 0.9453,
+      "step": 5756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544580721760744e-05,
+      "loss": 0.7422,
+      "step": 5757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5543019000670012e-05,
+      "loss": 0.7539,
+      "step": 5758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554145708440102e-05,
+      "loss": 0.832,
+      "step": 5759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539894973008766e-05,
+      "loss": 0.9297,
+      "step": 5760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553833266654826e-05,
+      "loss": 0.7852,
+      "step": 5761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536770165074505e-05,
+      "loss": 0.8438,
+      "step": 5762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5535207468642526e-05,
+      "loss": 0.9414,
+      "step": 5763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533644577307347e-05,
+      "loss": 0.9922,
+      "step": 5764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5532081491123995e-05,
+      "loss": 0.9492,
+      "step": 5765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5530518210147518e-05,
+      "loss": 0.9062,
+      "step": 5766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5528954734432952e-05,
+      "loss": 0.9023,
+      "step": 5767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527391064035354e-05,
+      "loss": 0.9102,
+      "step": 5768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552582719900978e-05,
+      "loss": 0.8867,
+      "step": 5769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55242631394113e-05,
+      "loss": 1.0078,
+      "step": 5770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5522698885294988e-05,
+      "loss": 0.8828,
+      "step": 5771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552113443671592e-05,
+      "loss": 0.9414,
+      "step": 5772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519569793729182e-05,
+      "loss": 0.8359,
+      "step": 5773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518004956389868e-05,
+      "loss": 1.0,
+      "step": 5774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551643992475308e-05,
+      "loss": 0.8008,
+      "step": 5775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551487469887393e-05,
+      "loss": 0.8828,
+      "step": 5776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5513309278807526e-05,
+      "loss": 0.8633,
+      "step": 5777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511743664608986e-05,
+      "loss": 0.918,
+      "step": 5778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551017785633344e-05,
+      "loss": 0.8438,
+      "step": 5779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5508611854036032e-05,
+      "loss": 0.9453,
+      "step": 5780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507045657771887e-05,
+      "loss": 0.9219,
+      "step": 5781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5505479267596165e-05,
+      "loss": 0.8281,
+      "step": 5782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550391268356402e-05,
+      "loss": 1.1484,
+      "step": 5783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502345905730608e-05,
+      "loss": 0.9023,
+      "step": 5784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500778934151098e-05,
+      "loss": 0.7969,
+      "step": 5785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499211768880676e-05,
+      "loss": 1.0859,
+      "step": 5786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497644409974513e-05,
+      "loss": 0.7734,
+      "step": 5787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54960768574878e-05,
+      "loss": 0.875,
+      "step": 5788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549450911147574e-05,
+      "loss": 0.8242,
+      "step": 5789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492941171993527e-05,
+      "loss": 0.8906,
+      "step": 5790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491373039096376e-05,
+      "loss": 0.9492,
+      "step": 5791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54898047128395e-05,
+      "loss": 0.875,
+      "step": 5792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5488236193278123e-05,
+      "loss": 0.8281,
+      "step": 5793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5486667480467478e-05,
+      "loss": 0.957,
+      "step": 5794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54850985744628e-05,
+      "loss": 0.8516,
+      "step": 5795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483529475319333e-05,
+      "loss": 0.9219,
+      "step": 5796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5481960183092323e-05,
+      "loss": 0.8398,
+      "step": 5797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5480390697837033e-05,
+      "loss": 0.9141,
+      "step": 5798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478821019608724e-05,
+      "loss": 0.9297,
+      "step": 5799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547725114846267e-05,
+      "loss": 0.957,
+      "step": 5800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475681084454145e-05,
+      "loss": 0.9219,
+      "step": 5801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5474110827638433e-05,
+      "loss": 0.9648,
+      "step": 5802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472540378070826e-05,
+      "loss": 0.7812,
+      "step": 5803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547096973580663e-05,
+      "loss": 0.8047,
+      "step": 5804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469398900901135e-05,
+      "loss": 0.9414,
+      "step": 5805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467827873409657e-05,
+      "loss": 0.9727,
+      "step": 5806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466256653387525e-05,
+      "loss": 0.8828,
+      "step": 5807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546468524089005e-05,
+      "loss": 0.8477,
+      "step": 5808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463113635972577e-05,
+      "loss": 0.8164,
+      "step": 5809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546154183869043e-05,
+      "loss": 0.8984,
+      "step": 5810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545996984909897e-05,
+      "loss": 0.9102,
+      "step": 5811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545839766725354e-05,
+      "loss": 0.832,
+      "step": 5812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54568252932095e-05,
+      "loss": 0.8438,
+      "step": 5813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5455252727022212e-05,
+      "loss": 0.8906,
+      "step": 5814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545367996874706e-05,
+      "loss": 1.0859,
+      "step": 5815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5452107018439406e-05,
+      "loss": 0.9141,
+      "step": 5816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450533876154656e-05,
+      "loss": 0.8672,
+      "step": 5817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448960541948184e-05,
+      "loss": 0.918,
+      "step": 5818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5447387015875405e-05,
+      "loss": 0.875,
+      "step": 5819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445813297991713e-05,
+      "loss": 0.875,
+      "step": 5820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5444239388352528e-05,
+      "loss": 0.8516,
+      "step": 5821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5442665287013267e-05,
+      "loss": 0.8555,
+      "step": 5822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441090994029362e-05,
+      "loss": 0.8945,
+      "step": 5823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439516509456237e-05,
+      "loss": 0.875,
+      "step": 5824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437941833349342e-05,
+      "loss": 0.8867,
+      "step": 5825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436366965764117e-05,
+      "loss": 0.9883,
+      "step": 5826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5434791906756016e-05,
+      "loss": 0.9609,
+      "step": 5827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433216656380502e-05,
+      "loss": 1.0234,
+      "step": 5828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5431641214693042e-05,
+      "loss": 0.8203,
+      "step": 5829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430065581749104e-05,
+      "loss": 0.7266,
+      "step": 5830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428489757604177e-05,
+      "loss": 0.9297,
+      "step": 5831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426913742313744e-05,
+      "loss": 0.8984,
+      "step": 5832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425337535933298e-05,
+      "loss": 0.9688,
+      "step": 5833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542376113851834e-05,
+      "loss": 1.0078,
+      "step": 5834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422184550124378e-05,
+      "loss": 0.8828,
+      "step": 5835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420607770806926e-05,
+      "loss": 0.8477,
+      "step": 5836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541903080062151e-05,
+      "loss": 0.9492,
+      "step": 5837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417453639623646e-05,
+      "loss": 0.8945,
+      "step": 5838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5415876287868877e-05,
+      "loss": 0.7969,
+      "step": 5839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541429874541274e-05,
+      "loss": 0.957,
+      "step": 5840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5412721012310787e-05,
+      "loss": 0.9297,
+      "step": 5841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541114308861857e-05,
+      "loss": 0.9609,
+      "step": 5842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540956497439165e-05,
+      "loss": 0.9531,
+      "step": 5843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407986669685594e-05,
+      "loss": 0.8555,
+      "step": 5844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.8984,
+      "step": 5845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540482948905838e-05,
+      "loss": 0.8633,
+      "step": 5846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403250613248397e-05,
+      "loss": 0.9062,
+      "step": 5847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401671547181613e-05,
+      "loss": 0.9531,
+      "step": 5848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540009229091363e-05,
+      "loss": 0.832,
+      "step": 5849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398512844500066e-05,
+      "loss": 0.9648,
+      "step": 5850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396933207996528e-05,
+      "loss": 0.957,
+      "step": 5851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5395353381458638e-05,
+      "loss": 0.9375,
+      "step": 5852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393773364942025e-05,
+      "loss": 0.8281,
+      "step": 5853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5392193158502323e-05,
+      "loss": 0.9062,
+      "step": 5854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539061276219518e-05,
+      "loss": 0.832,
+      "step": 5855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389032176076234e-05,
+      "loss": 0.8008,
+      "step": 5856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5387451400201144e-05,
+      "loss": 0.8945,
+      "step": 5857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538587043462557e-05,
+      "loss": 0.9023,
+      "step": 5858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5384289279405186e-05,
+      "loss": 0.9141,
+      "step": 5859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382707934595663e-05,
+      "loss": 0.9844,
+      "step": 5860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538112640025268e-05,
+      "loss": 0.9531,
+      "step": 5861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537954467643193e-05,
+      "loss": 0.8945,
+      "step": 5862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377962763189103e-05,
+      "loss": 0.9141,
+      "step": 5863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5376380660579907e-05,
+      "loss": 0.9805,
+      "step": 5864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374798368660047e-05,
+      "loss": 0.875,
+      "step": 5865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373215887485237e-05,
+      "loss": 0.8477,
+      "step": 5866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5371633217111198e-05,
+      "loss": 0.8984,
+      "step": 5867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537005035759366e-05,
+      "loss": 0.8984,
+      "step": 5868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536846730898836e-05,
+      "loss": 0.9023,
+      "step": 5869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5366884071351033e-05,
+      "loss": 0.9141,
+      "step": 5870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365300644737434e-05,
+      "loss": 0.8125,
+      "step": 5871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363717029203317e-05,
+      "loss": 1.0234,
+      "step": 5872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362133224804438e-05,
+      "loss": 0.9648,
+      "step": 5873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360549231596572e-05,
+      "loss": 0.8633,
+      "step": 5874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535896504963549e-05,
+      "loss": 0.8945,
+      "step": 5875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357380678976977e-05,
+      "loss": 0.8281,
+      "step": 5876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355796119676813e-05,
+      "loss": 0.9062,
+      "step": 5877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53542113717908e-05,
+      "loss": 0.8555,
+      "step": 5878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535262643537474e-05,
+      "loss": 0.8633,
+      "step": 5879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351041310484438e-05,
+      "loss": 0.918,
+      "step": 5880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534945599717571e-05,
+      "loss": 0.8672,
+      "step": 5881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347870495504375e-05,
+      "loss": 1.0781,
+      "step": 5882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346284805526265e-05,
+      "loss": 0.8984,
+      "step": 5883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534469892729721e-05,
+      "loss": 0.8789,
+      "step": 5884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343112860873054e-05,
+      "loss": 0.8633,
+      "step": 5885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341526606309646e-05,
+      "loss": 0.9688,
+      "step": 5886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5339940163662837e-05,
+      "loss": 0.8125,
+      "step": 5887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5338353532988492e-05,
+      "loss": 0.8477,
+      "step": 5888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336766714342474e-05,
+      "loss": 0.7969,
+      "step": 5889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5335179707780664e-05,
+      "loss": 0.9062,
+      "step": 5890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333592513358934e-05,
+      "loss": 1.1172,
+      "step": 5891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533200513113318e-05,
+      "loss": 1.0312,
+      "step": 5892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5330417561159288e-05,
+      "loss": 0.9062,
+      "step": 5893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328829803493174e-05,
+      "loss": 0.9219,
+      "step": 5894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5327241858190726e-05,
+      "loss": 0.8555,
+      "step": 5895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325653725307868e-05,
+      "loss": 0.8711,
+      "step": 5896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532406540490052e-05,
+      "loss": 0.7852,
+      "step": 5897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532247689702461e-05,
+      "loss": 0.9766,
+      "step": 5898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532088820173607e-05,
+      "loss": 1.0703,
+      "step": 5899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531929931909084e-05,
+      "loss": 0.8633,
+      "step": 5900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531771024914487e-05,
+      "loss": 0.8867,
+      "step": 5901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531612099195411e-05,
+      "loss": 0.8594,
+      "step": 5902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5314531547574524e-05,
+      "loss": 0.9141,
+      "step": 5903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5312941916062076e-05,
+      "loss": 0.9062,
+      "step": 5904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531135209747274e-05,
+      "loss": 0.8906,
+      "step": 5905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5309762091862495e-05,
+      "loss": 0.875,
+      "step": 5906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530817189928733e-05,
+      "loss": 0.8711,
+      "step": 5907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306581519803233e-05,
+      "loss": 0.832,
+      "step": 5908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304990953466207e-05,
+      "loss": 0.8281,
+      "step": 5909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303400200332264e-05,
+      "loss": 0.9219,
+      "step": 5910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301809260457408e-05,
+      "loss": 0.9414,
+      "step": 5911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5300218133897663e-05,
+      "loss": 0.8164,
+      "step": 5912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298626820709054e-05,
+      "loss": 0.7188,
+      "step": 5913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5297035320947613e-05,
+      "loss": 0.875,
+      "step": 5914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529544363466938e-05,
+      "loss": 0.9922,
+      "step": 5915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52938517619304e-05,
+      "loss": 0.9531,
+      "step": 5916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529225970278673e-05,
+      "loss": 1.125,
+      "step": 5917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529066745729442e-05,
+      "loss": 1.0312,
+      "step": 5918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528907502550954e-05,
+      "loss": 0.9766,
+      "step": 5919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528748240748816e-05,
+      "loss": 0.8164,
+      "step": 5920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5285889603286364e-05,
+      "loss": 0.8984,
+      "step": 5921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284296612960235e-05,
+      "loss": 0.8359,
+      "step": 5922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282703436565863e-05,
+      "loss": 0.9297,
+      "step": 5923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5281110074159343e-05,
+      "loss": 0.8164,
+      "step": 5924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527951652579679e-05,
+      "loss": 0.8438,
+      "step": 5925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52779227915343e-05,
+      "loss": 1.1328,
+      "step": 5926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5276328871428002e-05,
+      "loss": 0.7852,
+      "step": 5927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527473476553402e-05,
+      "loss": 1.0156,
+      "step": 5928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273140473908484e-05,
+      "loss": 0.8984,
+      "step": 5929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271545996607525e-05,
+      "loss": 0.8984,
+      "step": 5930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52699513336873e-05,
+      "loss": 1.0547,
+      "step": 5931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268356485203942e-05,
+      "loss": 0.7617,
+      "step": 5932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5266761451213625e-05,
+      "loss": 0.9258,
+      "step": 5933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5265166231772505e-05,
+      "loss": 0.9219,
+      "step": 5934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526357082693675e-05,
+      "loss": 0.9297,
+      "step": 5935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526197523676254e-05,
+      "loss": 1.0078,
+      "step": 5936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260379461306055e-05,
+      "loss": 0.8594,
+      "step": 5937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258783500623491e-05,
+      "loss": 0.9023,
+      "step": 5938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257187354771042e-05,
+      "loss": 0.8633,
+      "step": 5939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5255591023804903e-05,
+      "loss": 0.9453,
+      "step": 5940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253994507781294e-05,
+      "loss": 0.7695,
+      "step": 5941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252397806756426e-05,
+      "loss": 0.9258,
+      "step": 5942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525080092078652e-05,
+      "loss": 0.8281,
+      "step": 5943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5249203849927807e-05,
+      "loss": 0.9297,
+      "step": 5944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524760659423652e-05,
+      "loss": 0.8008,
+      "step": 5945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5246009153768903e-05,
+      "loss": 0.8555,
+      "step": 5946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244411528581208e-05,
+      "loss": 0.7617,
+      "step": 5947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242813718729682e-05,
+      "loss": 0.8945,
+      "step": 5948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524121572427059e-05,
+      "loss": 0.8438,
+      "step": 5949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239617545260201e-05,
+      "loss": 0.9062,
+      "step": 5950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238019181754789e-05,
+      "loss": 0.7617,
+      "step": 5951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236420633810633e-05,
+      "loss": 0.8828,
+      "step": 5952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523482190148402e-05,
+      "loss": 1.1094,
+      "step": 5953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5233222984831249e-05,
+      "loss": 0.9102,
+      "step": 5954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231623883908614e-05,
+      "loss": 0.8008,
+      "step": 5955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230024598772424e-05,
+      "loss": 1.0156,
+      "step": 5956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5228425129478996e-05,
+      "loss": 0.8906,
+      "step": 5957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522682547608464e-05,
+      "loss": 0.8477,
+      "step": 5958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225225638645698e-05,
+      "loss": 0.8945,
+      "step": 5959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223625617218485e-05,
+      "loss": 0.8398,
+      "step": 5960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222025411859353e-05,
+      "loss": 0.9805,
+      "step": 5961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220425022624642e-05,
+      "loss": 0.9844,
+      "step": 5962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218824449570709e-05,
+      "loss": 0.957,
+      "step": 5963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217223692753904e-05,
+      "loss": 0.9727,
+      "step": 5964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215622752230603e-05,
+      "loss": 0.9141,
+      "step": 5965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214021628057168e-05,
+      "loss": 0.8203,
+      "step": 5966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212420320289983e-05,
+      "loss": 0.9414,
+      "step": 5967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5210818828985429e-05,
+      "loss": 0.9883,
+      "step": 5968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.52092171541999e-05,
+      "loss": 0.9219,
+      "step": 5969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207615295989793e-05,
+      "loss": 0.9844,
+      "step": 5970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206013254411506e-05,
+      "loss": 0.7891,
+      "step": 5971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520441102952146e-05,
+      "loss": 0.8906,
+      "step": 5972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202808621376063e-05,
+      "loss": 0.9688,
+      "step": 5973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520120603003174e-05,
+      "loss": 0.9219,
+      "step": 5974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199603255544927e-05,
+      "loss": 1.0156,
+      "step": 5975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198000297972053e-05,
+      "loss": 0.8086,
+      "step": 5976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196397157369558e-05,
+      "loss": 0.8594,
+      "step": 5977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194793833793902e-05,
+      "loss": 0.918,
+      "step": 5978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193190327301532e-05,
+      "loss": 1.0859,
+      "step": 5979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191586637948912e-05,
+      "loss": 0.8125,
+      "step": 5980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518998276579251e-05,
+      "loss": 0.7617,
+      "step": 5981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188378710888805e-05,
+      "loss": 0.9688,
+      "step": 5982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186774473294273e-05,
+      "loss": 0.8047,
+      "step": 5983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185170053065404e-05,
+      "loss": 1.0625,
+      "step": 5984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183565450258689e-05,
+      "loss": 0.8906,
+      "step": 5985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181960664930634e-05,
+      "loss": 0.9453,
+      "step": 5986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180355697137738e-05,
+      "loss": 0.8711,
+      "step": 5987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178750546936526e-05,
+      "loss": 0.8516,
+      "step": 5988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177145214383506e-05,
+      "loss": 0.9609,
+      "step": 5989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175539699535211e-05,
+      "loss": 0.9297,
+      "step": 5990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173934002448171e-05,
+      "loss": 0.8594,
+      "step": 5991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517232812317893e-05,
+      "loss": 0.9453,
+      "step": 5992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170722061784029e-05,
+      "loss": 0.8477,
+      "step": 5993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169115818320022e-05,
+      "loss": 0.9062,
+      "step": 5994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516750939284346e-05,
+      "loss": 0.7656,
+      "step": 5995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516590278541092e-05,
+      "loss": 0.8828,
+      "step": 5996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164295996078961e-05,
+      "loss": 0.9023,
+      "step": 5997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162689024904171e-05,
+      "loss": 0.9961,
+      "step": 5998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161081871943127e-05,
+      "loss": 0.793,
+      "step": 5999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515947453725242e-05,
+      "loss": 0.8945,
+      "step": 6000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157867020888653e-05,
+      "loss": 0.8438,
+      "step": 6001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156259322908421e-05,
+      "loss": 0.9219,
+      "step": 6002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515465144336834e-05,
+      "loss": 0.8789,
+      "step": 6003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515304338232502e-05,
+      "loss": 0.8672,
+      "step": 6004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515143513983509e-05,
+      "loss": 0.9062,
+      "step": 6005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514982671595517e-05,
+      "loss": 0.8516,
+      "step": 6006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148218110741902e-05,
+      "loss": 1.0,
+      "step": 6007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146609324251925e-05,
+      "loss": 0.8203,
+      "step": 6008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145000356541888e-05,
+      "loss": 0.8828,
+      "step": 6009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143391207668443e-05,
+      "loss": 1.2031,
+      "step": 6010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141781877688255e-05,
+      "loss": 0.8594,
+      "step": 6011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140172366657986e-05,
+      "loss": 0.9375,
+      "step": 6012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138562674634314e-05,
+      "loss": 0.9023,
+      "step": 6013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136952801673914e-05,
+      "loss": 0.9492,
+      "step": 6014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135342747833477e-05,
+      "loss": 0.9023,
+      "step": 6015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133732513169691e-05,
+      "loss": 0.957,
+      "step": 6016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5132122097739258e-05,
+      "loss": 0.9844,
+      "step": 6017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513051150159888e-05,
+      "loss": 1.0625,
+      "step": 6018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128900724805271e-05,
+      "loss": 0.875,
+      "step": 6019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512728976741515e-05,
+      "loss": 0.875,
+      "step": 6020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125678629485237e-05,
+      "loss": 0.957,
+      "step": 6021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124067311072268e-05,
+      "loss": 0.9492,
+      "step": 6022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122455812232975e-05,
+      "loss": 0.8828,
+      "step": 6023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120844133024103e-05,
+      "loss": 0.8672,
+      "step": 6024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119232273502407e-05,
+      "loss": 0.9297,
+      "step": 6025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117620233724637e-05,
+      "loss": 0.8906,
+      "step": 6026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116008013747556e-05,
+      "loss": 0.9336,
+      "step": 6027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114395613627935e-05,
+      "loss": 0.8945,
+      "step": 6028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112783033422547e-05,
+      "loss": 0.8438,
+      "step": 6029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111170273188171e-05,
+      "loss": 0.9258,
+      "step": 6030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109557332981604e-05,
+      "loss": 0.9609,
+      "step": 6031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107944212859629e-05,
+      "loss": 0.8555,
+      "step": 6032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106330912879058e-05,
+      "loss": 0.75,
+      "step": 6033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104717433096685e-05,
+      "loss": 0.8555,
+      "step": 6034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103103773569334e-05,
+      "loss": 0.7852,
+      "step": 6035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101489934353817e-05,
+      "loss": 0.9258,
+      "step": 6036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099875915506963e-05,
+      "loss": 0.8555,
+      "step": 6037
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098261717085604e-05,
+      "loss": 0.8906,
+      "step": 6038
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509664733914658e-05,
+      "loss": 0.9492,
+      "step": 6039
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095032781746732e-05,
+      "loss": 0.8906,
+      "step": 6040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093418044942914e-05,
+      "loss": 0.9258,
+      "step": 6041
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509180312879198e-05,
+      "loss": 0.9844,
+      "step": 6042
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5090188033350801e-05,
+      "loss": 0.8828,
+      "step": 6043
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088572758676238e-05,
+      "loss": 0.8633,
+      "step": 6044
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086957304825171e-05,
+      "loss": 0.8242,
+      "step": 6045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085341671854485e-05,
+      "loss": 0.8672,
+      "step": 6046
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508372585982107e-05,
+      "loss": 0.8984,
+      "step": 6047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082109868781811e-05,
+      "loss": 0.8555,
+      "step": 6048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080493698793623e-05,
+      "loss": 0.7734,
+      "step": 6049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078877349913406e-05,
+      "loss": 0.8984,
+      "step": 6050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077260822198074e-05,
+      "loss": 0.9062,
+      "step": 6051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507564411570455e-05,
+      "loss": 0.9766,
+      "step": 6052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074027230489763e-05,
+      "loss": 0.9492,
+      "step": 6053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072410166610638e-05,
+      "loss": 0.8203,
+      "step": 6054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070792924124122e-05,
+      "loss": 0.875,
+      "step": 6055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5069175503087157e-05,
+      "loss": 0.8906,
+      "step": 6056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067557903556696e-05,
+      "loss": 0.8438,
+      "step": 6057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5065940125589697e-05,
+      "loss": 0.7227,
+      "step": 6058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064322169243127e-05,
+      "loss": 0.9609,
+      "step": 6059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506270403457395e-05,
+      "loss": 0.9766,
+      "step": 6060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061085721639148e-05,
+      "loss": 0.9219,
+      "step": 6061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059467230495703e-05,
+      "loss": 0.9453,
+      "step": 6062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057848561200608e-05,
+      "loss": 0.8945,
+      "step": 6063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056229713810852e-05,
+      "loss": 0.9609,
+      "step": 6064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054610688383445e-05,
+      "loss": 0.9219,
+      "step": 6065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052991484975387e-05,
+      "loss": 0.8516,
+      "step": 6066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50513721036437e-05,
+      "loss": 0.8945,
+      "step": 6067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049752544445402e-05,
+      "loss": 0.8828,
+      "step": 6068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048132807437518e-05,
+      "loss": 0.9531,
+      "step": 6069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5046512892677085e-05,
+      "loss": 0.9609,
+      "step": 6070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5044892800221143e-05,
+      "loss": 0.8711,
+      "step": 6071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043272530126733e-05,
+      "loss": 0.9336,
+      "step": 6072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5041652082450913e-05,
+      "loss": 0.8438,
+      "step": 6073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040031457250736e-05,
+      "loss": 0.8281,
+      "step": 6074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038410654583273e-05,
+      "loss": 0.8594,
+      "step": 6075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036789674505588e-05,
+      "loss": 0.9102,
+      "step": 6076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5035168517074765e-05,
+      "loss": 0.8789,
+      "step": 6077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503354718234788e-05,
+      "loss": 0.832,
+      "step": 6078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5031925670382033e-05,
+      "loss": 0.9883,
+      "step": 6079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503030398123431e-05,
+      "loss": 0.9531,
+      "step": 6080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028682114961819e-05,
+      "loss": 0.9062,
+      "step": 6081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027060071621665e-05,
+      "loss": 0.9492,
+      "step": 6082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5025437851270965e-05,
+      "loss": 0.9336,
+      "step": 6083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023815453966837e-05,
+      "loss": 0.875,
+      "step": 6084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022192879766412e-05,
+      "loss": 0.9844,
+      "step": 6085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5020570128726822e-05,
+      "loss": 0.9414,
+      "step": 6086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018947200905203e-05,
+      "loss": 0.8281,
+      "step": 6087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017324096358707e-05,
+      "loss": 0.8008,
+      "step": 6088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501570081514448e-05,
+      "loss": 0.8086,
+      "step": 6089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014077357319688e-05,
+      "loss": 0.957,
+      "step": 6090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012453722941484e-05,
+      "loss": 0.9961,
+      "step": 6091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501082991206705e-05,
+      "loss": 0.8438,
+      "step": 6092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009205924753554e-05,
+      "loss": 0.9844,
+      "step": 6093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007581761058188e-05,
+      "loss": 0.9219,
+      "step": 6094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005957421038134e-05,
+      "loss": 0.9023,
+      "step": 6095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500433290475059e-05,
+      "loss": 0.9141,
+      "step": 6096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002708212252756e-05,
+      "loss": 0.9609,
+      "step": 6097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001083343601845e-05,
+      "loss": 0.8398,
+      "step": 6098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4999458298855065e-05,
+      "loss": 0.8906,
+      "step": 6099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4997833078069644e-05,
+      "loss": 1.1172,
+      "step": 6100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996207681302799e-05,
+      "loss": 0.8047,
+      "step": 6101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994582108611768e-05,
+      "loss": 0.8477,
+      "step": 6102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499295636005379e-05,
+      "loss": 0.8164,
+      "step": 6103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991330435686111e-05,
+      "loss": 0.9141,
+      "step": 6104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989704335565977e-05,
+      "loss": 0.8945,
+      "step": 6105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988078059750652e-05,
+      "loss": 0.9062,
+      "step": 6106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986451608297396e-05,
+      "loss": 1.0703,
+      "step": 6107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4984824981263481e-05,
+      "loss": 0.8867,
+      "step": 6108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983198178706181e-05,
+      "loss": 0.8516,
+      "step": 6109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981571200682781e-05,
+      "loss": 0.832,
+      "step": 6110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4979944047250565e-05,
+      "loss": 0.9375,
+      "step": 6111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497831671846683e-05,
+      "loss": 0.8438,
+      "step": 6112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976689214388877e-05,
+      "loss": 0.8438,
+      "step": 6113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4975061535074016e-05,
+      "loss": 0.918,
+      "step": 6114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973433680579552e-05,
+      "loss": 0.7773,
+      "step": 6115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4971805650962813e-05,
+      "loss": 1.0,
+      "step": 6116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970177446281118e-05,
+      "loss": 0.8945,
+      "step": 6117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4968549066591802e-05,
+      "loss": 0.9102,
+      "step": 6118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966920511952202e-05,
+      "loss": 1.0078,
+      "step": 6119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4965291782419664e-05,
+      "loss": 0.8438,
+      "step": 6120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963662878051532e-05,
+      "loss": 0.8164,
+      "step": 6121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962033798905169e-05,
+      "loss": 0.9688,
+      "step": 6122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960404545037935e-05,
+      "loss": 0.8281,
+      "step": 6123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958775116507196e-05,
+      "loss": 0.8672,
+      "step": 6124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957145513370331e-05,
+      "loss": 0.957,
+      "step": 6125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4955515735684715e-05,
+      "loss": 0.9492,
+      "step": 6126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4953885783507743e-05,
+      "loss": 0.8906,
+      "step": 6127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952255656896801e-05,
+      "loss": 0.8047,
+      "step": 6128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950625355909291e-05,
+      "loss": 0.8242,
+      "step": 6129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494899488060262e-05,
+      "loss": 0.7812,
+      "step": 6130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947364231034196e-05,
+      "loss": 0.9102,
+      "step": 6131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945733407261442e-05,
+      "loss": 0.8164,
+      "step": 6132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4944102409341776e-05,
+      "loss": 0.9609,
+      "step": 6133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942471237332632e-05,
+      "loss": 0.9492,
+      "step": 6134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940839891291443e-05,
+      "loss": 0.9531,
+      "step": 6135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939208371275653e-05,
+      "loss": 0.9688,
+      "step": 6136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493757667734271e-05,
+      "loss": 0.9531,
+      "step": 6137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935944809550071e-05,
+      "loss": 0.9023,
+      "step": 6138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934312767955193e-05,
+      "loss": 0.8867,
+      "step": 6139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4932680552615542e-05,
+      "loss": 0.957,
+      "step": 6140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931048163588596e-05,
+      "loss": 0.9375,
+      "step": 6141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929415600931826e-05,
+      "loss": 0.8164,
+      "step": 6142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927782864702727e-05,
+      "loss": 0.8555,
+      "step": 6143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492614995495878e-05,
+      "loss": 0.7969,
+      "step": 6144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924516871757491e-05,
+      "loss": 0.8828,
+      "step": 6145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922883615156356e-05,
+      "loss": 0.8242,
+      "step": 6146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921250185212892e-05,
+      "loss": 0.8203,
+      "step": 6147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4919616581984606e-05,
+      "loss": 0.8398,
+      "step": 6148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917982805529028e-05,
+      "loss": 0.8164,
+      "step": 6149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916348855903679e-05,
+      "loss": 0.8047,
+      "step": 6150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49147147331661e-05,
+      "loss": 0.8867,
+      "step": 6151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913080437373824e-05,
+      "loss": 0.8633,
+      "step": 6152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911445968584398e-05,
+      "loss": 1.0938,
+      "step": 6153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4909811326855382e-05,
+      "loss": 0.8242,
+      "step": 6154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908176512244325e-05,
+      "loss": 1.0312,
+      "step": 6155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4906541524808793e-05,
+      "loss": 0.9062,
+      "step": 6156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904906364606362e-05,
+      "loss": 1.0156,
+      "step": 6157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903271031694601e-05,
+      "loss": 0.9922,
+      "step": 6158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49016355261311e-05,
+      "loss": 0.8633,
+      "step": 6159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899999847973444e-05,
+      "loss": 1.0,
+      "step": 6160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898363997279226e-05,
+      "loss": 1.0547,
+      "step": 6161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896727974106046e-05,
+      "loss": 0.9141,
+      "step": 6162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4895091778511518e-05,
+      "loss": 0.9375,
+      "step": 6163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893455410553248e-05,
+      "loss": 1.0,
+      "step": 6164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891818870288858e-05,
+      "loss": 0.8516,
+      "step": 6165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890182157775974e-05,
+      "loss": 0.8594,
+      "step": 6166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888545273072226e-05,
+      "loss": 0.9062,
+      "step": 6167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886908216235248e-05,
+      "loss": 0.9414,
+      "step": 6168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885270987322689e-05,
+      "loss": 0.9492,
+      "step": 6169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883633586392195e-05,
+      "loss": 0.9062,
+      "step": 6170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488199601350142e-05,
+      "loss": 0.8945,
+      "step": 6171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488035826870803e-05,
+      "loss": 0.8594,
+      "step": 6172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878720352069692e-05,
+      "loss": 0.9258,
+      "step": 6173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877082263644076e-05,
+      "loss": 0.9688,
+      "step": 6174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875444003488861e-05,
+      "loss": 0.8203,
+      "step": 6175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873805571661737e-05,
+      "loss": 0.9609,
+      "step": 6176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872166968220394e-05,
+      "loss": 0.9141,
+      "step": 6177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487052819322253e-05,
+      "loss": 0.8789,
+      "step": 6178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868889246725843e-05,
+      "loss": 1.0312,
+      "step": 6179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867250128788055e-05,
+      "loss": 0.8125,
+      "step": 6180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486561083946687e-05,
+      "loss": 0.8633,
+      "step": 6181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863971378820016e-05,
+      "loss": 0.9414,
+      "step": 6182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862331746905221e-05,
+      "loss": 0.7891,
+      "step": 6183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4860691943780219e-05,
+      "loss": 0.9062,
+      "step": 6184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859051969502747e-05,
+      "loss": 0.8359,
+      "step": 6185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857411824130555e-05,
+      "loss": 0.8828,
+      "step": 6186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4855771507721388e-05,
+      "loss": 0.9141,
+      "step": 6187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854131020333013e-05,
+      "loss": 0.9023,
+      "step": 6188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4852490362023191e-05,
+      "loss": 1.0547,
+      "step": 6189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850849532849687e-05,
+      "loss": 0.875,
+      "step": 6190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849208532870282e-05,
+      "loss": 0.9883,
+      "step": 6191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847567362142758e-05,
+      "loss": 0.9766,
+      "step": 6192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845926020724906e-05,
+      "loss": 0.8594,
+      "step": 6193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484428450867451e-05,
+      "loss": 0.8438,
+      "step": 6194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842642826049382e-05,
+      "loss": 0.8555,
+      "step": 6195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4841000972907323e-05,
+      "loss": 0.7969,
+      "step": 6196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839358949306144e-05,
+      "loss": 0.8984,
+      "step": 6197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837716755303664e-05,
+      "loss": 0.9062,
+      "step": 6198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836074390957709e-05,
+      "loss": 0.9766,
+      "step": 6199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834431856326106e-05,
+      "loss": 0.9453,
+      "step": 6200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4832789151466695e-05,
+      "loss": 0.9688,
+      "step": 6201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831146276437315e-05,
+      "loss": 0.9922,
+      "step": 6202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4829503231295817e-05,
+      "loss": 0.9141,
+      "step": 6203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4827860016100051e-05,
+      "loss": 0.8867,
+      "step": 6204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826216630907879e-05,
+      "loss": 0.8633,
+      "step": 6205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824573075777172e-05,
+      "loss": 0.8828,
+      "step": 6206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4822929350765794e-05,
+      "loss": 0.8867,
+      "step": 6207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821285455931629e-05,
+      "loss": 0.8633,
+      "step": 6208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4819641391332557e-05,
+      "loss": 0.8594,
+      "step": 6209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481799715702647e-05,
+      "loss": 0.9531,
+      "step": 6210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816352753071265e-05,
+      "loss": 0.8281,
+      "step": 6211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814708179524846e-05,
+      "loss": 1.125,
+      "step": 6212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813063436445115e-05,
+      "loss": 0.8516,
+      "step": 6213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481141852388999e-05,
+      "loss": 0.8438,
+      "step": 6214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480977344191739e-05,
+      "loss": 0.9492,
+      "step": 6215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480812819058524e-05,
+      "loss": 0.8633,
+      "step": 6216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806482769951474e-05,
+      "loss": 0.8633,
+      "step": 6217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804837180074028e-05,
+      "loss": 0.9844,
+      "step": 6218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803191421010845e-05,
+      "loss": 0.9023,
+      "step": 6219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801545492819878e-05,
+      "loss": 0.8555,
+      "step": 6220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799899395559078e-05,
+      "loss": 0.9102,
+      "step": 6221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798253129286413e-05,
+      "loss": 0.9844,
+      "step": 6222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796606694059844e-05,
+      "loss": 0.9102,
+      "step": 6223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794960089937347e-05,
+      "loss": 0.8672,
+      "step": 6224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793313316976903e-05,
+      "loss": 0.8828,
+      "step": 6225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791666375236498e-05,
+      "loss": 0.9297,
+      "step": 6226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790019264774116e-05,
+      "loss": 0.9922,
+      "step": 6227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4788371985647768e-05,
+      "loss": 0.8828,
+      "step": 6228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478672453791544e-05,
+      "loss": 0.8633,
+      "step": 6229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785076921635157e-05,
+      "loss": 0.8594,
+      "step": 6230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783429136864924e-05,
+      "loss": 0.9219,
+      "step": 6231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4781781183662768e-05,
+      "loss": 0.7969,
+      "step": 6232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780133062086712e-05,
+      "loss": 0.957,
+      "step": 6233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778484772194792e-05,
+      "loss": 0.8281,
+      "step": 6234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776836314045045e-05,
+      "loss": 0.8086,
+      "step": 6235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775187687695516e-05,
+      "loss": 0.9375,
+      "step": 6236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773538893204253e-05,
+      "loss": 0.7578,
+      "step": 6237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771889930629317e-05,
+      "loss": 1.0,
+      "step": 6238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770240800028767e-05,
+      "loss": 0.9219,
+      "step": 6239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768591501460676e-05,
+      "loss": 0.9727,
+      "step": 6240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766942034983114e-05,
+      "loss": 0.8672,
+      "step": 6241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4765292400654166e-05,
+      "loss": 0.8711,
+      "step": 6242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763642598531909e-05,
+      "loss": 0.9531,
+      "step": 6243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761992628674447e-05,
+      "loss": 0.8008,
+      "step": 6244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760342491139867e-05,
+      "loss": 0.9961,
+      "step": 6245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4758692185986282e-05,
+      "loss": 0.918,
+      "step": 6246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757041713271796e-05,
+      "loss": 0.7617,
+      "step": 6247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755391073054526e-05,
+      "loss": 0.9297,
+      "step": 6248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753740265392595e-05,
+      "loss": 0.8594,
+      "step": 6249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475208929034413e-05,
+      "loss": 0.8164,
+      "step": 6250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750438147967263e-05,
+      "loss": 0.8242,
+      "step": 6251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4748786838320133e-05,
+      "loss": 0.9609,
+      "step": 6252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747135361460889e-05,
+      "loss": 0.9609,
+      "step": 6253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745483717447675e-05,
+      "loss": 0.9414,
+      "step": 6254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743831906338658e-05,
+      "loss": 0.8984,
+      "step": 6255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4742179928191989e-05,
+      "loss": 0.8281,
+      "step": 6256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740527783065849e-05,
+      "loss": 0.8359,
+      "step": 6257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738875471018402e-05,
+      "loss": 0.9297,
+      "step": 6258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737222992107833e-05,
+      "loss": 0.875,
+      "step": 6259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473557034639233e-05,
+      "loss": 0.9414,
+      "step": 6260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733917533930084e-05,
+      "loss": 0.9648,
+      "step": 6261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473226455477929e-05,
+      "loss": 0.9102,
+      "step": 6262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730611408998157e-05,
+      "loss": 0.9023,
+      "step": 6263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472895809664489e-05,
+      "loss": 0.8867,
+      "step": 6264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472730461777771e-05,
+      "loss": 0.918,
+      "step": 6265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472565097245483e-05,
+      "loss": 0.9336,
+      "step": 6266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723997160734488e-05,
+      "loss": 0.9141,
+      "step": 6267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722343182674909e-05,
+      "loss": 0.9492,
+      "step": 6268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4720689038334337e-05,
+      "loss": 0.8516,
+      "step": 6269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719034727771014e-05,
+      "loss": 1.0078,
+      "step": 6270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4717380251043194e-05,
+      "loss": 0.9219,
+      "step": 6271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471572560820913e-05,
+      "loss": 1.0156,
+      "step": 6272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714070799327088e-05,
+      "loss": 0.9648,
+      "step": 6273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712415824455332e-05,
+      "loss": 0.9102,
+      "step": 6274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471076068365214e-05,
+      "loss": 0.9375,
+      "step": 6275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709105376975792e-05,
+      "loss": 0.8164,
+      "step": 6276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707449904484575e-05,
+      "loss": 0.9258,
+      "step": 6277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705794266236773e-05,
+      "loss": 0.9766,
+      "step": 6278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704138462290694e-05,
+      "loss": 0.9141,
+      "step": 6279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4702482492704633e-05,
+      "loss": 0.8203,
+      "step": 6280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700826357536906e-05,
+      "loss": 0.8945,
+      "step": 6281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699170056845822e-05,
+      "loss": 0.9023,
+      "step": 6282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697513590689707e-05,
+      "loss": 0.8477,
+      "step": 6283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4695856959126884e-05,
+      "loss": 0.8672,
+      "step": 6284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694200162215689e-05,
+      "loss": 0.875,
+      "step": 6285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692543200014455e-05,
+      "loss": 0.8516,
+      "step": 6286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690886072581534e-05,
+      "loss": 0.9961,
+      "step": 6287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689228779975266e-05,
+      "loss": 0.8906,
+      "step": 6288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4687571322254017e-05,
+      "loss": 0.832,
+      "step": 6289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685913699476144e-05,
+      "loss": 0.957,
+      "step": 6290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684255911700014e-05,
+      "loss": 0.8789,
+      "step": 6291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4682597958984e-05,
+      "loss": 0.9375,
+      "step": 6292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680939841386484e-05,
+      "loss": 0.8242,
+      "step": 6293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679281558965846e-05,
+      "loss": 0.875,
+      "step": 6294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4677623111780481e-05,
+      "loss": 0.8008,
+      "step": 6295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675964499888782e-05,
+      "loss": 0.9141,
+      "step": 6296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674305723349159e-05,
+      "loss": 0.8086,
+      "step": 6297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672646782220009e-05,
+      "loss": 0.9258,
+      "step": 6298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670987676559754e-05,
+      "loss": 0.8828,
+      "step": 6299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4669328406426808e-05,
+      "loss": 0.9922,
+      "step": 6300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4667668971879601e-05,
+      "loss": 1.0234,
+      "step": 6301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666009372976562e-05,
+      "loss": 0.8711,
+      "step": 6302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4664349609776132e-05,
+      "loss": 0.9336,
+      "step": 6303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662689682336746e-05,
+      "loss": 1.0859,
+      "step": 6304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661029590716859e-05,
+      "loss": 0.793,
+      "step": 6305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659369334974924e-05,
+      "loss": 0.875,
+      "step": 6306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46577089151694e-05,
+      "loss": 0.9961,
+      "step": 6307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656048331358755e-05,
+      "loss": 0.8359,
+      "step": 6308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654387583601456e-05,
+      "loss": 0.9141,
+      "step": 6309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652726671955988e-05,
+      "loss": 1.0391,
+      "step": 6310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651065596480827e-05,
+      "loss": 0.8594,
+      "step": 6311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649404357234467e-05,
+      "loss": 0.9609,
+      "step": 6312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46477429542754e-05,
+      "loss": 0.8945,
+      "step": 6313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646081387662128e-05,
+      "loss": 0.8086,
+      "step": 6314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4644419657453155e-05,
+      "loss": 1.0078,
+      "step": 6315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642757763706998e-05,
+      "loss": 0.9297,
+      "step": 6316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641095706482169e-05,
+      "loss": 0.9297,
+      "step": 6317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639433485837195e-05,
+      "loss": 0.9102,
+      "step": 6318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637771101830604e-05,
+      "loss": 0.8945,
+      "step": 6319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636108554520936e-05,
+      "loss": 0.9375,
+      "step": 6320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634445843966723e-05,
+      "loss": 0.8867,
+      "step": 6321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632782970226518e-05,
+      "loss": 0.8164,
+      "step": 6322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463111993335887e-05,
+      "loss": 0.8711,
+      "step": 6323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629456733422343e-05,
+      "loss": 0.8594,
+      "step": 6324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627793370475492e-05,
+      "loss": 1.0312,
+      "step": 6325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626129844576894e-05,
+      "loss": 0.9609,
+      "step": 6326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624466155785118e-05,
+      "loss": 0.9688,
+      "step": 6327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622802304158752e-05,
+      "loss": 0.8359,
+      "step": 6328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621138289756378e-05,
+      "loss": 0.9219,
+      "step": 6329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619474112636591e-05,
+      "loss": 1.0781,
+      "step": 6330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617809772857985e-05,
+      "loss": 0.8281,
+      "step": 6331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616145270479169e-05,
+      "loss": 0.918,
+      "step": 6332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614480605558749e-05,
+      "loss": 0.9531,
+      "step": 6333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612815778155346e-05,
+      "loss": 0.8906,
+      "step": 6334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611150788327575e-05,
+      "loss": 0.9258,
+      "step": 6335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4609485636134066e-05,
+      "loss": 0.7969,
+      "step": 6336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607820321633447e-05,
+      "loss": 0.9805,
+      "step": 6337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606154844884365e-05,
+      "loss": 0.9414,
+      "step": 6338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604489205945455e-05,
+      "loss": 0.8477,
+      "step": 6339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602823404875375e-05,
+      "loss": 0.9062,
+      "step": 6340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601157441732773e-05,
+      "loss": 0.8359,
+      "step": 6341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4599491316576314e-05,
+      "loss": 0.8477,
+      "step": 6342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597825029464666e-05,
+      "loss": 0.8945,
+      "step": 6343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596158580456496e-05,
+      "loss": 0.8125,
+      "step": 6344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459449196961049e-05,
+      "loss": 0.9453,
+      "step": 6345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4592825196985325e-05,
+      "loss": 0.8203,
+      "step": 6346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591158262639696e-05,
+      "loss": 0.9141,
+      "step": 6347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589491166632292e-05,
+      "loss": 0.8398,
+      "step": 6348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587823909021821e-05,
+      "loss": 0.9688,
+      "step": 6349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586156489866984e-05,
+      "loss": 0.8906,
+      "step": 6350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45844889092265e-05,
+      "loss": 0.8203,
+      "step": 6351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458282116715908e-05,
+      "loss": 0.9531,
+      "step": 6352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581153263723454e-05,
+      "loss": 0.8477,
+      "step": 6353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579485198978343e-05,
+      "loss": 0.7773,
+      "step": 6354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577816972982493e-05,
+      "loss": 0.9453,
+      "step": 6355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576148585794634e-05,
+      "loss": 0.793,
+      "step": 6356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4574480037473521e-05,
+      "loss": 0.8477,
+      "step": 6357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572811328077903e-05,
+      "loss": 0.9219,
+      "step": 6358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571142457666536e-05,
+      "loss": 0.8008,
+      "step": 6359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569473426298187e-05,
+      "loss": 0.9727,
+      "step": 6360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567804234031621e-05,
+      "loss": 0.918,
+      "step": 6361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456613488092562e-05,
+      "loss": 0.8633,
+      "step": 6362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4564465367038953e-05,
+      "loss": 0.8477,
+      "step": 6363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456279569243042e-05,
+      "loss": 0.8711,
+      "step": 6364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45611258571588e-05,
+      "loss": 0.9023,
+      "step": 6365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559455861282898e-05,
+      "loss": 0.9883,
+      "step": 6366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557785704861515e-05,
+      "loss": 0.9453,
+      "step": 6367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556115387953464e-05,
+      "loss": 0.8906,
+      "step": 6368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554444910617552e-05,
+      "loss": 0.8359,
+      "step": 6369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552774272912602e-05,
+      "loss": 0.8828,
+      "step": 6370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551103474897442e-05,
+      "loss": 0.9219,
+      "step": 6371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549432516630901e-05,
+      "loss": 0.8555,
+      "step": 6372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547761398171816e-05,
+      "loss": 0.9883,
+      "step": 6373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546090119579033e-05,
+      "loss": 0.8906,
+      "step": 6374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4544418680911396e-05,
+      "loss": 0.9219,
+      "step": 6375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542747082227762e-05,
+      "loss": 0.9961,
+      "step": 6376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4541075323586988e-05,
+      "loss": 0.9297,
+      "step": 6377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453940340504794e-05,
+      "loss": 0.9336,
+      "step": 6378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453773132666949e-05,
+      "loss": 0.9258,
+      "step": 6379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4536059088510514e-05,
+      "loss": 0.9609,
+      "step": 6380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534386690629893e-05,
+      "loss": 0.9648,
+      "step": 6381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4532714133086517e-05,
+      "loss": 0.8789,
+      "step": 6382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531041415939275e-05,
+      "loss": 0.8125,
+      "step": 6383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452936853924707e-05,
+      "loss": 1.1641,
+      "step": 6384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4527695503068808e-05,
+      "loss": 0.8359,
+      "step": 6385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526022307463395e-05,
+      "loss": 0.8398,
+      "step": 6386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524348952489749e-05,
+      "loss": 0.8906,
+      "step": 6387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522675438206789e-05,
+      "loss": 0.957,
+      "step": 6388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4521001764673445e-05,
+      "loss": 0.875,
+      "step": 6389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519327931948648e-05,
+      "loss": 0.9297,
+      "step": 6390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517653940091339e-05,
+      "loss": 0.9062,
+      "step": 6391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515979789160455e-05,
+      "loss": 1.0,
+      "step": 6392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514305479214953e-05,
+      "loss": 0.9102,
+      "step": 6393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4512631010313785e-05,
+      "loss": 0.8125,
+      "step": 6394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451095638251591e-05,
+      "loss": 0.8945,
+      "step": 6395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509281595880295e-05,
+      "loss": 0.8477,
+      "step": 6396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507606650465917e-05,
+      "loss": 0.6484,
+      "step": 6397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505931546331746e-05,
+      "loss": 0.8867,
+      "step": 6398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4504256283536768e-05,
+      "loss": 0.9531,
+      "step": 6399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502580862139972e-05,
+      "loss": 0.8984,
+      "step": 6400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500905282200355e-05,
+      "loss": 0.8945,
+      "step": 6401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449922954377691e-05,
+      "loss": 0.8555,
+      "step": 6402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4497553646928648e-05,
+      "loss": 0.9062,
+      "step": 6403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495877591714576e-05,
+      "loss": 0.8516,
+      "step": 6404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494201378193715e-05,
+      "loss": 0.8867,
+      "step": 6405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449252500642508e-05,
+      "loss": 0.75,
+      "step": 6406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490848476467707e-05,
+      "loss": 0.9023,
+      "step": 6407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489171788380624e-05,
+      "loss": 0.9062,
+      "step": 6408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448749494222287e-05,
+      "loss": 0.8281,
+      "step": 6409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485817938053493e-05,
+      "loss": 0.9453,
+      "step": 6410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484140775931538e-05,
+      "loss": 1.0078,
+      "step": 6411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482463455916064e-05,
+      "loss": 1.0312,
+      "step": 6412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4480785978066131e-05,
+      "loss": 0.8086,
+      "step": 6413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479108342440804e-05,
+      "loss": 0.8633,
+      "step": 6414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477430549099159e-05,
+      "loss": 0.8828,
+      "step": 6415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447575259810027e-05,
+      "loss": 0.9297,
+      "step": 6416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447407448950322e-05,
+      "loss": 0.8203,
+      "step": 6417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472396223367102e-05,
+      "loss": 0.8242,
+      "step": 6418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470717799751004e-05,
+      "loss": 0.9453,
+      "step": 6419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469039218714034e-05,
+      "loss": 0.8984,
+      "step": 6420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446736048031529e-05,
+      "loss": 0.793,
+      "step": 6421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465681584613886e-05,
+      "loss": 0.8672,
+      "step": 6422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446400253166894e-05,
+      "loss": 0.8359,
+      "step": 6423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462323321539572e-05,
+      "loss": 0.9531,
+      "step": 6424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460643954284907e-05,
+      "loss": 0.8438,
+      "step": 6425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458964429964085e-05,
+      "loss": 0.9453,
+      "step": 6426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457284748636238e-05,
+      "loss": 0.832,
+      "step": 6427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455604910360514e-05,
+      "loss": 0.9062,
+      "step": 6428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445392491519606e-05,
+      "loss": 0.9805,
+      "step": 6429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452244763202036e-05,
+      "loss": 0.918,
+      "step": 6430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44505644544376e-05,
+      "loss": 0.8906,
+      "step": 6431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448883988961915e-05,
+      "loss": 0.9141,
+      "step": 6432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447203366834156e-05,
+      "loss": 0.8438,
+      "step": 6433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445522588113503e-05,
+      "loss": 0.9141,
+      "step": 6434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4443841652859134e-05,
+      "loss": 1.0391,
+      "step": 6435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442160561130238e-05,
+      "loss": 1.1016,
+      "step": 6436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444047931298601e-05,
+      "loss": 0.9609,
+      "step": 6437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438797908485652e-05,
+      "loss": 0.8242,
+      "step": 6438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437116347688363e-05,
+      "loss": 1.0234,
+      "step": 6439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443543463065336e-05,
+      "loss": 0.832,
+      "step": 6440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433752757439851e-05,
+      "loss": 0.8984,
+      "step": 6441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432070728107067e-05,
+      "loss": 0.9688,
+      "step": 6442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430388542714225e-05,
+      "loss": 0.9375,
+      "step": 6443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428706201320564e-05,
+      "loss": 0.8984,
+      "step": 6444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427023703985317e-05,
+      "loss": 0.8906,
+      "step": 6445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425341050767734e-05,
+      "loss": 0.9844,
+      "step": 6446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423658241727056e-05,
+      "loss": 0.9453,
+      "step": 6447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4421975276922547e-05,
+      "loss": 0.9688,
+      "step": 6448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420292156413454e-05,
+      "loss": 0.8828,
+      "step": 6449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418608880259054e-05,
+      "loss": 0.9688,
+      "step": 6450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441692544851861e-05,
+      "loss": 0.9258,
+      "step": 6451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415241861251406e-05,
+      "loss": 0.8438,
+      "step": 6452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4413558118516713e-05,
+      "loss": 0.9492,
+      "step": 6453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411874220373828e-05,
+      "loss": 0.8633,
+      "step": 6454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4410190166882037e-05,
+      "loss": 0.9023,
+      "step": 6455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4408505958100645e-05,
+      "loss": 0.8945,
+      "step": 6456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406821594088946e-05,
+      "loss": 0.8047,
+      "step": 6457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405137074906259e-05,
+      "loss": 0.9492,
+      "step": 6458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403452400611892e-05,
+      "loss": 0.875,
+      "step": 6459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4401767571265169e-05,
+      "loss": 1.0,
+      "step": 6460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400082586925413e-05,
+      "loss": 0.8281,
+      "step": 6461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4398397447651955e-05,
+      "loss": 0.875,
+      "step": 6462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396712153504129e-05,
+      "loss": 0.9102,
+      "step": 6463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395026704541283e-05,
+      "loss": 0.7969,
+      "step": 6464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393341100822759e-05,
+      "loss": 0.8906,
+      "step": 6465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4391655342407914e-05,
+      "loss": 0.9531,
+      "step": 6466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389969429356101e-05,
+      "loss": 0.9297,
+      "step": 6467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438828336172669e-05,
+      "loss": 0.9609,
+      "step": 6468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386597139579041e-05,
+      "loss": 0.8516,
+      "step": 6469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4384910762972538e-05,
+      "loss": 0.9414,
+      "step": 6470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4383224231966554e-05,
+      "loss": 0.9336,
+      "step": 6471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381537546620479e-05,
+      "loss": 0.875,
+      "step": 6472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43798507069937e-05,
+      "loss": 0.9453,
+      "step": 6473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378163713145618e-05,
+      "loss": 0.8594,
+      "step": 6474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437647656513563e-05,
+      "loss": 0.9375,
+      "step": 6475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374789263023146e-05,
+      "loss": 0.9805,
+      "step": 6476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373101806867577e-05,
+      "loss": 0.8086,
+      "step": 6477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437141419672834e-05,
+      "loss": 0.793,
+      "step": 6478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369726432664862e-05,
+      "loss": 0.9062,
+      "step": 6479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368038514736566e-05,
+      "loss": 0.8906,
+      "step": 6480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4366350443002892e-05,
+      "loss": 0.8281,
+      "step": 6481
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364662217523278e-05,
+      "loss": 1.0234,
+      "step": 6482
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362973838357169e-05,
+      "loss": 0.8711,
+      "step": 6483
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436128530556401e-05,
+      "loss": 1.0312,
+      "step": 6484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359596619203266e-05,
+      "loss": 1.0234,
+      "step": 6485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435790777933439e-05,
+      "loss": 0.8242,
+      "step": 6486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4356218786016859e-05,
+      "loss": 0.8281,
+      "step": 6487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435452963931013e-05,
+      "loss": 0.7812,
+      "step": 6488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352840339273697e-05,
+      "loss": 0.8555,
+      "step": 6489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351150885967029e-05,
+      "loss": 0.9531,
+      "step": 6490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349461279449624e-05,
+      "loss": 0.8594,
+      "step": 6491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434777151978097e-05,
+      "loss": 0.875,
+      "step": 6492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434608160702057e-05,
+      "loss": 0.9727,
+      "step": 6493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344391541227922e-05,
+      "loss": 0.8984,
+      "step": 6494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342701322462543e-05,
+      "loss": 0.8945,
+      "step": 6495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341010950783944e-05,
+      "loss": 0.8828,
+      "step": 6496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339320426251648e-05,
+      "loss": 0.9102,
+      "step": 6497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337629748925179e-05,
+      "loss": 1.0078,
+      "step": 6498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4335938918864072e-05,
+      "loss": 0.8633,
+      "step": 6499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334247936127858e-05,
+      "loss": 0.8477,
+      "step": 6500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332556800776085e-05,
+      "loss": 0.9609,
+      "step": 6501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330865512868293e-05,
+      "loss": 0.9297,
+      "step": 6502
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329174072464045e-05,
+      "loss": 0.9102,
+      "step": 6503
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327482479622888e-05,
+      "loss": 1.0078,
+      "step": 6504
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4325790734404396e-05,
+      "loss": 0.8555,
+      "step": 6505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432409883686813e-05,
+      "loss": 0.8945,
+      "step": 6506
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432240678707367e-05,
+      "loss": 0.9297,
+      "step": 6507
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432071458508059e-05,
+      "loss": 0.7891,
+      "step": 6508
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319022230948483e-05,
+      "loss": 0.793,
+      "step": 6509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431732972473693e-05,
+      "loss": 0.8594,
+      "step": 6510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315637066505534e-05,
+      "loss": 0.9805,
+      "step": 6511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4313944256313893e-05,
+      "loss": 0.9922,
+      "step": 6512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312251294221615e-05,
+      "loss": 0.8594,
+      "step": 6513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431055818028831e-05,
+      "loss": 0.875,
+      "step": 6514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308864914573595e-05,
+      "loss": 0.8867,
+      "step": 6515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307171497137096e-05,
+      "loss": 0.8711,
+      "step": 6516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305477928038436e-05,
+      "loss": 0.8203,
+      "step": 6517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430378420733725e-05,
+      "loss": 0.8789,
+      "step": 6518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302090335093179e-05,
+      "loss": 0.9102,
+      "step": 6519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300396311365864e-05,
+      "loss": 0.8945,
+      "step": 6520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298702136214958e-05,
+      "loss": 0.957,
+      "step": 6521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429700780970011e-05,
+      "loss": 0.8672,
+      "step": 6522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429531333188098e-05,
+      "loss": 0.8711,
+      "step": 6523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429361870281724e-05,
+      "loss": 0.8789,
+      "step": 6524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291923922568551e-05,
+      "loss": 1.0391,
+      "step": 6525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42902289911946e-05,
+      "loss": 1.0078,
+      "step": 6526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288533908755058e-05,
+      "loss": 0.9844,
+      "step": 6527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286838675309618e-05,
+      "loss": 0.8516,
+      "step": 6528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285143290917966e-05,
+      "loss": 0.9023,
+      "step": 6529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4283447755639808e-05,
+      "loss": 0.8984,
+      "step": 6530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281752069534835e-05,
+      "loss": 0.8633,
+      "step": 6531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280056232662765e-05,
+      "loss": 0.9883,
+      "step": 6532
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4278360245083302e-05,
+      "loss": 0.8047,
+      "step": 6533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276664106856176e-05,
+      "loss": 0.8047,
+      "step": 6534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42749678180411e-05,
+      "loss": 0.8477,
+      "step": 6535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273271378697804e-05,
+      "loss": 0.9922,
+      "step": 6536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271574788886029e-05,
+      "loss": 0.8086,
+      "step": 6537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269878048665506e-05,
+      "loss": 0.9727,
+      "step": 6538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268181158095989e-05,
+      "loss": 0.8555,
+      "step": 6539
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266484117237218e-05,
+      "loss": 0.9922,
+      "step": 6540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264786926148956e-05,
+      "loss": 0.9297,
+      "step": 6541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263089584890964e-05,
+      "loss": 0.9531,
+      "step": 6542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4261392093523005e-05,
+      "loss": 0.832,
+      "step": 6543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259694452104849e-05,
+      "loss": 0.8906,
+      "step": 6544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257996660696275e-05,
+      "loss": 0.9297,
+      "step": 6545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4256298719357063e-05,
+      "loss": 0.8594,
+      "step": 6546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254600628147003e-05,
+      "loss": 0.9141,
+      "step": 6547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4252902387125886e-05,
+      "loss": 0.8359,
+      "step": 6548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251203996353509e-05,
+      "loss": 0.7656,
+      "step": 6549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249505455889676e-05,
+      "loss": 0.8477,
+      "step": 6550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247806765794194e-05,
+      "loss": 0.8008,
+      "step": 6551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4246107926126878e-05,
+      "loss": 0.7852,
+      "step": 6552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244408936947545e-05,
+      "loss": 0.8477,
+      "step": 6553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424270979831602e-05,
+      "loss": 0.8906,
+      "step": 6554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424101051029213e-05,
+      "loss": 0.8203,
+      "step": 6555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239311072935716e-05,
+      "loss": 1.0156,
+      "step": 6556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423761148630661e-05,
+      "loss": 0.832,
+      "step": 6557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235911750464664e-05,
+      "loss": 0.8125,
+      "step": 6558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234211865469722e-05,
+      "loss": 0.9375,
+      "step": 6559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232511831381648e-05,
+      "loss": 0.8477,
+      "step": 6560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4230811648260292e-05,
+      "loss": 0.8047,
+      "step": 6561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229111316165526e-05,
+      "loss": 0.8867,
+      "step": 6562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227410835157223e-05,
+      "loss": 0.793,
+      "step": 6563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225710205295258e-05,
+      "loss": 0.9023,
+      "step": 6564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224009426639509e-05,
+      "loss": 0.9531,
+      "step": 6565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222308499249869e-05,
+      "loss": 0.9805,
+      "step": 6566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220607423186224e-05,
+      "loss": 0.8516,
+      "step": 6567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421890619850848e-05,
+      "loss": 0.8203,
+      "step": 6568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421720482527653e-05,
+      "loss": 0.8398,
+      "step": 6569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421550330355029e-05,
+      "loss": 0.9141,
+      "step": 6570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213801633389668e-05,
+      "loss": 0.8711,
+      "step": 6571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4212099814854585e-05,
+      "loss": 0.8242,
+      "step": 6572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210397848004964e-05,
+      "loss": 0.9492,
+      "step": 6573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208695732900734e-05,
+      "loss": 0.9102,
+      "step": 6574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420699346960183e-05,
+      "loss": 0.8711,
+      "step": 6575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4205291058168192e-05,
+      "loss": 0.8164,
+      "step": 6576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203588498659758e-05,
+      "loss": 0.7383,
+      "step": 6577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201885791136487e-05,
+      "loss": 0.8711,
+      "step": 6578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4200182935658327e-05,
+      "loss": 0.8672,
+      "step": 6579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419847993228524e-05,
+      "loss": 0.8242,
+      "step": 6580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196776781077196e-05,
+      "loss": 0.7812,
+      "step": 6581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195073482094159e-05,
+      "loss": 0.8711,
+      "step": 6582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193370035396106e-05,
+      "loss": 0.8516,
+      "step": 6583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191666441043022e-05,
+      "loss": 0.8594,
+      "step": 6584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418996269909489e-05,
+      "loss": 1.0938,
+      "step": 6585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188258809611704e-05,
+      "loss": 0.832,
+      "step": 6586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186554772653458e-05,
+      "loss": 0.8945,
+      "step": 6587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418485058828015e-05,
+      "loss": 0.9453,
+      "step": 6588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183146256551796e-05,
+      "loss": 0.9375,
+      "step": 6589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181441777528402e-05,
+      "loss": 0.8398,
+      "step": 6590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179737151269988e-05,
+      "loss": 0.9141,
+      "step": 6591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178032377836574e-05,
+      "loss": 0.918,
+      "step": 6592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176327457288191e-05,
+      "loss": 0.8828,
+      "step": 6593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4174622389684866e-05,
+      "loss": 0.8828,
+      "step": 6594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172917175086646e-05,
+      "loss": 0.8242,
+      "step": 6595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171211813553564e-05,
+      "loss": 0.918,
+      "step": 6596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416950630514568e-05,
+      "loss": 0.957,
+      "step": 6597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167800649923036e-05,
+      "loss": 0.832,
+      "step": 6598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41660948479457e-05,
+      "loss": 0.8867,
+      "step": 6599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416438889927373e-05,
+      "loss": 0.9648,
+      "step": 6600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4162682803967197e-05,
+      "loss": 0.8594,
+      "step": 6601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160976562086173e-05,
+      "loss": 0.9609,
+      "step": 6602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159270173690743e-05,
+      "loss": 0.8164,
+      "step": 6603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4157563638840987e-05,
+      "loss": 0.8359,
+      "step": 6604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4155856957596999e-05,
+      "loss": 0.9258,
+      "step": 6605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.9219,
+      "step": 6606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152443156166695e-05,
+      "loss": 0.8438,
+      "step": 6607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415073603610059e-05,
+      "loss": 0.9141,
+      "step": 6608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414902876988066e-05,
+      "loss": 0.8906,
+      "step": 6609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147321357567019e-05,
+      "loss": 0.7852,
+      "step": 6610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4145613799219794e-05,
+      "loss": 0.9141,
+      "step": 6611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143906094899103e-05,
+      "loss": 0.7852,
+      "step": 6612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414219824466508e-05,
+      "loss": 0.9844,
+      "step": 6613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414049024857786e-05,
+      "loss": 0.8398,
+      "step": 6614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4138782106697591e-05,
+      "loss": 0.6836,
+      "step": 6615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137073819084407e-05,
+      "loss": 0.8359,
+      "step": 6616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135365385798472e-05,
+      "loss": 1.0469,
+      "step": 6617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133656806899933e-05,
+      "loss": 0.875,
+      "step": 6618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131948082448958e-05,
+      "loss": 1.0859,
+      "step": 6619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413023921250571e-05,
+      "loss": 1.1094,
+      "step": 6620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4128530197130361e-05,
+      "loss": 0.8555,
+      "step": 6621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412682103638309e-05,
+      "loss": 0.9297,
+      "step": 6622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4125111730324082e-05,
+      "loss": 0.9609,
+      "step": 6623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123402279013517e-05,
+      "loss": 0.832,
+      "step": 6624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121692682511592e-05,
+      "loss": 0.9258,
+      "step": 6625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4119982940878503e-05,
+      "loss": 0.8516,
+      "step": 6626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118273054174455e-05,
+      "loss": 0.8203,
+      "step": 6627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116563022459653e-05,
+      "loss": 1.0547,
+      "step": 6628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4114852845794312e-05,
+      "loss": 0.8359,
+      "step": 6629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4113142524238647e-05,
+      "loss": 0.9219,
+      "step": 6630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111432057852885e-05,
+      "loss": 0.8906,
+      "step": 6631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4109721446697254e-05,
+      "loss": 0.8711,
+      "step": 6632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4108010690831981e-05,
+      "loss": 0.8711,
+      "step": 6633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410629979031731e-05,
+      "loss": 0.832,
+      "step": 6634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104588745213488e-05,
+      "loss": 1.0156,
+      "step": 6635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4102877555580752e-05,
+      "loss": 0.8477,
+      "step": 6636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101166221479368e-05,
+      "loss": 0.957,
+      "step": 6637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4099454742969585e-05,
+      "loss": 0.918,
+      "step": 6638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097743120111674e-05,
+      "loss": 0.9062,
+      "step": 6639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096031352965898e-05,
+      "loss": 0.9062,
+      "step": 6640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094319441592536e-05,
+      "loss": 0.8242,
+      "step": 6641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092607386051866e-05,
+      "loss": 0.9961,
+      "step": 6642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090895186404168e-05,
+      "loss": 0.9141,
+      "step": 6643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089182842709736e-05,
+      "loss": 0.9219,
+      "step": 6644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.408747035502886e-05,
+      "loss": 0.8477,
+      "step": 6645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085757723421847e-05,
+      "loss": 0.8906,
+      "step": 6646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084044947948991e-05,
+      "loss": 1.0312,
+      "step": 6647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4082332028670608e-05,
+      "loss": 0.8945,
+      "step": 6648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080618965647012e-05,
+      "loss": 0.8555,
+      "step": 6649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078905758938521e-05,
+      "loss": 0.8164,
+      "step": 6650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077192408605461e-05,
+      "loss": 0.8828,
+      "step": 6651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407547891470816e-05,
+      "loss": 0.8359,
+      "step": 6652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073765277306954e-05,
+      "loss": 0.8125,
+      "step": 6653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072051496462185e-05,
+      "loss": 0.9492,
+      "step": 6654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070337572234195e-05,
+      "loss": 0.9297,
+      "step": 6655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068623504683334e-05,
+      "loss": 0.9336,
+      "step": 6656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066909293869955e-05,
+      "loss": 0.9883,
+      "step": 6657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065194939854423e-05,
+      "loss": 1.0469,
+      "step": 6658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40634804426971e-05,
+      "loss": 0.8789,
+      "step": 6659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061765802458358e-05,
+      "loss": 0.9297,
+      "step": 6660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060051019198566e-05,
+      "loss": 0.8398,
+      "step": 6661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058336092978116e-05,
+      "loss": 0.9727,
+      "step": 6662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405662102385738e-05,
+      "loss": 0.8203,
+      "step": 6663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054905811896757e-05,
+      "loss": 0.918,
+      "step": 6664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053190457156638e-05,
+      "loss": 0.9141,
+      "step": 6665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405147495969743e-05,
+      "loss": 0.9336,
+      "step": 6666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049759319579526e-05,
+      "loss": 0.9805,
+      "step": 6667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048043536863349e-05,
+      "loss": 0.8945,
+      "step": 6668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4046327611609305e-05,
+      "loss": 1.0391,
+      "step": 6669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044611543877822e-05,
+      "loss": 0.9141,
+      "step": 6670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042895333729318e-05,
+      "loss": 0.9766,
+      "step": 6671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404117898122423e-05,
+      "loss": 0.832,
+      "step": 6672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039462486422988e-05,
+      "loss": 0.8438,
+      "step": 6673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037745849386038e-05,
+      "loss": 0.8398,
+      "step": 6674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403602907017382e-05,
+      "loss": 0.8906,
+      "step": 6675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034312148846787e-05,
+      "loss": 0.957,
+      "step": 6676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032595085465395e-05,
+      "loss": 0.8359,
+      "step": 6677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030877880090104e-05,
+      "loss": 0.8516,
+      "step": 6678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402916053278138e-05,
+      "loss": 0.918,
+      "step": 6679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027443043599691e-05,
+      "loss": 0.9102,
+      "step": 6680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4025725412605514e-05,
+      "loss": 0.8984,
+      "step": 6681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402400763985933e-05,
+      "loss": 0.957,
+      "step": 6682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022289725421623e-05,
+      "loss": 0.8828,
+      "step": 6683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020571669352885e-05,
+      "loss": 0.8633,
+      "step": 6684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018853471713611e-05,
+      "loss": 1.0391,
+      "step": 6685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4017135132564301e-05,
+      "loss": 0.9375,
+      "step": 6686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4015416651965461e-05,
+      "loss": 0.8711,
+      "step": 6687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4013698029977603e-05,
+      "loss": 0.8203,
+      "step": 6688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011979266661235e-05,
+      "loss": 0.7969,
+      "step": 6689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010260362076886e-05,
+      "loss": 0.8906,
+      "step": 6690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008541316285076e-05,
+      "loss": 0.8906,
+      "step": 6691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400682212934634e-05,
+      "loss": 0.8203,
+      "step": 6692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005102801321208e-05,
+      "loss": 0.8789,
+      "step": 6693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003383332270224e-05,
+      "loss": 0.7852,
+      "step": 6694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400166372225393e-05,
+      "loss": 1.0625,
+      "step": 6695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399994397133288e-05,
+      "loss": 0.7539,
+      "step": 6696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998224079567625e-05,
+      "loss": 1.0469,
+      "step": 6697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996504047018726e-05,
+      "loss": 0.9883,
+      "step": 6698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3994783873746754e-05,
+      "loss": 0.8477,
+      "step": 6699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993063559812268e-05,
+      "loss": 0.8125,
+      "step": 6700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991343105275852e-05,
+      "loss": 1.0,
+      "step": 6701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3989622510198079e-05,
+      "loss": 0.7656,
+      "step": 6702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987901774639543e-05,
+      "loss": 0.7773,
+      "step": 6703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3986180898660824e-05,
+      "loss": 0.9883,
+      "step": 6704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3984459882322523e-05,
+      "loss": 1.0078,
+      "step": 6705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982738725685235e-05,
+      "loss": 0.9258,
+      "step": 6706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398101742880957e-05,
+      "loss": 0.8086,
+      "step": 6707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979295991756134e-05,
+      "loss": 0.9414,
+      "step": 6708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977574414585542e-05,
+      "loss": 0.8477,
+      "step": 6709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3975852697358414e-05,
+      "loss": 0.8203,
+      "step": 6710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974130840135371e-05,
+      "loss": 1.0,
+      "step": 6711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972408842977047e-05,
+      "loss": 0.9336,
+      "step": 6712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970686705944074e-05,
+      "loss": 0.7852,
+      "step": 6713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3968964429097094e-05,
+      "loss": 0.7891,
+      "step": 6714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967242012496745e-05,
+      "loss": 0.8867,
+      "step": 6715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396551945620368e-05,
+      "loss": 0.9258,
+      "step": 6716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3963796760278551e-05,
+      "loss": 1.0938,
+      "step": 6717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962073924782023e-05,
+      "loss": 0.8867,
+      "step": 6718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960350949774748e-05,
+      "loss": 0.8047,
+      "step": 6719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3958627835317407e-05,
+      "loss": 0.8945,
+      "step": 6720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956904581470667e-05,
+      "loss": 0.9102,
+      "step": 6721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955181188295203e-05,
+      "loss": 0.8242,
+      "step": 6722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953457655851707e-05,
+      "loss": 0.9141,
+      "step": 6723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951733984200859e-05,
+      "loss": 0.8594,
+      "step": 6724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395001017340336e-05,
+      "loss": 0.8945,
+      "step": 6725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39482862235199e-05,
+      "loss": 0.9023,
+      "step": 6726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394656213461119e-05,
+      "loss": 0.8828,
+      "step": 6727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394483790673793e-05,
+      "loss": 0.9648,
+      "step": 6728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394311353996084e-05,
+      "loss": 0.8672,
+      "step": 6729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3941389034340632e-05,
+      "loss": 0.8633,
+      "step": 6730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393966438993803e-05,
+      "loss": 0.8242,
+      "step": 6731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937939606813762e-05,
+      "loss": 0.9219,
+      "step": 6732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936214685028562e-05,
+      "loss": 0.9453,
+      "step": 6733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3934489624643165e-05,
+      "loss": 0.8984,
+      "step": 6734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932764425718312e-05,
+      "loss": 0.8828,
+      "step": 6735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931039088314751e-05,
+      "loss": 0.8984,
+      "step": 6736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929313612493239e-05,
+      "loss": 1.0547,
+      "step": 6737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927587998314522e-05,
+      "loss": 0.75,
+      "step": 6738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392586224583937e-05,
+      "loss": 0.7031,
+      "step": 6739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924136355128546e-05,
+      "loss": 0.9375,
+      "step": 6740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922410326242823e-05,
+      "loss": 1.0469,
+      "step": 6741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3920684159242977e-05,
+      "loss": 0.8984,
+      "step": 6742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3918957854189787e-05,
+      "loss": 0.8477,
+      "step": 6743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391723141114404e-05,
+      "loss": 1.0312,
+      "step": 6744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391550483016653e-05,
+      "loss": 0.8672,
+      "step": 6745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3913778111318044e-05,
+      "loss": 0.9648,
+      "step": 6746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3912051254659391e-05,
+      "loss": 0.8281,
+      "step": 6747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910324260251371e-05,
+      "loss": 0.9141,
+      "step": 6748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908597128154799e-05,
+      "loss": 1.0625,
+      "step": 6749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3906869858430485e-05,
+      "loss": 0.7812,
+      "step": 6750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905142451139253e-05,
+      "loss": 0.9297,
+      "step": 6751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903414906341923e-05,
+      "loss": 0.9062,
+      "step": 6752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901687224099331e-05,
+      "loss": 0.8398,
+      "step": 6753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899959404472305e-05,
+      "loss": 0.9492,
+      "step": 6754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898231447521689e-05,
+      "loss": 0.8281,
+      "step": 6755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3896503353308326e-05,
+      "loss": 0.8008,
+      "step": 6756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389477512189306e-05,
+      "loss": 0.8242,
+      "step": 6757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389304675333675e-05,
+      "loss": 0.9727,
+      "step": 6758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891318247700252e-05,
+      "loss": 1.0312,
+      "step": 6759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3889589605044431e-05,
+      "loss": 0.8555,
+      "step": 6760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887860825430155e-05,
+      "loss": 0.9453,
+      "step": 6761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3886131908918296e-05,
+      "loss": 0.918,
+      "step": 6762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884402855569732e-05,
+      "loss": 0.9141,
+      "step": 6763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882673665445348e-05,
+      "loss": 0.9922,
+      "step": 6764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880944338606027e-05,
+      "loss": 0.8125,
+      "step": 6765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3879214875112665e-05,
+      "loss": 0.9688,
+      "step": 6766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877485275026154e-05,
+      "loss": 0.9727,
+      "step": 6767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875755538407405e-05,
+      "loss": 0.8008,
+      "step": 6768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874025665317317e-05,
+      "loss": 0.8398,
+      "step": 6769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872295655816803e-05,
+      "loss": 1.0547,
+      "step": 6770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387056550996678e-05,
+      "loss": 0.9219,
+      "step": 6771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868835227828172e-05,
+      "loss": 0.8164,
+      "step": 6772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38671048094619e-05,
+      "loss": 0.8359,
+      "step": 6773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865374254928899e-05,
+      "loss": 0.8203,
+      "step": 6774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863643564290098e-05,
+      "loss": 0.8398,
+      "step": 6775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861912737606446e-05,
+      "loss": 0.7812,
+      "step": 6776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860181774938881e-05,
+      "loss": 0.8672,
+      "step": 6777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858450676348359e-05,
+      "loss": 0.8984,
+      "step": 6778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856719441895827e-05,
+      "loss": 0.8438,
+      "step": 6779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854988071642253e-05,
+      "loss": 0.8828,
+      "step": 6780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853256565648593e-05,
+      "loss": 0.9766,
+      "step": 6781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3851524923975823e-05,
+      "loss": 0.8633,
+      "step": 6782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849793146684912e-05,
+      "loss": 0.8984,
+      "step": 6783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384806123383684e-05,
+      "loss": 1.0391,
+      "step": 6784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846329185492591e-05,
+      "loss": 0.832,
+      "step": 6785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844597001713154e-05,
+      "loss": 0.9297,
+      "step": 6786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3842864682559517e-05,
+      "loss": 0.9062,
+      "step": 6787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841132228092684e-05,
+      "loss": 0.9648,
+      "step": 6788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839399638373654e-05,
+      "loss": 0.8672,
+      "step": 6789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3837666913463434e-05,
+      "loss": 0.9844,
+      "step": 6790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835934053423038e-05,
+      "loss": 0.7891,
+      "step": 6791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834201058313482e-05,
+      "loss": 0.7852,
+      "step": 6792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832467928195783e-05,
+      "loss": 0.8008,
+      "step": 6793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830734663130975e-05,
+      "loss": 0.8242,
+      "step": 6794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829001263180083e-05,
+      "loss": 0.8281,
+      "step": 6795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3827267728404144e-05,
+      "loss": 0.8359,
+      "step": 6796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3825534058864199e-05,
+      "loss": 0.9492,
+      "step": 6797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823800254621295e-05,
+      "loss": 0.9531,
+      "step": 6798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822066315736477e-05,
+      "loss": 0.9297,
+      "step": 6799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820332242270806e-05,
+      "loss": 0.8438,
+      "step": 6800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3818598034285335e-05,
+      "loss": 0.9219,
+      "step": 6801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816863691841136e-05,
+      "loss": 1.0078,
+      "step": 6802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381512921499927e-05,
+      "loss": 0.8516,
+      "step": 6803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813394603820815e-05,
+      "loss": 0.9922,
+      "step": 6804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811659858366846e-05,
+      "loss": 0.9023,
+      "step": 6805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809924978698451e-05,
+      "loss": 0.8984,
+      "step": 6806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808189964876712e-05,
+      "loss": 0.8125,
+      "step": 6807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806454816962727e-05,
+      "loss": 0.8203,
+      "step": 6808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380471953501759e-05,
+      "loss": 0.9297,
+      "step": 6809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802984119102403e-05,
+      "loss": 0.9102,
+      "step": 6810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801248569278276e-05,
+      "loss": 0.8789,
+      "step": 6811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799512885606314e-05,
+      "loss": 0.9023,
+      "step": 6812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379777706814764e-05,
+      "loss": 0.8984,
+      "step": 6813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3796041116963368e-05,
+      "loss": 0.8203,
+      "step": 6814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379430503211463e-05,
+      "loss": 1.0078,
+      "step": 6815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3792568813662553e-05,
+      "loss": 0.875,
+      "step": 6816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790832461668273e-05,
+      "loss": 0.9531,
+      "step": 6817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3789095976192928e-05,
+      "loss": 0.875,
+      "step": 6818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787359357297663e-05,
+      "loss": 0.7539,
+      "step": 6819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785622605043628e-05,
+      "loss": 0.8984,
+      "step": 6820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783885719491977e-05,
+      "loss": 0.9766,
+      "step": 6821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3782148700703863e-05,
+      "loss": 0.8594,
+      "step": 6822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378041154874046e-05,
+      "loss": 1.0156,
+      "step": 6823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778674263662926e-05,
+      "loss": 0.9219,
+      "step": 6824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776936845532439e-05,
+      "loss": 0.957,
+      "step": 6825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377519929441017e-05,
+      "loss": 0.9961,
+      "step": 6826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773461610357308e-05,
+      "loss": 0.8555,
+      "step": 6827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771723793435037e-05,
+      "loss": 0.8555,
+      "step": 6828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769985843704547e-05,
+      "loss": 0.7773,
+      "step": 6829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376824776122703e-05,
+      "loss": 0.8203,
+      "step": 6830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3766509546063699e-05,
+      "loss": 0.9609,
+      "step": 6831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3764771198275745e-05,
+      "loss": 0.8828,
+      "step": 6832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376303271792439e-05,
+      "loss": 0.8633,
+      "step": 6833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761294105070839e-05,
+      "loss": 0.8789,
+      "step": 6834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759555359776317e-05,
+      "loss": 0.8633,
+      "step": 6835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3757816482102044e-05,
+      "loss": 0.8398,
+      "step": 6836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756077472109252e-05,
+      "loss": 0.875,
+      "step": 6837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754338329859174e-05,
+      "loss": 1.1172,
+      "step": 6838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752599055413045e-05,
+      "loss": 0.8828,
+      "step": 6839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3750859648832108e-05,
+      "loss": 0.9102,
+      "step": 6840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749120110177616e-05,
+      "loss": 0.8125,
+      "step": 6841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374738043951081e-05,
+      "loss": 0.8203,
+      "step": 6842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745640636892959e-05,
+      "loss": 0.9961,
+      "step": 6843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3743900702385313e-05,
+      "loss": 1.0234,
+      "step": 6844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742160636049147e-05,
+      "loss": 1.0078,
+      "step": 6845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740420437945723e-05,
+      "loss": 0.9102,
+      "step": 6846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738680108136324e-05,
+      "loss": 0.8281,
+      "step": 6847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736939646682222e-05,
+      "loss": 0.9219,
+      "step": 6848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373519905364471e-05,
+      "loss": 0.9414,
+      "step": 6849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733458329085068e-05,
+      "loss": 0.9375,
+      "step": 6850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3731717473064597e-05,
+      "loss": 0.8828,
+      "step": 6851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729976485644587e-05,
+      "loss": 0.7812,
+      "step": 6852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372823536688635e-05,
+      "loss": 1.0469,
+      "step": 6853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726494116851189e-05,
+      "loss": 0.8555,
+      "step": 6854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724752735600415e-05,
+      "loss": 0.8711,
+      "step": 6855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723011223195349e-05,
+      "loss": 0.9141,
+      "step": 6856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721269579697306e-05,
+      "loss": 0.8555,
+      "step": 6857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719527805167616e-05,
+      "loss": 0.8633,
+      "step": 6858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717785899667614e-05,
+      "loss": 0.7812,
+      "step": 6859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3716043863258624e-05,
+      "loss": 0.8125,
+      "step": 6860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714301696001995e-05,
+      "loss": 0.9375,
+      "step": 6861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712559397959066e-05,
+      "loss": 0.8008,
+      "step": 6862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3710816969191192e-05,
+      "loss": 0.7695,
+      "step": 6863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370907440975972e-05,
+      "loss": 0.8047,
+      "step": 6864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3707331719726014e-05,
+      "loss": 0.8438,
+      "step": 6865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3705588899151432e-05,
+      "loss": 0.8633,
+      "step": 6866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703845948097343e-05,
+      "loss": 0.7812,
+      "step": 6867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702102866625122e-05,
+      "loss": 0.8594,
+      "step": 6868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700359654796141e-05,
+      "loss": 0.9258,
+      "step": 6869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3698616312671784e-05,
+      "loss": 0.8633,
+      "step": 6870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696872840313433e-05,
+      "loss": 0.8203,
+      "step": 6871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3695129237782485e-05,
+      "loss": 0.8984,
+      "step": 6872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693385505140332e-05,
+      "loss": 0.8438,
+      "step": 6873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369164164244837e-05,
+      "loss": 0.8359,
+      "step": 6874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689897649768008e-05,
+      "loss": 0.8164,
+      "step": 6875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368815352716065e-05,
+      "loss": 0.7578,
+      "step": 6876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3686409274687712e-05,
+      "loss": 0.9414,
+      "step": 6877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684664892410616e-05,
+      "loss": 0.8672,
+      "step": 6878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682920380390776e-05,
+      "loss": 0.875,
+      "step": 6879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3681175738689626e-05,
+      "loss": 0.8984,
+      "step": 6880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679430967368592e-05,
+      "loss": 0.9883,
+      "step": 6881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677686066489111e-05,
+      "loss": 1.0,
+      "step": 6882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675941036112629e-05,
+      "loss": 0.8516,
+      "step": 6883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674195876300585e-05,
+      "loss": 0.9805,
+      "step": 6884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672450587114435e-05,
+      "loss": 0.8477,
+      "step": 6885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670705168615623e-05,
+      "loss": 1.0078,
+      "step": 6886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366895962086562e-05,
+      "loss": 0.8906,
+      "step": 6887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366721394392588e-05,
+      "loss": 0.9531,
+      "step": 6888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665468137857877e-05,
+      "loss": 0.9219,
+      "step": 6889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663722202723082e-05,
+      "loss": 0.7891,
+      "step": 6890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661976138582972e-05,
+      "loss": 0.8633,
+      "step": 6891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3660229945499026e-05,
+      "loss": 0.8867,
+      "step": 6892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658483623532735e-05,
+      "loss": 0.8242,
+      "step": 6893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3656737172745584e-05,
+      "loss": 1.0234,
+      "step": 6894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654990593199072e-05,
+      "loss": 0.9219,
+      "step": 6895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36532438849547e-05,
+      "loss": 0.8867,
+      "step": 6896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651497048073968e-05,
+      "loss": 0.8281,
+      "step": 6897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649750082618387e-05,
+      "loss": 0.9805,
+      "step": 6898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648002988649469e-05,
+      "loss": 0.8711,
+      "step": 6899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3646255766228735e-05,
+      "loss": 0.793,
+      "step": 6900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644508415417705e-05,
+      "loss": 0.9062,
+      "step": 6901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642760936277907e-05,
+      "loss": 0.9844,
+      "step": 6902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641013328870873e-05,
+      "loss": 0.9609,
+      "step": 6903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363926559325814e-05,
+      "loss": 0.8672,
+      "step": 6904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637517729501242e-05,
+      "loss": 0.9688,
+      "step": 6905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363576973766173e-05,
+      "loss": 0.9688,
+      "step": 6906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634021617801152e-05,
+      "loss": 0.9766,
+      "step": 6907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632273369981062e-05,
+      "loss": 0.75,
+      "step": 6908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363052499426302e-05,
+      "loss": 0.7617,
+      "step": 6909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628776490708584e-05,
+      "loss": 0.957,
+      "step": 6910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627027859379327e-05,
+      "loss": 1.0156,
+      "step": 6911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3625279100336818e-05,
+      "loss": 0.9336,
+      "step": 6912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623530213642636e-05,
+      "loss": 0.918,
+      "step": 6913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362178119935836e-05,
+      "loss": 0.8008,
+      "step": 6914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620032057545576e-05,
+      "loss": 0.8477,
+      "step": 6915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618282788265873e-05,
+      "loss": 0.8242,
+      "step": 6916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616533391580846e-05,
+      "loss": 0.8633,
+      "step": 6917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614783867552097e-05,
+      "loss": 1.0312,
+      "step": 6918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613034216241223e-05,
+      "loss": 0.9062,
+      "step": 6919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361128443770984e-05,
+      "loss": 0.875,
+      "step": 6920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609534532019553e-05,
+      "loss": 0.8008,
+      "step": 6921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607784499231985e-05,
+      "loss": 0.7734,
+      "step": 6922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606034339408751e-05,
+      "loss": 0.9219,
+      "step": 6923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360428405261148e-05,
+      "loss": 0.9141,
+      "step": 6924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602533638901806e-05,
+      "loss": 0.8672,
+      "step": 6925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3600783098341359e-05,
+      "loss": 0.8594,
+      "step": 6926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599032430991779e-05,
+      "loss": 0.9883,
+      "step": 6927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359728163691471e-05,
+      "loss": 0.9375,
+      "step": 6928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595530716171801e-05,
+      "loss": 0.9023,
+      "step": 6929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593779668824702e-05,
+      "loss": 0.9766,
+      "step": 6930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3592028494935074e-05,
+      "loss": 0.9531,
+      "step": 6931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590277194564576e-05,
+      "loss": 0.8477,
+      "step": 6932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588525767774875e-05,
+      "loss": 0.8477,
+      "step": 6933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3586774214627641e-05,
+      "loss": 0.875,
+      "step": 6934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585022535184549e-05,
+      "loss": 0.9531,
+      "step": 6935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3583270729507279e-05,
+      "loss": 0.8984,
+      "step": 6936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581518797657514e-05,
+      "loss": 0.832,
+      "step": 6937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579766739696938e-05,
+      "loss": 0.9219,
+      "step": 6938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578014555687254e-05,
+      "loss": 0.9688,
+      "step": 6939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3576262245690148e-05,
+      "loss": 0.8203,
+      "step": 6940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574509809767331e-05,
+      "loss": 0.8477,
+      "step": 6941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572757247980502e-05,
+      "loss": 0.8945,
+      "step": 6942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571004560391374e-05,
+      "loss": 1.0156,
+      "step": 6943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569251747061662e-05,
+      "loss": 0.8359,
+      "step": 6944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567498808053085e-05,
+      "loss": 1.0859,
+      "step": 6945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3565745743427368e-05,
+      "loss": 0.8164,
+      "step": 6946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356399255324624e-05,
+      "loss": 1.0312,
+      "step": 6947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3562239237571426e-05,
+      "loss": 1.0391,
+      "step": 6948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560485796464671e-05,
+      "loss": 1.0703,
+      "step": 6949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558732229987714e-05,
+      "loss": 0.9688,
+      "step": 6950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35569785382023e-05,
+      "loss": 1.0,
+      "step": 6951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355522472117018e-05,
+      "loss": 1.0,
+      "step": 6952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3553470778953109e-05,
+      "loss": 0.8906,
+      "step": 6953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551716711612843e-05,
+      "loss": 0.9258,
+      "step": 6954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549962519211151e-05,
+      "loss": 1.1406,
+      "step": 6955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548208201809795e-05,
+      "loss": 0.9297,
+      "step": 6956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354645375947055e-05,
+      "loss": 0.793,
+      "step": 6957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3544699192255193e-05,
+      "loss": 0.875,
+      "step": 6958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542944500225503e-05,
+      "loss": 0.9336,
+      "step": 6959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541189683443268e-05,
+      "loss": 0.875,
+      "step": 6960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539434741970278e-05,
+      "loss": 0.9609,
+      "step": 6961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537679675868324e-05,
+      "loss": 0.9688,
+      "step": 6962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535924485199208e-05,
+      "loss": 0.8945,
+      "step": 6963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353416917002473e-05,
+      "loss": 1.0156,
+      "step": 6964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532413730406702e-05,
+      "loss": 0.8477,
+      "step": 6965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353065816640693e-05,
+      "loss": 0.8828,
+      "step": 6966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3528902478087231e-05,
+      "loss": 0.9062,
+      "step": 6967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3527146665509428e-05,
+      "loss": 1.0156,
+      "step": 6968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525390728735345e-05,
+      "loss": 0.8477,
+      "step": 6969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3523634667826813e-05,
+      "loss": 0.8281,
+      "step": 6970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521878482845665e-05,
+      "loss": 0.8828,
+      "step": 6971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520122173853734e-05,
+      "loss": 0.7969,
+      "step": 6972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3518365740912868e-05,
+      "loss": 0.9648,
+      "step": 6973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3516609184084913e-05,
+      "loss": 0.9102,
+      "step": 6974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514852503431717e-05,
+      "loss": 0.8711,
+      "step": 6975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513095699015138e-05,
+      "loss": 0.9336,
+      "step": 6976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511338770897036e-05,
+      "loss": 0.9141,
+      "step": 6977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509581719139273e-05,
+      "loss": 0.8047,
+      "step": 6978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507824543803721e-05,
+      "loss": 0.8672,
+      "step": 6979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506067244952251e-05,
+      "loss": 0.8516,
+      "step": 6980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3504309822646737e-05,
+      "loss": 0.8945,
+      "step": 6981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502552276949066e-05,
+      "loss": 0.9062,
+      "step": 6982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350079460792112e-05,
+      "loss": 0.9531,
+      "step": 6983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349903681562479e-05,
+      "loss": 0.9258,
+      "step": 6984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497278900121972e-05,
+      "loss": 0.832,
+      "step": 6985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495520861474565e-05,
+      "loss": 0.9375,
+      "step": 6986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349376269974447e-05,
+      "loss": 0.9609,
+      "step": 6987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492004414993597e-05,
+      "loss": 0.8594,
+      "step": 6988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3490246007283857e-05,
+      "loss": 0.9648,
+      "step": 6989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3488487476677169e-05,
+      "loss": 0.9141,
+      "step": 6990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3486728823235444e-05,
+      "loss": 1.1719,
+      "step": 6991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348497004702062e-05,
+      "loss": 0.8359,
+      "step": 6992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3483211148094615e-05,
+      "loss": 0.9883,
+      "step": 6993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348145212651937e-05,
+      "loss": 0.7617,
+      "step": 6994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347969298235682e-05,
+      "loss": 0.9219,
+      "step": 6995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477933715668907e-05,
+      "loss": 1.0234,
+      "step": 6996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476174326517578e-05,
+      "loss": 1.0781,
+      "step": 6997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3474414814964784e-05,
+      "loss": 0.9688,
+      "step": 6998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347265518107248e-05,
+      "loss": 0.9141,
+      "step": 6999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470895424902626e-05,
+      "loss": 0.8281,
+      "step": 7000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469135546517183e-05,
+      "loss": 1.0156,
+      "step": 7001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467375545978123e-05,
+      "loss": 0.8906,
+      "step": 7002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465615423347416e-05,
+      "loss": 0.8594,
+      "step": 7003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463855178687037e-05,
+      "loss": 0.8984,
+      "step": 7004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346209481205897e-05,
+      "loss": 0.9102,
+      "step": 7005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460334323525202e-05,
+      "loss": 1.0,
+      "step": 7006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458573713147716e-05,
+      "loss": 0.7773,
+      "step": 7007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3456812980988513e-05,
+      "loss": 0.9414,
+      "step": 7008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455052127109584e-05,
+      "loss": 0.7891,
+      "step": 7009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453291151572938e-05,
+      "loss": 0.8359,
+      "step": 7010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3451530054440576e-05,
+      "loss": 0.8906,
+      "step": 7011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3449768835774514e-05,
+      "loss": 0.7656,
+      "step": 7012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448007495636763e-05,
+      "loss": 0.9414,
+      "step": 7013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446246034089347e-05,
+      "loss": 0.832,
+      "step": 7014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3444484451194284e-05,
+      "loss": 0.9961,
+      "step": 7015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344272274701361e-05,
+      "loss": 0.9531,
+      "step": 7016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440960921609348e-05,
+      "loss": 0.9648,
+      "step": 7017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3439198975043542e-05,
+      "loss": 0.8516,
+      "step": 7018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437436907378225e-05,
+      "loss": 0.9336,
+      "step": 7019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435674718675453e-05,
+      "loss": 0.9492,
+      "step": 7020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3433912408997265e-05,
+      "loss": 0.8906,
+      "step": 7021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432149978405722e-05,
+      "loss": 0.8789,
+      "step": 7022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430387426962877e-05,
+      "loss": 1.0391,
+      "step": 7023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3428624754730797e-05,
+      "loss": 0.9492,
+      "step": 7024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426861961771543e-05,
+      "loss": 0.8555,
+      "step": 7025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342509904814719e-05,
+      "loss": 0.793,
+      "step": 7026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342333601391981e-05,
+      "loss": 0.8906,
+      "step": 7027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421572859151484e-05,
+      "loss": 0.9102,
+      "step": 7028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419809583904293e-05,
+      "loss": 1.0078,
+      "step": 7029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418046188240328e-05,
+      "loss": 0.9258,
+      "step": 7030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416282672221679e-05,
+      "loss": 1.0156,
+      "step": 7031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414519035910445e-05,
+      "loss": 0.8789,
+      "step": 7032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3412755279368718e-05,
+      "loss": 0.9336,
+      "step": 7033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3410991402658614e-05,
+      "loss": 0.8516,
+      "step": 7034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340922740584223e-05,
+      "loss": 0.9102,
+      "step": 7035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3407463288981692e-05,
+      "loss": 0.918,
+      "step": 7036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340569905213911e-05,
+      "loss": 1.0312,
+      "step": 7037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403934695376601e-05,
+      "loss": 0.8945,
+      "step": 7038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3402170218756299e-05,
+      "loss": 0.8594,
+      "step": 7039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340040562234033e-05,
+      "loss": 0.9141,
+      "step": 7040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398640906190828e-05,
+      "loss": 0.875,
+      "step": 7041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396876070369933e-05,
+      "loss": 0.8398,
+      "step": 7042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395111114939786e-05,
+      "loss": 0.9062,
+      "step": 7043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393346039962533e-05,
+      "loss": 0.875,
+      "step": 7044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3391580845500332e-05,
+      "loss": 0.8828,
+      "step": 7045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389815531615326e-05,
+      "loss": 0.9648,
+      "step": 7046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388050098369686e-05,
+      "loss": 0.9492,
+      "step": 7047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338628454582557e-05,
+      "loss": 0.8281,
+      "step": 7048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338451887404515e-05,
+      "loss": 0.7969,
+      "step": 7049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382753083090588e-05,
+      "loss": 0.9219,
+      "step": 7050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380987173024071e-05,
+      "loss": 0.8008,
+      "step": 7051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379221143907775e-05,
+      "loss": 0.7969,
+      "step": 7052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377454995803884e-05,
+      "loss": 0.8672,
+      "step": 7053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337568872877459e-05,
+      "loss": 0.8789,
+      "step": 7054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373922342882082e-05,
+      "loss": 0.8438,
+      "step": 7055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337215583818856e-05,
+      "loss": 0.8359,
+      "step": 7056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370389214756223e-05,
+      "loss": 0.875,
+      "step": 7057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336862247264728e-05,
+      "loss": 0.9648,
+      "step": 7058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366855611923938e-05,
+      "loss": 0.9648,
+      "step": 7059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336508863264841e-05,
+      "loss": 0.8789,
+      "step": 7060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363321534882914e-05,
+      "loss": 0.9062,
+      "step": 7061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3361554318689679e-05,
+      "loss": 0.8281,
+      "step": 7062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335978698413092e-05,
+      "loss": 0.8555,
+      "step": 7063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358019531268878e-05,
+      "loss": 0.9648,
+      "step": 7064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356251960165782e-05,
+      "loss": 0.8711,
+      "step": 7065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354484270883874e-05,
+      "loss": 0.7969,
+      "step": 7066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352716463485396e-05,
+      "loss": 0.9688,
+      "step": 7067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350948538032591e-05,
+      "loss": 0.8828,
+      "step": 7068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334918049458772e-05,
+      "loss": 0.8359,
+      "step": 7069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347412333213025e-05,
+      "loss": 0.7734,
+      "step": 7070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334564405397078e-05,
+      "loss": 0.9961,
+      "step": 7071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343875656923237e-05,
+      "loss": 0.8555,
+      "step": 7072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342107142132674e-05,
+      "loss": 0.9102,
+      "step": 7073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3340338509661355e-05,
+      "loss": 0.9453,
+      "step": 7074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333856975957156e-05,
+      "loss": 0.8477,
+      "step": 7075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333680089192557e-05,
+      "loss": 0.8945,
+      "step": 7076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335031906785669e-05,
+      "loss": 0.8633,
+      "step": 7077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333262804214144e-05,
+      "loss": 0.9258,
+      "step": 7078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333149358427329e-05,
+      "loss": 0.8906,
+      "step": 7079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329724247025402e-05,
+      "loss": 0.957,
+      "step": 7080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327954792532785e-05,
+      "loss": 0.8945,
+      "step": 7081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332618522085774e-05,
+      "loss": 0.9531,
+      "step": 7082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324415532062575e-05,
+      "loss": 0.8633,
+      "step": 7083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332264572620961e-05,
+      "loss": 0.8477,
+      "step": 7084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320875803361158e-05,
+      "loss": 1.0,
+      "step": 7085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319105763579539e-05,
+      "loss": 1.0,
+      "step": 7086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317335606927084e-05,
+      "loss": 0.8633,
+      "step": 7087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315565333466118e-05,
+      "loss": 1.1797,
+      "step": 7088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331379494325898e-05,
+      "loss": 0.8242,
+      "step": 7089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312024436368003e-05,
+      "loss": 0.8555,
+      "step": 7090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310253812855534e-05,
+      "loss": 0.9219,
+      "step": 7091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308483072783918e-05,
+      "loss": 0.832,
+      "step": 7092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306712216215502e-05,
+      "loss": 0.9688,
+      "step": 7093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304941243212645e-05,
+      "loss": 0.9062,
+      "step": 7094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3303170153837703e-05,
+      "loss": 0.8047,
+      "step": 7095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301398948153042e-05,
+      "loss": 0.918,
+      "step": 7096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3299627626221024e-05,
+      "loss": 0.8945,
+      "step": 7097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297856188104023e-05,
+      "loss": 0.8594,
+      "step": 7098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296084633864415e-05,
+      "loss": 0.8398,
+      "step": 7099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294312963564578e-05,
+      "loss": 0.8906,
+      "step": 7100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292541177266893e-05,
+      "loss": 0.7812,
+      "step": 7101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290769275033753e-05,
+      "loss": 0.8086,
+      "step": 7102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288997256927542e-05,
+      "loss": 0.9141,
+      "step": 7103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287225123010664e-05,
+      "loss": 0.8828,
+      "step": 7104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285452873345508e-05,
+      "loss": 0.793,
+      "step": 7105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328368050799449e-05,
+      "loss": 0.8477,
+      "step": 7106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281908027020005e-05,
+      "loss": 0.8047,
+      "step": 7107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280135430484476e-05,
+      "loss": 1.1406,
+      "step": 7108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3278362718450311e-05,
+      "loss": 0.9023,
+      "step": 7109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276589890979935e-05,
+      "loss": 0.9531,
+      "step": 7110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3274816948135766e-05,
+      "loss": 0.8906,
+      "step": 7111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327304388998024e-05,
+      "loss": 1.0391,
+      "step": 7112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271270716575786e-05,
+      "loss": 0.75,
+      "step": 7113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269497427984836e-05,
+      "loss": 0.8906,
+      "step": 7114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267724024269834e-05,
+      "loss": 0.9922,
+      "step": 7115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265950505493226e-05,
+      "loss": 1.0,
+      "step": 7116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3264176871717456e-05,
+      "loss": 0.8984,
+      "step": 7117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326240312300498e-05,
+      "loss": 0.8477,
+      "step": 7118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3260629259418252e-05,
+      "loss": 1.0312,
+      "step": 7119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258855281019734e-05,
+      "loss": 0.8516,
+      "step": 7120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257081187871888e-05,
+      "loss": 0.9297,
+      "step": 7121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3255306980037186e-05,
+      "loss": 0.9258,
+      "step": 7122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253532657578098e-05,
+      "loss": 0.7539,
+      "step": 7123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251758220557102e-05,
+      "loss": 1.0156,
+      "step": 7124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249983669036678e-05,
+      "loss": 0.7578,
+      "step": 7125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248209003079313e-05,
+      "loss": 0.9844,
+      "step": 7126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246434222747494e-05,
+      "loss": 0.8438,
+      "step": 7127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244659328103713e-05,
+      "loss": 1.0156,
+      "step": 7128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242884319210463e-05,
+      "loss": 0.8047,
+      "step": 7129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241109196130256e-05,
+      "loss": 0.8789,
+      "step": 7130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239333958925584e-05,
+      "loss": 0.9297,
+      "step": 7131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237558607658965e-05,
+      "loss": 0.8633,
+      "step": 7132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235783142392906e-05,
+      "loss": 0.9062,
+      "step": 7133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323400756318993e-05,
+      "loss": 0.9336,
+      "step": 7134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232231870112554e-05,
+      "loss": 0.8633,
+      "step": 7135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230456063223303e-05,
+      "loss": 0.9805,
+      "step": 7136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228680142584703e-05,
+      "loss": 0.875,
+      "step": 7137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3226904108259292e-05,
+      "loss": 0.957,
+      "step": 7138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225127960309605e-05,
+      "loss": 0.7852,
+      "step": 7139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223351698798186e-05,
+      "loss": 1.0,
+      "step": 7140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3221575323787572e-05,
+      "loss": 0.8477,
+      "step": 7141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219798835340321e-05,
+      "loss": 0.8125,
+      "step": 7142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218022233518979e-05,
+      "loss": 0.7812,
+      "step": 7143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3216245518386105e-05,
+      "loss": 0.8438,
+      "step": 7144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214468690004263e-05,
+      "loss": 0.8789,
+      "step": 7145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212691748436015e-05,
+      "loss": 0.9766,
+      "step": 7146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210914693743929e-05,
+      "loss": 0.8086,
+      "step": 7147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209137525990579e-05,
+      "loss": 0.9336,
+      "step": 7148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207360245238543e-05,
+      "loss": 0.832,
+      "step": 7149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3205582851550398e-05,
+      "loss": 0.8828,
+      "step": 7150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3203805344988734e-05,
+      "loss": 0.9336,
+      "step": 7151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202027725616135e-05,
+      "loss": 1.0938,
+      "step": 7152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200249993495198e-05,
+      "loss": 1.0469,
+      "step": 7153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198472148688515e-05,
+      "loss": 0.8516,
+      "step": 7154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3196694191258691e-05,
+      "loss": 0.9336,
+      "step": 7155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194916121268328e-05,
+      "loss": 0.9922,
+      "step": 7156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193137938780036e-05,
+      "loss": 1.0938,
+      "step": 7157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191359643856428e-05,
+      "loss": 0.8672,
+      "step": 7158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3189581236560118e-05,
+      "loss": 0.8438,
+      "step": 7159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187802716953728e-05,
+      "loss": 0.957,
+      "step": 7160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3186024085099885e-05,
+      "loss": 0.7891,
+      "step": 7161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318424534106121e-05,
+      "loss": 0.875,
+      "step": 7162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3182466484900345e-05,
+      "loss": 0.918,
+      "step": 7163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180687516679918e-05,
+      "loss": 0.9258,
+      "step": 7164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178908436462575e-05,
+      "loss": 0.9453,
+      "step": 7165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3177129244310954e-05,
+      "loss": 0.9414,
+      "step": 7166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317534994028771e-05,
+      "loss": 1.0,
+      "step": 7167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173570524455491e-05,
+      "loss": 0.8945,
+      "step": 7168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3171790996876954e-05,
+      "loss": 0.8789,
+      "step": 7169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170011357614758e-05,
+      "loss": 0.8906,
+      "step": 7170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3168231606731569e-05,
+      "loss": 0.8984,
+      "step": 7171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3166451744290052e-05,
+      "loss": 0.9062,
+      "step": 7172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164671770352881e-05,
+      "loss": 0.8789,
+      "step": 7173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316289168498273e-05,
+      "loss": 0.9453,
+      "step": 7174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316111148824228e-05,
+      "loss": 0.8672,
+      "step": 7175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159331180194212e-05,
+      "loss": 0.8867,
+      "step": 7176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157550760901219e-05,
+      "loss": 0.9102,
+      "step": 7177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155770230425984e-05,
+      "loss": 0.9414,
+      "step": 7178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3153989588831211e-05,
+      "loss": 0.7656,
+      "step": 7179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315220883617959e-05,
+      "loss": 0.8086,
+      "step": 7180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150427972533832e-05,
+      "loss": 0.8633,
+      "step": 7181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314864699795664e-05,
+      "loss": 0.8945,
+      "step": 7182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146865912510727e-05,
+      "loss": 0.8516,
+      "step": 7183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3145084716258804e-05,
+      "loss": 0.957,
+      "step": 7184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3143303409263594e-05,
+      "loss": 0.9961,
+      "step": 7185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141521991587817e-05,
+      "loss": 0.957,
+      "step": 7186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139740463294201e-05,
+      "loss": 0.8867,
+      "step": 7187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3137958824445476e-05,
+      "loss": 0.9297,
+      "step": 7188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136177075104375e-05,
+      "loss": 0.8125,
+      "step": 7189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3134395215333635e-05,
+      "loss": 0.7266,
+      "step": 7190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132613245196005e-05,
+      "loss": 0.9492,
+      "step": 7191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130831164754222e-05,
+      "loss": 0.8906,
+      "step": 7192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129048974071039e-05,
+      "loss": 0.8945,
+      "step": 7193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312726667320921e-05,
+      "loss": 1.0625,
+      "step": 7194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125484262231498e-05,
+      "loss": 0.8633,
+      "step": 7195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123701741200653e-05,
+      "loss": 0.7969,
+      "step": 7196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121919110179453e-05,
+      "loss": 0.9297,
+      "step": 7197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120136369230656e-05,
+      "loss": 0.8203,
+      "step": 7198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118353518417042e-05,
+      "loss": 0.75,
+      "step": 7199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3116570557801384e-05,
+      "loss": 1.0234,
+      "step": 7200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3114787487446468e-05,
+      "loss": 0.8711,
+      "step": 7201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311300430741507e-05,
+      "loss": 0.75,
+      "step": 7202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311122101776999e-05,
+      "loss": 0.9219,
+      "step": 7203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109437618574014e-05,
+      "loss": 0.75,
+      "step": 7204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3107654109889935e-05,
+      "loss": 0.8789,
+      "step": 7205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105870491780559e-05,
+      "loss": 0.832,
+      "step": 7206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104086764308686e-05,
+      "loss": 0.8242,
+      "step": 7207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102302927537127e-05,
+      "loss": 0.8984,
+      "step": 7208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310051898152869e-05,
+      "loss": 0.9883,
+      "step": 7209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3098734926346192e-05,
+      "loss": 0.8203,
+      "step": 7210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096950762052454e-05,
+      "loss": 0.8555,
+      "step": 7211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095166488710297e-05,
+      "loss": 0.9492,
+      "step": 7212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309338210638255e-05,
+      "loss": 0.8633,
+      "step": 7213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3091597615132044e-05,
+      "loss": 0.8906,
+      "step": 7214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3089813015021611e-05,
+      "loss": 0.9258,
+      "step": 7215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3088028306114091e-05,
+      "loss": 0.8828,
+      "step": 7216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086243488472326e-05,
+      "loss": 1.0547,
+      "step": 7217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084458562159165e-05,
+      "loss": 0.8164,
+      "step": 7218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082673527237452e-05,
+      "loss": 0.9453,
+      "step": 7219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308088838377005e-05,
+      "loss": 0.9609,
+      "step": 7220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3079103131819805e-05,
+      "loss": 0.8203,
+      "step": 7221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307731777144959e-05,
+      "loss": 0.9102,
+      "step": 7222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307553230272226e-05,
+      "loss": 0.8906,
+      "step": 7223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073746725700693e-05,
+      "loss": 0.9766,
+      "step": 7224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071961040447754e-05,
+      "loss": 0.9102,
+      "step": 7225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070175247026326e-05,
+      "loss": 0.9453,
+      "step": 7226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068389345499291e-05,
+      "loss": 0.8906,
+      "step": 7227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3066603335929522e-05,
+      "loss": 0.8086,
+      "step": 7228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064817218379919e-05,
+      "loss": 0.8359,
+      "step": 7229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063030992913368e-05,
+      "loss": 0.8203,
+      "step": 7230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061244659592769e-05,
+      "loss": 0.8828,
+      "step": 7231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059458218481016e-05,
+      "loss": 0.8672,
+      "step": 7232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057671669641018e-05,
+      "loss": 0.9102,
+      "step": 7233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055885013135679e-05,
+      "loss": 1.0156,
+      "step": 7234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054098249027909e-05,
+      "loss": 1.0078,
+      "step": 7235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052311377380624e-05,
+      "loss": 0.8906,
+      "step": 7236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050524398256745e-05,
+      "loss": 0.8594,
+      "step": 7237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048737311719189e-05,
+      "loss": 0.793,
+      "step": 7238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046950117830888e-05,
+      "loss": 0.9062,
+      "step": 7239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045162816654769e-05,
+      "loss": 0.8633,
+      "step": 7240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304337540825376e-05,
+      "loss": 0.8125,
+      "step": 7241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304158789269081e-05,
+      "loss": 0.8945,
+      "step": 7242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303980027002885e-05,
+      "loss": 0.8789,
+      "step": 7243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038012540330834e-05,
+      "loss": 0.8516,
+      "step": 7244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036224703659702e-05,
+      "loss": 0.8242,
+      "step": 7245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303443676007841e-05,
+      "loss": 0.9219,
+      "step": 7246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032648709649913e-05,
+      "loss": 1.0078,
+      "step": 7247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030860552437177e-05,
+      "loss": 0.8359,
+      "step": 7248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302907228850316e-05,
+      "loss": 1.0781,
+      "step": 7249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302728391791083e-05,
+      "loss": 0.8438,
+      "step": 7250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025495440723157e-05,
+      "loss": 0.668,
+      "step": 7251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023706857003121e-05,
+      "loss": 0.9141,
+      "step": 7252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021918166813699e-05,
+      "loss": 0.875,
+      "step": 7253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3020129370217871e-05,
+      "loss": 0.8828,
+      "step": 7254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018340467278622e-05,
+      "loss": 0.9297,
+      "step": 7255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016551458058951e-05,
+      "loss": 0.8398,
+      "step": 7256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014762342621841e-05,
+      "loss": 0.9883,
+      "step": 7257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012973121030295e-05,
+      "loss": 0.9062,
+      "step": 7258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011183793347313e-05,
+      "loss": 0.8281,
+      "step": 7259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009394359635903e-05,
+      "loss": 0.9023,
+      "step": 7260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300760481995907e-05,
+      "loss": 0.8828,
+      "step": 7261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005815174379826e-05,
+      "loss": 0.8984,
+      "step": 7262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004025422961189e-05,
+      "loss": 0.9219,
+      "step": 7263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3002235565766179e-05,
+      "loss": 0.7812,
+      "step": 7264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000445602857821e-05,
+      "loss": 0.8047,
+      "step": 7265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998655534299137e-05,
+      "loss": 0.7656,
+      "step": 7266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996865360153166e-05,
+      "loss": 0.9375,
+      "step": 7267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2995075080482935e-05,
+      "loss": 1.0156,
+      "step": 7268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993284695351487e-05,
+      "loss": 1.0703,
+      "step": 7269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2991494204821863e-05,
+      "loss": 1.0078,
+      "step": 7270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2989703608957109e-05,
+      "loss": 0.8594,
+      "step": 7271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987912907820274e-05,
+      "loss": 0.8555,
+      "step": 7272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2986122101474416e-05,
+      "loss": 0.8438,
+      "step": 7273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2984331189982587e-05,
+      "loss": 0.8672,
+      "step": 7274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982540173407849e-05,
+      "loss": 0.8906,
+      "step": 7275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980749051813265e-05,
+      "loss": 0.957,
+      "step": 7276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2978957825261905e-05,
+      "loss": 0.8203,
+      "step": 7277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977166493816839e-05,
+      "loss": 0.9688,
+      "step": 7278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2975375057541146e-05,
+      "loss": 0.8906,
+      "step": 7279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2973583516497903e-05,
+      "loss": 1.0,
+      "step": 7280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971791870750194e-05,
+      "loss": 0.9102,
+      "step": 7281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970000120361101e-05,
+      "loss": 0.8594,
+      "step": 7282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2968208265393724e-05,
+      "loss": 0.8242,
+      "step": 7283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966416305911147e-05,
+      "loss": 0.8398,
+      "step": 7284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964624241976473e-05,
+      "loss": 0.9805,
+      "step": 7285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29628320736528e-05,
+      "loss": 0.8164,
+      "step": 7286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961039801003239e-05,
+      "loss": 0.9727,
+      "step": 7287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959247424090892e-05,
+      "loss": 0.9023,
+      "step": 7288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957454942978875e-05,
+      "loss": 0.9414,
+      "step": 7289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2955662357730303e-05,
+      "loss": 0.7969,
+      "step": 7290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953869668408296e-05,
+      "loss": 0.9258,
+      "step": 7291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952076875075976e-05,
+      "loss": 0.8516,
+      "step": 7292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950283977796475e-05,
+      "loss": 0.9219,
+      "step": 7293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948490976632914e-05,
+      "loss": 0.832,
+      "step": 7294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946697871648439e-05,
+      "loss": 0.9375,
+      "step": 7295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944904662906178e-05,
+      "loss": 0.918,
+      "step": 7296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943111350469279e-05,
+      "loss": 0.75,
+      "step": 7297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941317934400885e-05,
+      "loss": 0.8438,
+      "step": 7298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939524414764146e-05,
+      "loss": 0.8516,
+      "step": 7299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2937730791622212e-05,
+      "loss": 0.7969,
+      "step": 7300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935937065038243e-05,
+      "loss": 0.8984,
+      "step": 7301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934143235075393e-05,
+      "loss": 0.8477,
+      "step": 7302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2932349301796833e-05,
+      "loss": 0.8828,
+      "step": 7303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930555265265725e-05,
+      "loss": 1.0156,
+      "step": 7304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928761125545244e-05,
+      "loss": 0.8359,
+      "step": 7305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2926966882698557e-05,
+      "loss": 0.8828,
+      "step": 7306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925172536788851e-05,
+      "loss": 1.0391,
+      "step": 7307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923378087879301e-05,
+      "loss": 0.9102,
+      "step": 7308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921583536033098e-05,
+      "loss": 0.8242,
+      "step": 7309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919788881313423e-05,
+      "loss": 0.875,
+      "step": 7310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917994123783476e-05,
+      "loss": 1.0391,
+      "step": 7311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291619926350645e-05,
+      "loss": 0.918,
+      "step": 7312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2914404300545549e-05,
+      "loss": 0.8008,
+      "step": 7313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912609234963967e-05,
+      "loss": 0.875,
+      "step": 7314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291081406682492e-05,
+      "loss": 0.7891,
+      "step": 7315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2909018796191616e-05,
+      "loss": 0.8906,
+      "step": 7316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907223423127266e-05,
+      "loss": 0.9023,
+      "step": 7317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905427947695092e-05,
+      "loss": 1.0625,
+      "step": 7318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903632369958315e-05,
+      "loss": 0.8906,
+      "step": 7319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2901836689980158e-05,
+      "loss": 0.75,
+      "step": 7320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290004090782385e-05,
+      "loss": 0.918,
+      "step": 7321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898245023552626e-05,
+      "loss": 0.9922,
+      "step": 7322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2896449037229717e-05,
+      "loss": 0.8594,
+      "step": 7323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2894652948918369e-05,
+      "loss": 0.8086,
+      "step": 7324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892856758681818e-05,
+      "loss": 0.8633,
+      "step": 7325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891060466583317e-05,
+      "loss": 0.7773,
+      "step": 7326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288926407268611e-05,
+      "loss": 0.8047,
+      "step": 7327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887467577053455e-05,
+      "loss": 0.8281,
+      "step": 7328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885670979748608e-05,
+      "loss": 0.9258,
+      "step": 7329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2883874280834832e-05,
+      "loss": 0.9258,
+      "step": 7330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882077480375388e-05,
+      "loss": 0.8555,
+      "step": 7331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2880280578433548e-05,
+      "loss": 0.8672,
+      "step": 7332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287848357507258e-05,
+      "loss": 0.9531,
+      "step": 7333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287668647035576e-05,
+      "loss": 1.0469,
+      "step": 7334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2874889264346371e-05,
+      "loss": 0.9453,
+      "step": 7335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.287309195710769e-05,
+      "loss": 0.8438,
+      "step": 7336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871294548703007e-05,
+      "loss": 0.793,
+      "step": 7337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869497039195609e-05,
+      "loss": 0.9453,
+      "step": 7338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2867699428648788e-05,
+      "loss": 0.8984,
+      "step": 7339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865901717125843e-05,
+      "loss": 0.8906,
+      "step": 7340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864103904690073e-05,
+      "loss": 0.8516,
+      "step": 7341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862305991404784e-05,
+      "loss": 1.0312,
+      "step": 7342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860507977333282e-05,
+      "loss": 0.8125,
+      "step": 7343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2858709862538875e-05,
+      "loss": 0.9727,
+      "step": 7344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856911647084881e-05,
+      "loss": 0.9141,
+      "step": 7345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855113331034618e-05,
+      "loss": 0.9414,
+      "step": 7346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853314914451408e-05,
+      "loss": 0.8945,
+      "step": 7347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851516397398571e-05,
+      "loss": 0.8633,
+      "step": 7348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849717779939439e-05,
+      "loss": 0.8906,
+      "step": 7349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847919062137346e-05,
+      "loss": 0.8555,
+      "step": 7350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846120244055624e-05,
+      "loss": 0.8359,
+      "step": 7351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844321325757615e-05,
+      "loss": 0.8672,
+      "step": 7352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284252230730666e-05,
+      "loss": 0.8086,
+      "step": 7353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2840723188766107e-05,
+      "loss": 0.8086,
+      "step": 7354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2838923970199301e-05,
+      "loss": 0.8828,
+      "step": 7355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28371246516696e-05,
+      "loss": 0.9258,
+      "step": 7356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2835325233240362e-05,
+      "loss": 0.7891,
+      "step": 7357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833525714974944e-05,
+      "loss": 0.8672,
+      "step": 7358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831726096936708e-05,
+      "loss": 0.875,
+      "step": 7359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829926379189025e-05,
+      "loss": 0.8281,
+      "step": 7360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828126561795266e-05,
+      "loss": 0.8086,
+      "step": 7361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2826326644818806e-05,
+      "loss": 0.9727,
+      "step": 7362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2824526628323018e-05,
+      "loss": 0.7891,
+      "step": 7363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822726512371286e-05,
+      "loss": 0.8828,
+      "step": 7364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2820926297026996e-05,
+      "loss": 0.8711,
+      "step": 7365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819125982353537e-05,
+      "loss": 1.0078,
+      "step": 7366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 1.0703,
+      "step": 7367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2815525055272677e-05,
+      "loss": 0.8828,
+      "step": 7368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813724442992068e-05,
+      "loss": 0.8242,
+      "step": 7369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281192373163588e-05,
+      "loss": 0.9062,
+      "step": 7370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810122921267516e-05,
+      "loss": 0.875,
+      "step": 7371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808322011950386e-05,
+      "loss": 1.0078,
+      "step": 7372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28065210037479e-05,
+      "loss": 0.9375,
+      "step": 7373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804719896723476e-05,
+      "loss": 0.8242,
+      "step": 7374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802918690940538e-05,
+      "loss": 0.8398,
+      "step": 7375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801117386462499e-05,
+      "loss": 0.9258,
+      "step": 7376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799315983352796e-05,
+      "loss": 0.9844,
+      "step": 7377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797514481674853e-05,
+      "loss": 0.9141,
+      "step": 7378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795712881492109e-05,
+      "loss": 0.8555,
+      "step": 7379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793911182867992e-05,
+      "loss": 0.8711,
+      "step": 7380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792109385865954e-05,
+      "loss": 0.8984,
+      "step": 7381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2790307490549432e-05,
+      "loss": 0.8281,
+      "step": 7382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788505496981874e-05,
+      "loss": 0.832,
+      "step": 7383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786703405226732e-05,
+      "loss": 0.9141,
+      "step": 7384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2784901215347462e-05,
+      "loss": 0.8398,
+      "step": 7385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783098927407518e-05,
+      "loss": 0.8047,
+      "step": 7386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781296541470365e-05,
+      "loss": 0.8672,
+      "step": 7387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2779494057599466e-05,
+      "loss": 0.9023,
+      "step": 7388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777691475858292e-05,
+      "loss": 0.7695,
+      "step": 7389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277588879631031e-05,
+      "loss": 1.0234,
+      "step": 7390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774086019018999e-05,
+      "loss": 0.9414,
+      "step": 7391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772283144047832e-05,
+      "loss": 0.8555,
+      "step": 7392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2770480171460302e-05,
+      "loss": 0.9258,
+      "step": 7393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768677101319881e-05,
+      "loss": 0.793,
+      "step": 7394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276687393369007e-05,
+      "loss": 0.9492,
+      "step": 7395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765070668634351e-05,
+      "loss": 0.8477,
+      "step": 7396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763267306216227e-05,
+      "loss": 0.9297,
+      "step": 7397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761463846499193e-05,
+      "loss": 0.8398,
+      "step": 7398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759660289546756e-05,
+      "loss": 0.8789,
+      "step": 7399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275785663542242e-05,
+      "loss": 0.9766,
+      "step": 7400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756052884189689e-05,
+      "loss": 0.9727,
+      "step": 7401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2754249035912086e-05,
+      "loss": 0.875,
+      "step": 7402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2752445090653117e-05,
+      "loss": 0.8477,
+      "step": 7403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750641048476309e-05,
+      "loss": 0.9844,
+      "step": 7404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748836909445182e-05,
+      "loss": 0.875,
+      "step": 7405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2747032673623262e-05,
+      "loss": 0.9258,
+      "step": 7406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745228341074082e-05,
+      "loss": 0.8164,
+      "step": 7407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743423911861175e-05,
+      "loss": 0.8047,
+      "step": 7408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741619386048075e-05,
+      "loss": 0.9414,
+      "step": 7409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739814763698322e-05,
+      "loss": 0.8906,
+      "step": 7410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738010044875461e-05,
+      "loss": 1.0859,
+      "step": 7411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736205229643042e-05,
+      "loss": 0.8281,
+      "step": 7412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273440031806461e-05,
+      "loss": 0.9102,
+      "step": 7413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273259531020372e-05,
+      "loss": 0.832,
+      "step": 7414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273079020612393e-05,
+      "loss": 1.0547,
+      "step": 7415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27289850058888e-05,
+      "loss": 0.8086,
+      "step": 7416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727179709561897e-05,
+      "loss": 0.7773,
+      "step": 7417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725374317206782e-05,
+      "loss": 0.8008,
+      "step": 7418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723568828887034e-05,
+      "loss": 0.8984,
+      "step": 7419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721763244666218e-05,
+      "loss": 0.8516,
+      "step": 7420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719957564607918e-05,
+      "loss": 0.9375,
+      "step": 7421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718151788775712e-05,
+      "loss": 0.8672,
+      "step": 7422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2716345917233187e-05,
+      "loss": 0.8867,
+      "step": 7423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714539950043924e-05,
+      "loss": 0.9102,
+      "step": 7424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2712733887271522e-05,
+      "loss": 0.8633,
+      "step": 7425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2710927728979568e-05,
+      "loss": 0.9453,
+      "step": 7426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709121475231668e-05,
+      "loss": 0.8242,
+      "step": 7427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270731512609141e-05,
+      "loss": 0.9688,
+      "step": 7428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705508681622414e-05,
+      "loss": 1.0078,
+      "step": 7429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703702141888277e-05,
+      "loss": 0.9805,
+      "step": 7430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701895506952613e-05,
+      "loss": 0.9453,
+      "step": 7431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700088776879035e-05,
+      "loss": 0.8828,
+      "step": 7432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2698281951731164e-05,
+      "loss": 0.8828,
+      "step": 7433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696475031572622e-05,
+      "loss": 0.8359,
+      "step": 7434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694668016467025e-05,
+      "loss": 0.8672,
+      "step": 7435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269286090647801e-05,
+      "loss": 1.0547,
+      "step": 7436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2691053701669204e-05,
+      "loss": 0.8516,
+      "step": 7437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689246402104244e-05,
+      "loss": 0.8711,
+      "step": 7438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687439007846762e-05,
+      "loss": 0.7734,
+      "step": 7439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2685631518960408e-05,
+      "loss": 0.8047,
+      "step": 7440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683823935508817e-05,
+      "loss": 1.0234,
+      "step": 7441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682016257555645e-05,
+      "loss": 0.8555,
+      "step": 7442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680208485164537e-05,
+      "loss": 0.8164,
+      "step": 7443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678400618399152e-05,
+      "loss": 0.957,
+      "step": 7444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676592657323145e-05,
+      "loss": 0.7578,
+      "step": 7445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2674784602000179e-05,
+      "loss": 0.9492,
+      "step": 7446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2672976452493914e-05,
+      "loss": 0.793,
+      "step": 7447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671168208868024e-05,
+      "loss": 0.8984,
+      "step": 7448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669359871186175e-05,
+      "loss": 0.8477,
+      "step": 7449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667551439512047e-05,
+      "loss": 0.9258,
+      "step": 7450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665742913909308e-05,
+      "loss": 1.0234,
+      "step": 7451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2663934294441652e-05,
+      "loss": 0.957,
+      "step": 7452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662125581172753e-05,
+      "loss": 0.9961,
+      "step": 7453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660316774166303e-05,
+      "loss": 0.8867,
+      "step": 7454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265850787348599e-05,
+      "loss": 0.957,
+      "step": 7455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2656698879195513e-05,
+      "loss": 0.8945,
+      "step": 7456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2654889791358562e-05,
+      "loss": 0.9648,
+      "step": 7457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653080610038846e-05,
+      "loss": 1.0312,
+      "step": 7458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2651271335300063e-05,
+      "loss": 0.8516,
+      "step": 7459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649461967205924e-05,
+      "loss": 0.8516,
+      "step": 7460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264765250582014e-05,
+      "loss": 0.9062,
+      "step": 7461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645842951206423e-05,
+      "loss": 0.8125,
+      "step": 7462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2644033303428487e-05,
+      "loss": 0.9297,
+      "step": 7463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264222356255006e-05,
+      "loss": 0.832,
+      "step": 7464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2640413728634859e-05,
+      "loss": 1.0234,
+      "step": 7465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2638603801746616e-05,
+      "loss": 0.8086,
+      "step": 7466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636793781949056e-05,
+      "loss": 0.9531,
+      "step": 7467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263498366930592e-05,
+      "loss": 0.8438,
+      "step": 7468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2633173463880937e-05,
+      "loss": 0.9336,
+      "step": 7469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631363165737854e-05,
+      "loss": 0.9102,
+      "step": 7470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2629552774940408e-05,
+      "loss": 0.9766,
+      "step": 7471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627742291552351e-05,
+      "loss": 0.7578,
+      "step": 7472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625931715637428e-05,
+      "loss": 1.0625,
+      "step": 7473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26241210472594e-05,
+      "loss": 0.9922,
+      "step": 7474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2622310286482011e-05,
+      "loss": 0.8281,
+      "step": 7475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620499433369033e-05,
+      "loss": 0.8828,
+      "step": 7476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618688487984223e-05,
+      "loss": 0.8555,
+      "step": 7477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261687745039135e-05,
+      "loss": 0.8711,
+      "step": 7478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615066320654177e-05,
+      "loss": 0.9688,
+      "step": 7479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613255098836485e-05,
+      "loss": 0.8398,
+      "step": 7480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2611443785002045e-05,
+      "loss": 0.8516,
+      "step": 7481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609632379214636e-05,
+      "loss": 0.7891,
+      "step": 7482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607820881538042e-05,
+      "loss": 0.8828,
+      "step": 7483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260600929203605e-05,
+      "loss": 0.8203,
+      "step": 7484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604197610772447e-05,
+      "loss": 0.8477,
+      "step": 7485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602385837811024e-05,
+      "loss": 0.9922,
+      "step": 7486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600573973215578e-05,
+      "loss": 0.7695,
+      "step": 7487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598762017049906e-05,
+      "loss": 0.8945,
+      "step": 7488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596949969377814e-05,
+      "loss": 0.8086,
+      "step": 7489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.25951378302631e-05,
+      "loss": 0.9336,
+      "step": 7490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259332559976958e-05,
+      "loss": 0.9141,
+      "step": 7491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591513277961058e-05,
+      "loss": 0.9844,
+      "step": 7492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589700864901354e-05,
+      "loss": 1.0,
+      "step": 7493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2587888360654283e-05,
+      "loss": 0.8711,
+      "step": 7494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258607576528367e-05,
+      "loss": 0.8867,
+      "step": 7495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584263078853332e-05,
+      "loss": 0.9062,
+      "step": 7496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2582450301427105e-05,
+      "loss": 0.9258,
+      "step": 7497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580637433068812e-05,
+      "loss": 0.9023,
+      "step": 7498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2578824473842292e-05,
+      "loss": 0.8398,
+      "step": 7499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257701142381138e-05,
+      "loss": 0.9688,
+      "step": 7500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257519828303992e-05,
+      "loss": 0.8672,
+      "step": 7501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573385051591748e-05,
+      "loss": 0.832,
+      "step": 7502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571571729530718e-05,
+      "loss": 0.8086,
+      "step": 7503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2569758316920677e-05,
+      "loss": 0.8281,
+      "step": 7504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567944813825479e-05,
+      "loss": 0.8984,
+      "step": 7505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2566131220308975e-05,
+      "loss": 0.8945,
+      "step": 7506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2564317536435031e-05,
+      "loss": 0.9219,
+      "step": 7507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2562503762267506e-05,
+      "loss": 0.8594,
+      "step": 7508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.256068989787027e-05,
+      "loss": 0.7852,
+      "step": 7509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558875943307186e-05,
+      "loss": 0.8945,
+      "step": 7510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2557061898642132e-05,
+      "loss": 1.0156,
+      "step": 7511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555247763938976e-05,
+      "loss": 0.8516,
+      "step": 7512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553433539261606e-05,
+      "loss": 0.8789,
+      "step": 7513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2551619224673894e-05,
+      "loss": 0.8945,
+      "step": 7514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549804820239736e-05,
+      "loss": 0.9219,
+      "step": 7515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547990326023005e-05,
+      "loss": 0.9375,
+      "step": 7516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546175742087607e-05,
+      "loss": 0.8594,
+      "step": 7517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544361068497426e-05,
+      "loss": 0.8867,
+      "step": 7518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542546305316365e-05,
+      "loss": 0.8672,
+      "step": 7519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2540731452608323e-05,
+      "loss": 0.793,
+      "step": 7520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2538916510437202e-05,
+      "loss": 0.8477,
+      "step": 7521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537101478866914e-05,
+      "loss": 0.8984,
+      "step": 7522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2535286357961363e-05,
+      "loss": 0.8789,
+      "step": 7523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533471147784464e-05,
+      "loss": 0.8359,
+      "step": 7524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531655848400137e-05,
+      "loss": 0.9219,
+      "step": 7525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529840459872298e-05,
+      "loss": 0.9336,
+      "step": 7526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528024982264868e-05,
+      "loss": 0.9062,
+      "step": 7527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526209415641776e-05,
+      "loss": 0.9141,
+      "step": 7528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524393760066952e-05,
+      "loss": 0.8711,
+      "step": 7529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2522578015604325e-05,
+      "loss": 0.9414,
+      "step": 7530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2520762182317831e-05,
+      "loss": 0.8906,
+      "step": 7531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518946260271409e-05,
+      "loss": 0.9375,
+      "step": 7532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2517130249528999e-05,
+      "loss": 0.9062,
+      "step": 7533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515314150154547e-05,
+      "loss": 1.0078,
+      "step": 7534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513497962212003e-05,
+      "loss": 1.0078,
+      "step": 7535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2511681685765308e-05,
+      "loss": 0.7891,
+      "step": 7536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250986532087843e-05,
+      "loss": 0.8281,
+      "step": 7537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508048867615315e-05,
+      "loss": 0.8281,
+      "step": 7538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506232326039927e-05,
+      "loss": 0.8594,
+      "step": 7539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250441569621623e-05,
+      "loss": 0.918,
+      "step": 7540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502598978208189e-05,
+      "loss": 0.8398,
+      "step": 7541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2500782172079778e-05,
+      "loss": 0.8125,
+      "step": 7542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498965277894958e-05,
+      "loss": 0.9492,
+      "step": 7543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497148295717719e-05,
+      "loss": 1.0156,
+      "step": 7544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2495331225612028e-05,
+      "loss": 0.9414,
+      "step": 7545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2493514067641876e-05,
+      "loss": 0.8125,
+      "step": 7546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491696821871239e-05,
+      "loss": 0.8594,
+      "step": 7547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489879488364115e-05,
+      "loss": 0.9023,
+      "step": 7548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488062067184486e-05,
+      "loss": 0.7812,
+      "step": 7549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486244558396351e-05,
+      "loss": 0.9219,
+      "step": 7550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484426962063704e-05,
+      "loss": 0.9648,
+      "step": 7551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482609278250553e-05,
+      "loss": 0.8398,
+      "step": 7552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480791507020893e-05,
+      "loss": 0.8594,
+      "step": 7553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2478973648438736e-05,
+      "loss": 0.8281,
+      "step": 7554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477155702568085e-05,
+      "loss": 0.9805,
+      "step": 7555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247533766947296e-05,
+      "loss": 0.9453,
+      "step": 7556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473519549217373e-05,
+      "loss": 0.9219,
+      "step": 7557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471701341865342e-05,
+      "loss": 0.8281,
+      "step": 7558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469883047480893e-05,
+      "loss": 0.9648,
+      "step": 7559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468064666128048e-05,
+      "loss": 0.9297,
+      "step": 7560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466246197870835e-05,
+      "loss": 0.8242,
+      "step": 7561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464427642773287e-05,
+      "loss": 0.8633,
+      "step": 7562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462609000899433e-05,
+      "loss": 0.793,
+      "step": 7563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460790272313317e-05,
+      "loss": 0.8906,
+      "step": 7564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458971457078975e-05,
+      "loss": 0.9023,
+      "step": 7565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457152555260451e-05,
+      "loss": 0.9062,
+      "step": 7566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245533356692179e-05,
+      "loss": 0.8516,
+      "step": 7567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453514492127047e-05,
+      "loss": 0.8359,
+      "step": 7568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451695330940268e-05,
+      "loss": 0.8672,
+      "step": 7569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2449876083425512e-05,
+      "loss": 0.8047,
+      "step": 7570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448056749646837e-05,
+      "loss": 0.9023,
+      "step": 7571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446237329668302e-05,
+      "loss": 0.9219,
+      "step": 7572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444417823553975e-05,
+      "loss": 0.9531,
+      "step": 7573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2442598231367922e-05,
+      "loss": 0.9023,
+      "step": 7574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440778553174215e-05,
+      "loss": 0.8438,
+      "step": 7575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438958789036926e-05,
+      "loss": 1.0,
+      "step": 7576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437138939020133e-05,
+      "loss": 0.8945,
+      "step": 7577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435319003187911e-05,
+      "loss": 0.8789,
+      "step": 7578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433498981604353e-05,
+      "loss": 0.9922,
+      "step": 7579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431678874333532e-05,
+      "loss": 0.9609,
+      "step": 7580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429858681439549e-05,
+      "loss": 0.8945,
+      "step": 7581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428038402986486e-05,
+      "loss": 0.7695,
+      "step": 7582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426218039038445e-05,
+      "loss": 0.8984,
+      "step": 7583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2424397589659519e-05,
+      "loss": 0.8906,
+      "step": 7584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242257705491381e-05,
+      "loss": 0.9453,
+      "step": 7585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420756434865427e-05,
+      "loss": 0.8359,
+      "step": 7586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418935729578468e-05,
+      "loss": 0.8359,
+      "step": 7587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417114939117048e-05,
+      "loss": 0.8906,
+      "step": 7588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2415294063545277e-05,
+      "loss": 0.9453,
+      "step": 7589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2413473102927276e-05,
+      "loss": 0.9141,
+      "step": 7590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241165205732716e-05,
+      "loss": 0.7891,
+      "step": 7591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240983092680905e-05,
+      "loss": 0.9219,
+      "step": 7592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408009711437072e-05,
+      "loss": 0.9766,
+      "step": 7593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406188411275357e-05,
+      "loss": 0.8555,
+      "step": 7594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2404367026388028e-05,
+      "loss": 0.8906,
+      "step": 7595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240254555683923e-05,
+      "loss": 0.8945,
+      "step": 7596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400724002693089e-05,
+      "loss": 0.9023,
+      "step": 7597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398902364013752e-05,
+      "loss": 0.9141,
+      "step": 7598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2397080640865355e-05,
+      "loss": 0.9648,
+      "step": 7599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395258833312052e-05,
+      "loss": 0.9727,
+      "step": 7600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393436941417985e-05,
+      "loss": 0.8086,
+      "step": 7601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391614965247307e-05,
+      "loss": 0.8945,
+      "step": 7602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389792904864175e-05,
+      "loss": 0.9297,
+      "step": 7603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2387970760332744e-05,
+      "loss": 0.875,
+      "step": 7604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386148531717177e-05,
+      "loss": 0.8125,
+      "step": 7605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384326219081633e-05,
+      "loss": 0.918,
+      "step": 7606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382503822490283e-05,
+      "loss": 0.8203,
+      "step": 7607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380681342007294e-05,
+      "loss": 0.7812,
+      "step": 7608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378858777696842e-05,
+      "loss": 1.0078,
+      "step": 7609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377036129623097e-05,
+      "loss": 0.8672,
+      "step": 7610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237521339785024e-05,
+      "loss": 1.0469,
+      "step": 7611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373390582442454e-05,
+      "loss": 0.9922,
+      "step": 7612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371567683463919e-05,
+      "loss": 0.7148,
+      "step": 7613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2369744700978826e-05,
+      "loss": 0.9453,
+      "step": 7614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367921635051361e-05,
+      "loss": 0.9062,
+      "step": 7615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366098485745719e-05,
+      "loss": 0.7266,
+      "step": 7616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364275253126099e-05,
+      "loss": 0.9883,
+      "step": 7617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362451937256693e-05,
+      "loss": 1.0312,
+      "step": 7618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360628538201709e-05,
+      "loss": 0.9258,
+      "step": 7619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2358805056025349e-05,
+      "loss": 0.8281,
+      "step": 7620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356981490791822e-05,
+      "loss": 0.8047,
+      "step": 7621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2355157842565335e-05,
+      "loss": 0.8789,
+      "step": 7622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353334111410107e-05,
+      "loss": 0.875,
+      "step": 7623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351510297390348e-05,
+      "loss": 0.7656,
+      "step": 7624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2349686400570285e-05,
+      "loss": 0.9062,
+      "step": 7625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347862421014132e-05,
+      "loss": 1.0,
+      "step": 7626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346038358786122e-05,
+      "loss": 0.8125,
+      "step": 7627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344214213950474e-05,
+      "loss": 1.1562,
+      "step": 7628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342389986571428e-05,
+      "loss": 0.7656,
+      "step": 7629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340565676713212e-05,
+      "loss": 0.7812,
+      "step": 7630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338741284440067e-05,
+      "loss": 0.8359,
+      "step": 7631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336916809816226e-05,
+      "loss": 1.0078,
+      "step": 7632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233509225290594e-05,
+      "loss": 0.9492,
+      "step": 7633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2333267613773449e-05,
+      "loss": 1.0234,
+      "step": 7634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331442892483004e-05,
+      "loss": 0.8516,
+      "step": 7635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329618089098854e-05,
+      "loss": 0.9766,
+      "step": 7636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327793203685254e-05,
+      "loss": 0.9414,
+      "step": 7637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232596823630646e-05,
+      "loss": 0.8242,
+      "step": 7638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2324143187026735e-05,
+      "loss": 0.7422,
+      "step": 7639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232231805591034e-05,
+      "loss": 0.9922,
+      "step": 7640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320492843021541e-05,
+      "loss": 0.832,
+      "step": 7641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318667548424603e-05,
+      "loss": 0.8203,
+      "step": 7642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316842172183806e-05,
+      "loss": 0.8594,
+      "step": 7643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315016714363412e-05,
+      "loss": 0.9492,
+      "step": 7644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231319117502771e-05,
+      "loss": 0.8281,
+      "step": 7645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311365554240972e-05,
+      "loss": 0.8008,
+      "step": 7646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2309539852067489e-05,
+      "loss": 0.9688,
+      "step": 7647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307714068571536e-05,
+      "loss": 0.9062,
+      "step": 7648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305888203817414e-05,
+      "loss": 0.9141,
+      "step": 7649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304062257869402e-05,
+      "loss": 0.8672,
+      "step": 7650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302236230791805e-05,
+      "loss": 0.9023,
+      "step": 7651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300410122648911e-05,
+      "loss": 0.8516,
+      "step": 7652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298583933505031e-05,
+      "loss": 0.8555,
+      "step": 7653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229675766342446e-05,
+      "loss": 0.9766,
+      "step": 7654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.22949313124715e-05,
+      "loss": 0.8242,
+      "step": 7655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229310488071047e-05,
+      "loss": 0.9336,
+      "step": 7656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2291278368205675e-05,
+      "loss": 0.8555,
+      "step": 7657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289451775021432e-05,
+      "loss": 0.8203,
+      "step": 7658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287625101222056e-05,
+      "loss": 0.8477,
+      "step": 7659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2285798346871869e-05,
+      "loss": 0.8789,
+      "step": 7660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283971512035191e-05,
+      "loss": 0.8672,
+      "step": 7661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228214459677635e-05,
+      "loss": 0.9531,
+      "step": 7662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280317601159674e-05,
+      "loss": 0.8438,
+      "step": 7663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278490525249498e-05,
+      "loss": 0.9102,
+      "step": 7664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276663369110149e-05,
+      "loss": 0.9375,
+      "step": 7665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274836132805969e-05,
+      "loss": 0.7812,
+      "step": 7666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273008816401298e-05,
+      "loss": 0.8398,
+      "step": 7667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2271181419960473e-05,
+      "loss": 0.8672,
+      "step": 7668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269353943547847e-05,
+      "loss": 0.8633,
+      "step": 7669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2267526387227765e-05,
+      "loss": 0.7734,
+      "step": 7670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265698751064575e-05,
+      "loss": 0.8828,
+      "step": 7671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263871035122636e-05,
+      "loss": 0.918,
+      "step": 7672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262043239466302e-05,
+      "loss": 0.7617,
+      "step": 7673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2260215364159933e-05,
+      "loss": 0.9414,
+      "step": 7674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225838740926789e-05,
+      "loss": 1.0156,
+      "step": 7675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225655937485454e-05,
+      "loss": 0.8047,
+      "step": 7676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254731260984249e-05,
+      "loss": 0.8711,
+      "step": 7677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252903067721391e-05,
+      "loss": 0.9805,
+      "step": 7678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251074795130339e-05,
+      "loss": 0.8672,
+      "step": 7679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2249246443275464e-05,
+      "loss": 0.8945,
+      "step": 7680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224741801222115e-05,
+      "loss": 0.7773,
+      "step": 7681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245589502031781e-05,
+      "loss": 0.9648,
+      "step": 7682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243760912771733e-05,
+      "loss": 0.8672,
+      "step": 7683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241932244505403e-05,
+      "loss": 0.8672,
+      "step": 7684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2240103497297175e-05,
+      "loss": 0.8242,
+      "step": 7685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238274671211447e-05,
+      "loss": 0.8516,
+      "step": 7686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223644576631261e-05,
+      "loss": 0.9258,
+      "step": 7687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2234616782665066e-05,
+      "loss": 0.7891,
+      "step": 7688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2232787720333212e-05,
+      "loss": 0.793,
+      "step": 7689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230958579381458e-05,
+      "loss": 0.8047,
+      "step": 7690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2229129359874203e-05,
+      "loss": 0.9102,
+      "step": 7691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227300061875866e-05,
+      "loss": 0.9297,
+      "step": 7692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225470685450853e-05,
+      "loss": 0.8203,
+      "step": 7693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2223641230663583e-05,
+      "loss": 0.8633,
+      "step": 7694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222181169757847e-05,
+      "loss": 0.957,
+      "step": 7695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219982086259939e-05,
+      "loss": 0.8906,
+      "step": 7696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218152396772408e-05,
+      "loss": 1.0391,
+      "step": 7697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216322629180309e-05,
+      "loss": 0.8281,
+      "step": 7698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214492783548064e-05,
+      "loss": 0.7734,
+      "step": 7699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212662859940113e-05,
+      "loss": 0.8281,
+      "step": 7700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210832858420884e-05,
+      "loss": 0.8906,
+      "step": 7701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220900277905482e-05,
+      "loss": 0.9414,
+      "step": 7702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2207172621906354e-05,
+      "loss": 0.8008,
+      "step": 7703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2205342387039933e-05,
+      "loss": 0.7773,
+      "step": 7704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220351207452e-05,
+      "loss": 0.8477,
+      "step": 7705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201681684411009e-05,
+      "loss": 0.9453,
+      "step": 7706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199851216777403e-05,
+      "loss": 0.8867,
+      "step": 7707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198020671683641e-05,
+      "loss": 0.8555,
+      "step": 7708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2196190049194177e-05,
+      "loss": 0.8867,
+      "step": 7709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219435934937347e-05,
+      "loss": 0.8672,
+      "step": 7710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192528572285985e-05,
+      "loss": 0.8555,
+      "step": 7711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190697717996179e-05,
+      "loss": 0.9609,
+      "step": 7712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2188866786568525e-05,
+      "loss": 0.8281,
+      "step": 7713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187035778067494e-05,
+      "loss": 0.7578,
+      "step": 7714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185204692557556e-05,
+      "loss": 0.832,
+      "step": 7715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183373530103183e-05,
+      "loss": 0.8633,
+      "step": 7716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181542290768862e-05,
+      "loss": 0.9414,
+      "step": 7717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2179710974619066e-05,
+      "loss": 0.9453,
+      "step": 7718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217787958171828e-05,
+      "loss": 0.8164,
+      "step": 7719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2176048112130994e-05,
+      "loss": 1.0312,
+      "step": 7720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2174216565921692e-05,
+      "loss": 0.918,
+      "step": 7721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217238494315487e-05,
+      "loss": 0.9688,
+      "step": 7722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170553243895017e-05,
+      "loss": 0.8281,
+      "step": 7723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168721468206635e-05,
+      "loss": 0.8516,
+      "step": 7724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166889616154223e-05,
+      "loss": 0.9023,
+      "step": 7725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216505768780228e-05,
+      "loss": 0.9805,
+      "step": 7726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163225683215314e-05,
+      "loss": 1.0,
+      "step": 7727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216139360245783e-05,
+      "loss": 0.7734,
+      "step": 7728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159561445594341e-05,
+      "loss": 0.75,
+      "step": 7729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157729212689358e-05,
+      "loss": 0.8828,
+      "step": 7730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21558969038074e-05,
+      "loss": 0.8828,
+      "step": 7731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2154064519012983e-05,
+      "loss": 0.9688,
+      "step": 7732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152232058370627e-05,
+      "loss": 0.8203,
+      "step": 7733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150399521944858e-05,
+      "loss": 0.9922,
+      "step": 7734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21485669098002e-05,
+      "loss": 0.8438,
+      "step": 7735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146734222001186e-05,
+      "loss": 0.7227,
+      "step": 7736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2144901458612346e-05,
+      "loss": 0.9883,
+      "step": 7737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143068619698213e-05,
+      "loss": 0.8125,
+      "step": 7738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141235705323324e-05,
+      "loss": 0.8633,
+      "step": 7739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213940271555222e-05,
+      "loss": 0.8242,
+      "step": 7740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137569650449445e-05,
+      "loss": 1.0391,
+      "step": 7741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213573651007954e-05,
+      "loss": 0.8867,
+      "step": 7742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133903294507056e-05,
+      "loss": 0.793,
+      "step": 7743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132070003796544e-05,
+      "loss": 0.9414,
+      "step": 7744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130236638012551e-05,
+      "loss": 0.8164,
+      "step": 7745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212840319721964e-05,
+      "loss": 0.9453,
+      "step": 7746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2126569681482365e-05,
+      "loss": 0.8906,
+      "step": 7747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212473609086529e-05,
+      "loss": 0.8555,
+      "step": 7748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2122902425432975e-05,
+      "loss": 0.8672,
+      "step": 7749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121068685249989e-05,
+      "loss": 0.9492,
+      "step": 7750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119234870380896e-05,
+      "loss": 1.0469,
+      "step": 7751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117400980890276e-05,
+      "loss": 0.9062,
+      "step": 7752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115567016842696e-05,
+      "loss": 0.875,
+      "step": 7753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113732978302735e-05,
+      "loss": 0.8828,
+      "step": 7754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211189886533497e-05,
+      "loss": 0.875,
+      "step": 7755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211006467800399e-05,
+      "loss": 0.9492,
+      "step": 7756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210823041637437e-05,
+      "loss": 0.9648,
+      "step": 7757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106396080510706e-05,
+      "loss": 0.9219,
+      "step": 7758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104561670477582e-05,
+      "loss": 0.7969,
+      "step": 7759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102727186339592e-05,
+      "loss": 1.0547,
+      "step": 7760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100892628161332e-05,
+      "loss": 0.8594,
+      "step": 7761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099057996007396e-05,
+      "loss": 0.918,
+      "step": 7762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209722328994239e-05,
+      "loss": 0.8477,
+      "step": 7763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2095388510030913e-05,
+      "loss": 0.9297,
+      "step": 7764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209355365633757e-05,
+      "loss": 0.9062,
+      "step": 7765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091718728926972e-05,
+      "loss": 0.8125,
+      "step": 7766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2089883727863727e-05,
+      "loss": 0.8672,
+      "step": 7767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088048653212449e-05,
+      "loss": 1.0859,
+      "step": 7768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086213505037755e-05,
+      "loss": 0.8398,
+      "step": 7769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2084378283404262e-05,
+      "loss": 0.8984,
+      "step": 7770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082542988376591e-05,
+      "loss": 0.8633,
+      "step": 7771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2080707620019368e-05,
+      "loss": 0.9414,
+      "step": 7772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078872178397215e-05,
+      "loss": 0.7734,
+      "step": 7773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077036663574764e-05,
+      "loss": 0.8086,
+      "step": 7774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2075201075616642e-05,
+      "loss": 0.8906,
+      "step": 7775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207336541458749e-05,
+      "loss": 0.8633,
+      "step": 7776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071529680551937e-05,
+      "loss": 0.8828,
+      "step": 7777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069693873574628e-05,
+      "loss": 1.0078,
+      "step": 7778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20678579937202e-05,
+      "loss": 0.9258,
+      "step": 7779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066022041053303e-05,
+      "loss": 0.8633,
+      "step": 7780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064186015638577e-05,
+      "loss": 0.8672,
+      "step": 7781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062349917540675e-05,
+      "loss": 0.8281,
+      "step": 7782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2060513746824247e-05,
+      "loss": 0.9688,
+      "step": 7783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205867750355395e-05,
+      "loss": 0.9414,
+      "step": 7784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205684118779444e-05,
+      "loss": 0.8633,
+      "step": 7785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055004799610374e-05,
+      "loss": 0.8125,
+      "step": 7786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053168339066417e-05,
+      "loss": 0.8516,
+      "step": 7787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051331806227234e-05,
+      "loss": 0.7734,
+      "step": 7788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2049495201157489e-05,
+      "loss": 0.918,
+      "step": 7789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047658523921855e-05,
+      "loss": 0.918,
+      "step": 7790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045821774585003e-05,
+      "loss": 0.9414,
+      "step": 7791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2043984953211607e-05,
+      "loss": 0.8867,
+      "step": 7792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042148059866343e-05,
+      "loss": 0.9414,
+      "step": 7793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040311094613894e-05,
+      "loss": 0.8516,
+      "step": 7794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203847405751894e-05,
+      "loss": 0.9219,
+      "step": 7795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036636948646172e-05,
+      "loss": 1.0,
+      "step": 7796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2034799768060268e-05,
+      "loss": 0.9922,
+      "step": 7797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032962515825924e-05,
+      "loss": 0.8672,
+      "step": 7798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031125192007828e-05,
+      "loss": 0.8828,
+      "step": 7799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029287796670684e-05,
+      "loss": 0.8125,
+      "step": 7800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2027450329879178e-05,
+      "loss": 0.9844,
+      "step": 7801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025612791698018e-05,
+      "loss": 0.9297,
+      "step": 7802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023775182191901e-05,
+      "loss": 0.8125,
+      "step": 7803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202193750142554e-05,
+      "loss": 0.9492,
+      "step": 7804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020099749463635e-05,
+      "loss": 0.8008,
+      "step": 7805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018261926370898e-05,
+      "loss": 0.9219,
+      "step": 7806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016424032212043e-05,
+      "loss": 0.9414,
+      "step": 7807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2014586067051786e-05,
+      "loss": 0.9141,
+      "step": 7808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012748030954842e-05,
+      "loss": 0.8516,
+      "step": 7809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010909923985931e-05,
+      "loss": 1.0625,
+      "step": 7810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009071746209778e-05,
+      "loss": 0.8555,
+      "step": 7811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007233497691106e-05,
+      "loss": 0.9961,
+      "step": 7812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005395178494643e-05,
+      "loss": 0.8594,
+      "step": 7813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2003556788685123e-05,
+      "loss": 0.8828,
+      "step": 7814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001718328327272e-05,
+      "loss": 0.8672,
+      "step": 7815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999879797485829e-05,
+      "loss": 1.0391,
+      "step": 7816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199804119622553e-05,
+      "loss": 0.9414,
+      "step": 7817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1996202524611115e-05,
+      "loss": 1.0469,
+      "step": 7818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994363782707326e-05,
+      "loss": 0.9609,
+      "step": 7819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992524970578911e-05,
+      "loss": 0.8281,
+      "step": 7820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1990686088290615e-05,
+      "loss": 0.8281,
+      "step": 7821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988847135907188e-05,
+      "loss": 0.9609,
+      "step": 7822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987008113493383e-05,
+      "loss": 0.6797,
+      "step": 7823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985169021113953e-05,
+      "loss": 0.9141,
+      "step": 7824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1983329858833658e-05,
+      "loss": 0.9141,
+      "step": 7825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981490626717257e-05,
+      "loss": 0.8555,
+      "step": 7826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979651324829511e-05,
+      "loss": 0.8711,
+      "step": 7827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977811953235182e-05,
+      "loss": 0.7812,
+      "step": 7828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975972511999048e-05,
+      "loss": 0.9844,
+      "step": 7829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974133001185864e-05,
+      "loss": 0.9258,
+      "step": 7830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972293420860412e-05,
+      "loss": 0.9727,
+      "step": 7831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970453771087463e-05,
+      "loss": 0.875,
+      "step": 7832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968614051931796e-05,
+      "loss": 0.8398,
+      "step": 7833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966774263458183e-05,
+      "loss": 0.7695,
+      "step": 7834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1964934405731417e-05,
+      "loss": 0.7891,
+      "step": 7835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1963094478816272e-05,
+      "loss": 0.9375,
+      "step": 7836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961254482777544e-05,
+      "loss": 1.0234,
+      "step": 7837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959414417680016e-05,
+      "loss": 0.9453,
+      "step": 7838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957574283588482e-05,
+      "loss": 0.9453,
+      "step": 7839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1955734080567734e-05,
+      "loss": 0.7656,
+      "step": 7840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1953893808682568e-05,
+      "loss": 0.9219,
+      "step": 7841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952053467997786e-05,
+      "loss": 0.8047,
+      "step": 7842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950213058578188e-05,
+      "loss": 0.8828,
+      "step": 7843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948372580488574e-05,
+      "loss": 0.8125,
+      "step": 7844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946532033793756e-05,
+      "loss": 1.0234,
+      "step": 7845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944691418558539e-05,
+      "loss": 0.9297,
+      "step": 7846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942850734847735e-05,
+      "loss": 0.8867,
+      "step": 7847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941009982726154e-05,
+      "loss": 0.8789,
+      "step": 7848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193916916225862e-05,
+      "loss": 0.8086,
+      "step": 7849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193732827350994e-05,
+      "loss": 0.9609,
+      "step": 7850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935487316544943e-05,
+      "loss": 0.9102,
+      "step": 7851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193364629142845e-05,
+      "loss": 0.9609,
+      "step": 7852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1931805198225283e-05,
+      "loss": 0.8711,
+      "step": 7853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1929964037000275e-05,
+      "loss": 0.7617,
+      "step": 7854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1928122807818251e-05,
+      "loss": 0.8633,
+      "step": 7855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926281510744046e-05,
+      "loss": 1.0547,
+      "step": 7856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924440145842497e-05,
+      "loss": 0.9141,
+      "step": 7857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922598713178438e-05,
+      "loss": 1.0312,
+      "step": 7858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920757212816711e-05,
+      "loss": 0.8438,
+      "step": 7859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918915644822156e-05,
+      "loss": 0.7969,
+      "step": 7860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917074009259617e-05,
+      "loss": 0.9961,
+      "step": 7861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915232306193942e-05,
+      "loss": 1.0703,
+      "step": 7862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913390535689983e-05,
+      "loss": 0.8242,
+      "step": 7863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911548697812587e-05,
+      "loss": 0.9492,
+      "step": 7864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909706792626612e-05,
+      "loss": 0.8906,
+      "step": 7865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190786482019691e-05,
+      "loss": 0.918,
+      "step": 7866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1906022780588345e-05,
+      "loss": 0.9023,
+      "step": 7867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1904180673865771e-05,
+      "loss": 0.9258,
+      "step": 7868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902338500094058e-05,
+      "loss": 0.875,
+      "step": 7869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900496259338068e-05,
+      "loss": 0.8867,
+      "step": 7870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1898653951662672e-05,
+      "loss": 0.9297,
+      "step": 7871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896811577132734e-05,
+      "loss": 0.918,
+      "step": 7872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1894969135813136e-05,
+      "loss": 0.8672,
+      "step": 7873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893126627768749e-05,
+      "loss": 0.9609,
+      "step": 7874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891284053064451e-05,
+      "loss": 0.8086,
+      "step": 7875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1889441411765119e-05,
+      "loss": 0.9531,
+      "step": 7876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887598703935639e-05,
+      "loss": 0.8125,
+      "step": 7877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1885755929640897e-05,
+      "loss": 1.0078,
+      "step": 7878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1883913088945771e-05,
+      "loss": 0.875,
+      "step": 7879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1882070181915163e-05,
+      "loss": 0.9453,
+      "step": 7880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188022720861395e-05,
+      "loss": 0.8203,
+      "step": 7881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187838416910704e-05,
+      "loss": 0.7773,
+      "step": 7882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876541063459322e-05,
+      "loss": 0.8477,
+      "step": 7883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1874697891735694e-05,
+      "loss": 0.8164,
+      "step": 7884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187285465400106e-05,
+      "loss": 0.8867,
+      "step": 7885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1871011350320324e-05,
+      "loss": 0.875,
+      "step": 7886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869167980758385e-05,
+      "loss": 0.9062,
+      "step": 7887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867324545380159e-05,
+      "loss": 0.9609,
+      "step": 7888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865481044250551e-05,
+      "loss": 0.8672,
+      "step": 7889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863637477434475e-05,
+      "loss": 0.9531,
+      "step": 7890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861793844996847e-05,
+      "loss": 0.9062,
+      "step": 7891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859950147002583e-05,
+      "loss": 0.9648,
+      "step": 7892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18581063835166e-05,
+      "loss": 0.8516,
+      "step": 7893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856262554603826e-05,
+      "loss": 1.0312,
+      "step": 7894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1854418660329178e-05,
+      "loss": 0.8164,
+      "step": 7895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185257470075759e-05,
+      "loss": 0.8516,
+      "step": 7896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1850730675953983e-05,
+      "loss": 0.8281,
+      "step": 7897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1848886585983294e-05,
+      "loss": 0.8594,
+      "step": 7898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847042430910451e-05,
+      "loss": 0.8945,
+      "step": 7899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845198210800394e-05,
+      "loss": 0.9336,
+      "step": 7900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843353925718058e-05,
+      "loss": 0.9688,
+      "step": 7901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841509575728385e-05,
+      "loss": 0.9258,
+      "step": 7902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1839665160896314e-05,
+      "loss": 0.8438,
+      "step": 7903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1837820681286796e-05,
+      "loss": 1.0078,
+      "step": 7904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835976136964771e-05,
+      "loss": 0.7344,
+      "step": 7905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834131527995195e-05,
+      "loss": 0.7969,
+      "step": 7906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1832286854443012e-05,
+      "loss": 0.8555,
+      "step": 7907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830442116373183e-05,
+      "loss": 1.0156,
+      "step": 7908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1828597313850658e-05,
+      "loss": 0.8789,
+      "step": 7909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826752446940401e-05,
+      "loss": 0.8477,
+      "step": 7910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824907515707369e-05,
+      "loss": 0.8516,
+      "step": 7911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823062520216526e-05,
+      "loss": 0.9336,
+      "step": 7912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821217460532836e-05,
+      "loss": 0.7773,
+      "step": 7913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181937233672127e-05,
+      "loss": 0.8516,
+      "step": 7914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181752714884679e-05,
+      "loss": 0.9062,
+      "step": 7915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815681896974376e-05,
+      "loss": 1.0547,
+      "step": 7916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813836581169e-05,
+      "loss": 0.8242,
+      "step": 7917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811991201495633e-05,
+      "loss": 0.8359,
+      "step": 7918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810145758019263e-05,
+      "loss": 0.8242,
+      "step": 7919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808300250804863e-05,
+      "loss": 0.957,
+      "step": 7920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806454679917421e-05,
+      "loss": 0.8906,
+      "step": 7921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1804609045421918e-05,
+      "loss": 0.9062,
+      "step": 7922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1802763347383348e-05,
+      "loss": 0.9531,
+      "step": 7923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800917585866693e-05,
+      "loss": 0.8867,
+      "step": 7924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799071760936953e-05,
+      "loss": 0.9961,
+      "step": 7925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797225872659115e-05,
+      "loss": 0.9258,
+      "step": 7926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795379921098182e-05,
+      "loss": 0.9492,
+      "step": 7927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1793533906319148e-05,
+      "loss": 0.9883,
+      "step": 7928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791687828387017e-05,
+      "loss": 0.8008,
+      "step": 7929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178984168736679e-05,
+      "loss": 0.9062,
+      "step": 7930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1787995483323475e-05,
+      "loss": 0.7812,
+      "step": 7931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786149216322079e-05,
+      "loss": 0.8438,
+      "step": 7932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178430288642761e-05,
+      "loss": 1.0,
+      "step": 7933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1782456493705082e-05,
+      "loss": 0.8516,
+      "step": 7934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1780610038219507e-05,
+      "loss": 0.793,
+      "step": 7935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778763520035905e-05,
+      "loss": 0.8516,
+      "step": 7936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1776916939219291e-05,
+      "loss": 0.8672,
+      "step": 7937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775070295834691e-05,
+      "loss": 0.8711,
+      "step": 7938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773223589947122e-05,
+      "loss": 0.8555,
+      "step": 7939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1771376821621616e-05,
+      "loss": 1.0938,
+      "step": 7940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769529990923197e-05,
+      "loss": 0.8516,
+      "step": 7941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767683097916893e-05,
+      "loss": 0.7695,
+      "step": 7942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765836142667737e-05,
+      "loss": 0.9609,
+      "step": 7943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763989125240766e-05,
+      "loss": 0.7578,
+      "step": 7944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762142045701014e-05,
+      "loss": 0.8789,
+      "step": 7945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760294904113519e-05,
+      "loss": 0.8945,
+      "step": 7946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758447700543325e-05,
+      "loss": 0.8516,
+      "step": 7947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1756600435055468e-05,
+      "loss": 0.9336,
+      "step": 7948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754753107715e-05,
+      "loss": 0.8516,
+      "step": 7949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752905718586965e-05,
+      "loss": 0.8906,
+      "step": 7950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1751058267736415e-05,
+      "loss": 0.875,
+      "step": 7951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749210755228398e-05,
+      "loss": 0.957,
+      "step": 7952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747363181127971e-05,
+      "loss": 0.9258,
+      "step": 7953
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745515545500186e-05,
+      "loss": 1.0,
+      "step": 7954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743667848410105e-05,
+      "loss": 0.9375,
+      "step": 7955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741820089922786e-05,
+      "loss": 0.8672,
+      "step": 7956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1739972270103295e-05,
+      "loss": 0.9141,
+      "step": 7957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738124389016692e-05,
+      "loss": 0.8867,
+      "step": 7958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1736276446728045e-05,
+      "loss": 0.8672,
+      "step": 7959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1734428443302424e-05,
+      "loss": 0.8359,
+      "step": 7960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732580378804901e-05,
+      "loss": 0.8945,
+      "step": 7961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730732253300544e-05,
+      "loss": 0.9531,
+      "step": 7962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1728884066854435e-05,
+      "loss": 0.9844,
+      "step": 7963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172703581953165e-05,
+      "loss": 0.8281,
+      "step": 7964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725187511397265e-05,
+      "loss": 0.8633,
+      "step": 7965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1723339142516363e-05,
+      "loss": 0.9141,
+      "step": 7966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172149071295403e-05,
+      "loss": 0.8086,
+      "step": 7967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171964222277535e-05,
+      "loss": 0.8164,
+      "step": 7968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717793672045416e-05,
+      "loss": 0.9141,
+      "step": 7969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715945060829312e-05,
+      "loss": 0.9844,
+      "step": 7970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714096389192133e-05,
+      "loss": 0.9375,
+      "step": 7971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1712247657198974e-05,
+      "loss": 0.8359,
+      "step": 7972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710398864914931e-05,
+      "loss": 0.7734,
+      "step": 7973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1708550012405104e-05,
+      "loss": 0.8203,
+      "step": 7974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706701099734593e-05,
+      "loss": 0.9023,
+      "step": 7975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17048521269685e-05,
+      "loss": 0.7891,
+      "step": 7976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1703003094171932e-05,
+      "loss": 0.7656,
+      "step": 7977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701154001409993e-05,
+      "loss": 0.8477,
+      "step": 7978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699304848747799e-05,
+      "loss": 0.875,
+      "step": 7979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697455636250456e-05,
+      "loss": 0.8867,
+      "step": 7980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695606363983081e-05,
+      "loss": 0.875,
+      "step": 7981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1693757032010783e-05,
+      "loss": 0.8086,
+      "step": 7982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169190764039869e-05,
+      "loss": 0.8867,
+      "step": 7983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1690058189211912e-05,
+      "loss": 0.7969,
+      "step": 7984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688208678515581e-05,
+      "loss": 0.9062,
+      "step": 7985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1686359108374811e-05,
+      "loss": 0.9453,
+      "step": 7986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1684509478854735e-05,
+      "loss": 0.9023,
+      "step": 7987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682659790020478e-05,
+      "loss": 0.9609,
+      "step": 7988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680810041937176e-05,
+      "loss": 0.8984,
+      "step": 7989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1678960234669953e-05,
+      "loss": 0.957,
+      "step": 7990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167711036828395e-05,
+      "loss": 0.8906,
+      "step": 7991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675260442844302e-05,
+      "loss": 0.7891,
+      "step": 7992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673410458416145e-05,
+      "loss": 1.0156,
+      "step": 7993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671560415064624e-05,
+      "loss": 0.9531,
+      "step": 7994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669710312854879e-05,
+      "loss": 0.9219,
+      "step": 7995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1667860151852058e-05,
+      "loss": 0.918,
+      "step": 7996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666009932121301e-05,
+      "loss": 0.9297,
+      "step": 7997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664159653727767e-05,
+      "loss": 0.8516,
+      "step": 7998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662309316736598e-05,
+      "loss": 0.8789,
+      "step": 7999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660458921212953e-05,
+      "loss": 1.0469,
+      "step": 8000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1658608467221985e-05,
+      "loss": 0.9414,
+      "step": 8001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1656757954828854e-05,
+      "loss": 0.957,
+      "step": 8002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1654907384098715e-05,
+      "loss": 0.7773,
+      "step": 8003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1653056755096735e-05,
+      "loss": 0.8398,
+      "step": 8004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165120606788807e-05,
+      "loss": 0.8828,
+      "step": 8005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649355322537894e-05,
+      "loss": 0.8438,
+      "step": 8006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1647504519111368e-05,
+      "loss": 0.8711,
+      "step": 8007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1645653657673666e-05,
+      "loss": 0.8477,
+      "step": 8008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1643802738289955e-05,
+      "loss": 1.1328,
+      "step": 8009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641951761025414e-05,
+      "loss": 0.9727,
+      "step": 8010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640100725945216e-05,
+      "loss": 0.9062,
+      "step": 8011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1638249633114538e-05,
+      "loss": 0.8398,
+      "step": 8012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163639848259856e-05,
+      "loss": 0.7969,
+      "step": 8013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634547274462468e-05,
+      "loss": 0.8906,
+      "step": 8014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632696008771443e-05,
+      "loss": 0.8633,
+      "step": 8015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630844685590667e-05,
+      "loss": 0.8945,
+      "step": 8016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628993304985335e-05,
+      "loss": 0.9531,
+      "step": 8017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1627141867020632e-05,
+      "loss": 0.8633,
+      "step": 8018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162529037176175e-05,
+      "loss": 0.9531,
+      "step": 8019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623438819273887e-05,
+      "loss": 0.957,
+      "step": 8020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621587209622237e-05,
+      "loss": 0.8906,
+      "step": 8021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1619735542871995e-05,
+      "loss": 0.8398,
+      "step": 8022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1617883819088364e-05,
+      "loss": 0.8242,
+      "step": 8023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616032038336548e-05,
+      "loss": 0.8906,
+      "step": 8024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614180200681746e-05,
+      "loss": 1.0703,
+      "step": 8025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1612328306189172e-05,
+      "loss": 1.0156,
+      "step": 8026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1610476354924023e-05,
+      "loss": 0.9492,
+      "step": 8027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608624346951519e-05,
+      "loss": 0.9023,
+      "step": 8028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606772282336865e-05,
+      "loss": 0.8672,
+      "step": 8029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1604920161145282e-05,
+      "loss": 0.875,
+      "step": 8030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603067983441978e-05,
+      "loss": 0.957,
+      "step": 8031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601215749292177e-05,
+      "loss": 0.9102,
+      "step": 8032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1599363458761094e-05,
+      "loss": 0.8438,
+      "step": 8033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159751111191396e-05,
+      "loss": 0.8555,
+      "step": 8034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159565870881599e-05,
+      "loss": 0.9609,
+      "step": 8035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593806249532414e-05,
+      "loss": 0.8984,
+      "step": 8036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1591953734128458e-05,
+      "loss": 0.8594,
+      "step": 8037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1590101162669354e-05,
+      "loss": 0.7617,
+      "step": 8038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588248535220333e-05,
+      "loss": 0.8594,
+      "step": 8039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586395851846628e-05,
+      "loss": 0.8867,
+      "step": 8040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584543112613476e-05,
+      "loss": 0.9648,
+      "step": 8041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582690317586116e-05,
+      "loss": 0.9375,
+      "step": 8042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580837466829782e-05,
+      "loss": 0.8711,
+      "step": 8043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1578984560409725e-05,
+      "loss": 0.8555,
+      "step": 8044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157713159839118e-05,
+      "loss": 0.9297,
+      "step": 8045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15752785808394e-05,
+      "loss": 1.0312,
+      "step": 8046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573425507819627e-05,
+      "loss": 1.0547,
+      "step": 8047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1571572379397112e-05,
+      "loss": 0.8906,
+      "step": 8048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569719195637108e-05,
+      "loss": 0.9258,
+      "step": 8049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567865956604867e-05,
+      "loss": 0.8516,
+      "step": 8050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566012662365645e-05,
+      "loss": 0.9453,
+      "step": 8051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1564159312984698e-05,
+      "loss": 0.9492,
+      "step": 8052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562305908527287e-05,
+      "loss": 0.918,
+      "step": 8053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560452449058672e-05,
+      "loss": 0.9219,
+      "step": 8054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155859893464412e-05,
+      "loss": 0.9883,
+      "step": 8055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556745365348892e-05,
+      "loss": 0.9688,
+      "step": 8056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554891741238255e-05,
+      "loss": 0.8672,
+      "step": 8057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553038062377478e-05,
+      "loss": 0.9141,
+      "step": 8058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1551184328831834e-05,
+      "loss": 1.0234,
+      "step": 8059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549330540666595e-05,
+      "loss": 0.8477,
+      "step": 8060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1547476697947038e-05,
+      "loss": 0.8828,
+      "step": 8061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545622800738431e-05,
+      "loss": 0.8984,
+      "step": 8062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543768849106066e-05,
+      "loss": 0.832,
+      "step": 8063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541914843115212e-05,
+      "loss": 0.9023,
+      "step": 8064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1540060782831157e-05,
+      "loss": 1.0156,
+      "step": 8065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538206668319182e-05,
+      "loss": 0.875,
+      "step": 8066
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1536352499644578e-05,
+      "loss": 0.8828,
+      "step": 8067
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534498276872628e-05,
+      "loss": 0.9062,
+      "step": 8068
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532644000068627e-05,
+      "loss": 0.8477,
+      "step": 8069
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530789669297862e-05,
+      "loss": 0.9219,
+      "step": 8070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1528935284625628e-05,
+      "loss": 0.9062,
+      "step": 8071
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527080846117224e-05,
+      "loss": 0.9062,
+      "step": 8072
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1525226353837948e-05,
+      "loss": 0.7188,
+      "step": 8073
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523371807853092e-05,
+      "loss": 0.9414,
+      "step": 8074
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521517208227967e-05,
+      "loss": 0.7969,
+      "step": 8075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1519662555027868e-05,
+      "loss": 0.9531,
+      "step": 8076
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1517807848318107e-05,
+      "loss": 1.0312,
+      "step": 8077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1515953088163988e-05,
+      "loss": 0.9688,
+      "step": 8078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514098274630821e-05,
+      "loss": 0.9805,
+      "step": 8079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1512243407783911e-05,
+      "loss": 1.0781,
+      "step": 8080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510388487688582e-05,
+      "loss": 0.8867,
+      "step": 8081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508533514410138e-05,
+      "loss": 0.8359,
+      "step": 8082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506678488013903e-05,
+      "loss": 0.9141,
+      "step": 8083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504823408565188e-05,
+      "loss": 0.9062,
+      "step": 8084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502968276129323e-05,
+      "loss": 0.9062,
+      "step": 8085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501113090771618e-05,
+      "loss": 0.7891,
+      "step": 8086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1499257852557409e-05,
+      "loss": 0.8008,
+      "step": 8087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497402561552012e-05,
+      "loss": 0.9609,
+      "step": 8088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1495547217820762e-05,
+      "loss": 0.9062,
+      "step": 8089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493691821428985e-05,
+      "loss": 1.0156,
+      "step": 8090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149183637244201e-05,
+      "loss": 0.8594,
+      "step": 8091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1489980870925177e-05,
+      "loss": 0.8633,
+      "step": 8092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488125316943812e-05,
+      "loss": 0.8945,
+      "step": 8093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486269710563262e-05,
+      "loss": 1.0625,
+      "step": 8094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1484414051848855e-05,
+      "loss": 1.0,
+      "step": 8095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1482558340865944e-05,
+      "loss": 0.8555,
+      "step": 8096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1480702577679859e-05,
+      "loss": 0.832,
+      "step": 8097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478846762355953e-05,
+      "loss": 0.8008,
+      "step": 8098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476990894959567e-05,
+      "loss": 0.8086,
+      "step": 8099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1475134975556054e-05,
+      "loss": 0.875,
+      "step": 8100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473279004210758e-05,
+      "loss": 0.9961,
+      "step": 8101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471422980989038e-05,
+      "loss": 0.8477,
+      "step": 8102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1469566905956238e-05,
+      "loss": 0.8594,
+      "step": 8103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467710779177722e-05,
+      "loss": 0.8477,
+      "step": 8104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465854600718842e-05,
+      "loss": 0.9297,
+      "step": 8105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1463998370644957e-05,
+      "loss": 0.8672,
+      "step": 8106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1462142089021431e-05,
+      "loss": 0.8828,
+      "step": 8107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146028575591362e-05,
+      "loss": 0.8945,
+      "step": 8108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14584293713869e-05,
+      "loss": 0.8711,
+      "step": 8109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456572935506624e-05,
+      "loss": 0.9219,
+      "step": 8110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454716448338167e-05,
+      "loss": 0.875,
+      "step": 8111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14528599099469e-05,
+      "loss": 0.9844,
+      "step": 8112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1451003320398192e-05,
+      "loss": 0.9609,
+      "step": 8113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449146679757414e-05,
+      "loss": 0.7969,
+      "step": 8114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447289988089948e-05,
+      "loss": 0.8633,
+      "step": 8115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445433245461164e-05,
+      "loss": 0.8125,
+      "step": 8116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1443576451936444e-05,
+      "loss": 0.9492,
+      "step": 8117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441719607581173e-05,
+      "loss": 0.8125,
+      "step": 8118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439862712460721e-05,
+      "loss": 0.8086,
+      "step": 8119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438005766640486e-05,
+      "loss": 0.8359,
+      "step": 8120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436148770185843e-05,
+      "loss": 1.0391,
+      "step": 8121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434291723162187e-05,
+      "loss": 0.875,
+      "step": 8122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432434625634903e-05,
+      "loss": 0.9727,
+      "step": 8123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1430577477669386e-05,
+      "loss": 0.7969,
+      "step": 8124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428720279331025e-05,
+      "loss": 0.8594,
+      "step": 8125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142686303068522e-05,
+      "loss": 0.9375,
+      "step": 8126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425005731797363e-05,
+      "loss": 0.8789,
+      "step": 8127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.8398,
+      "step": 8128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421290983557092e-05,
+      "loss": 0.793,
+      "step": 8129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1419433534335483e-05,
+      "loss": 0.875,
+      "step": 8130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417576035133424e-05,
+      "loss": 0.9062,
+      "step": 8131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1415718486016327e-05,
+      "loss": 0.8555,
+      "step": 8132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413860887049594e-05,
+      "loss": 0.75,
+      "step": 8133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412003238298639e-05,
+      "loss": 0.9766,
+      "step": 8134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410145539828867e-05,
+      "loss": 0.8203,
+      "step": 8135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408287791705694e-05,
+      "loss": 0.8594,
+      "step": 8136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1406429993994535e-05,
+      "loss": 0.918,
+      "step": 8137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404572146760804e-05,
+      "loss": 0.8789,
+      "step": 8138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402714250069917e-05,
+      "loss": 0.8086,
+      "step": 8139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400856303987298e-05,
+      "loss": 0.9688,
+      "step": 8140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1398998308578363e-05,
+      "loss": 0.8203,
+      "step": 8141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397140263908542e-05,
+      "loss": 0.8906,
+      "step": 8142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.139528217004325e-05,
+      "loss": 0.9531,
+      "step": 8143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393424027047921e-05,
+      "loss": 0.7969,
+      "step": 8144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391565834987978e-05,
+      "loss": 0.8906,
+      "step": 8145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389707593928858e-05,
+      "loss": 1.0312,
+      "step": 8146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1387849303935984e-05,
+      "loss": 0.8867,
+      "step": 8147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1385990965074796e-05,
+      "loss": 0.9219,
+      "step": 8148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384132577410724e-05,
+      "loss": 0.8594,
+      "step": 8149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382274141009208e-05,
+      "loss": 0.9961,
+      "step": 8150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1380415655935683e-05,
+      "loss": 0.7734,
+      "step": 8151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378557122255596e-05,
+      "loss": 0.8164,
+      "step": 8152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137669854003438e-05,
+      "loss": 0.8359,
+      "step": 8153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374839909337486e-05,
+      "loss": 0.8906,
+      "step": 8154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372981230230354e-05,
+      "loss": 0.9062,
+      "step": 8155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371122502778433e-05,
+      "loss": 0.9375,
+      "step": 8156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1369263727047171e-05,
+      "loss": 0.9062,
+      "step": 8157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367404903102022e-05,
+      "loss": 0.9531,
+      "step": 8158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365546031008434e-05,
+      "loss": 0.7344,
+      "step": 8159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1363687110831861e-05,
+      "loss": 0.8164,
+      "step": 8160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361828142637758e-05,
+      "loss": 0.8594,
+      "step": 8161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359969126491585e-05,
+      "loss": 0.8086,
+      "step": 8162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13581100624588e-05,
+      "loss": 0.8438,
+      "step": 8163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356250950604861e-05,
+      "loss": 0.8945,
+      "step": 8164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354391790995234e-05,
+      "loss": 0.9062,
+      "step": 8165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352532583695379e-05,
+      "loss": 0.7891,
+      "step": 8166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350673328770766e-05,
+      "loss": 0.9219,
+      "step": 8167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1348814026286855e-05,
+      "loss": 0.8438,
+      "step": 8168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346954676309122e-05,
+      "loss": 0.8477,
+      "step": 8169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1345095278903035e-05,
+      "loss": 0.7812,
+      "step": 8170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343235834134068e-05,
+      "loss": 0.8516,
+      "step": 8171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1341376342067692e-05,
+      "loss": 0.8594,
+      "step": 8172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1339516802769386e-05,
+      "loss": 0.9766,
+      "step": 8173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133765721630462e-05,
+      "loss": 0.9609,
+      "step": 8174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335797582738884e-05,
+      "loss": 0.8164,
+      "step": 8175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333937902137648e-05,
+      "loss": 0.8984,
+      "step": 8176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1332078174566403e-05,
+      "loss": 0.8984,
+      "step": 8177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330218400090625e-05,
+      "loss": 0.832,
+      "step": 8178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328358578775807e-05,
+      "loss": 0.8047,
+      "step": 8179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1326498710687428e-05,
+      "loss": 0.9062,
+      "step": 8180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324638795890986e-05,
+      "loss": 0.8281,
+      "step": 8181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1322778834451961e-05,
+      "loss": 0.9453,
+      "step": 8182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320918826435857e-05,
+      "loss": 0.8477,
+      "step": 8183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319058771908158e-05,
+      "loss": 0.8984,
+      "step": 8184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1317198670934364e-05,
+      "loss": 1.0156,
+      "step": 8185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315338523579971e-05,
+      "loss": 0.7734,
+      "step": 8186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313478329910479e-05,
+      "loss": 0.8438,
+      "step": 8187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311618089991385e-05,
+      "loss": 0.9609,
+      "step": 8188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309757803888193e-05,
+      "loss": 0.8672,
+      "step": 8189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1307897471666406e-05,
+      "loss": 0.8516,
+      "step": 8190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306037093391533e-05,
+      "loss": 0.8047,
+      "step": 8191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1304176669129074e-05,
+      "loss": 0.9805,
+      "step": 8192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1302316198944543e-05,
+      "loss": 0.8516,
+      "step": 8193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300455682903446e-05,
+      "loss": 0.8555,
+      "step": 8194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298595121071298e-05,
+      "loss": 0.9883,
+      "step": 8195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129673451351361e-05,
+      "loss": 0.8906,
+      "step": 8196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1294873860295898e-05,
+      "loss": 1.0469,
+      "step": 8197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293013161483676e-05,
+      "loss": 0.9023,
+      "step": 8198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1291152417142468e-05,
+      "loss": 0.8984,
+      "step": 8199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289291627337786e-05,
+      "loss": 0.8125,
+      "step": 8200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287430792135158e-05,
+      "loss": 0.7695,
+      "step": 8201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285569911600103e-05,
+      "loss": 0.7734,
+      "step": 8202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1283708985798147e-05,
+      "loss": 1.0156,
+      "step": 8203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281848014794813e-05,
+      "loss": 0.8711,
+      "step": 8204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279986998655637e-05,
+      "loss": 0.9141,
+      "step": 8205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1278125937446137e-05,
+      "loss": 0.8047,
+      "step": 8206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276264831231851e-05,
+      "loss": 0.9375,
+      "step": 8207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127440368007831e-05,
+      "loss": 1.0156,
+      "step": 8208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1272542484051048e-05,
+      "loss": 0.9102,
+      "step": 8209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1270681243215597e-05,
+      "loss": 0.9219,
+      "step": 8210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268819957637503e-05,
+      "loss": 0.8867,
+      "step": 8211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266958627382295e-05,
+      "loss": 0.8125,
+      "step": 8212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265097252515523e-05,
+      "loss": 0.875,
+      "step": 8213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263235833102716e-05,
+      "loss": 0.7695,
+      "step": 8214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126137436920943e-05,
+      "loss": 0.8125,
+      "step": 8215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1259512860901205e-05,
+      "loss": 0.9531,
+      "step": 8216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257651308243585e-05,
+      "loss": 0.793,
+      "step": 8217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1255789711302123e-05,
+      "loss": 1.0312,
+      "step": 8218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1253928070142364e-05,
+      "loss": 0.832,
+      "step": 8219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252066384829862e-05,
+      "loss": 0.8672,
+      "step": 8220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125020465543017e-05,
+      "loss": 0.9766,
+      "step": 8221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248342882008842e-05,
+      "loss": 0.8867,
+      "step": 8222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1246481064631434e-05,
+      "loss": 0.7617,
+      "step": 8223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244619203363503e-05,
+      "loss": 0.9297,
+      "step": 8224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242757298270605e-05,
+      "loss": 0.8672,
+      "step": 8225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1240895349418308e-05,
+      "loss": 0.8789,
+      "step": 8226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123903335687217e-05,
+      "loss": 0.7734,
+      "step": 8227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237171320697753e-05,
+      "loss": 0.9688,
+      "step": 8228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1235309240960621e-05,
+      "loss": 0.8164,
+      "step": 8229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1233447117726346e-05,
+      "loss": 0.9805,
+      "step": 8230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1231584951060493e-05,
+      "loss": 0.8398,
+      "step": 8231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229722741028633e-05,
+      "loss": 0.8672,
+      "step": 8232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1227860487696336e-05,
+      "loss": 0.8711,
+      "step": 8233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225998191129175e-05,
+      "loss": 0.8477,
+      "step": 8234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224135851392726e-05,
+      "loss": 1.0078,
+      "step": 8235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222273468552562e-05,
+      "loss": 0.9297,
+      "step": 8236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220411042674262e-05,
+      "loss": 0.8672,
+      "step": 8237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1218548573823406e-05,
+      "loss": 0.832,
+      "step": 8238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1216686062065574e-05,
+      "loss": 0.9258,
+      "step": 8239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1214823507466345e-05,
+      "loss": 0.8359,
+      "step": 8240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212960910091306e-05,
+      "loss": 0.918,
+      "step": 8241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211098270006043e-05,
+      "loss": 0.7383,
+      "step": 8242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209235587276135e-05,
+      "loss": 0.9375,
+      "step": 8243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207372861967179e-05,
+      "loss": 0.8867,
+      "step": 8244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205510094144756e-05,
+      "loss": 0.9922,
+      "step": 8245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203647283874465e-05,
+      "loss": 0.8359,
+      "step": 8246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201784431221893e-05,
+      "loss": 0.832,
+      "step": 8247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199921536252639e-05,
+      "loss": 0.9219,
+      "step": 8248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198058599032289e-05,
+      "loss": 0.9961,
+      "step": 8249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119619561962645e-05,
+      "loss": 0.8516,
+      "step": 8250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1194332598100714e-05,
+      "loss": 0.7344,
+      "step": 8251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192469534520687e-05,
+      "loss": 0.8906,
+      "step": 8252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1190606428951963e-05,
+      "loss": 1.0859,
+      "step": 8253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118874328146015e-05,
+      "loss": 0.8828,
+      "step": 8254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118688009211085e-05,
+      "loss": 0.7734,
+      "step": 8255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118501686096967e-05,
+      "loss": 1.0938,
+      "step": 8256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183153588102215e-05,
+      "loss": 0.8477,
+      "step": 8257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1181290273574098e-05,
+      "loss": 0.8867,
+      "step": 8258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179426917450924e-05,
+      "loss": 0.8711,
+      "step": 8259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177563519798309e-05,
+      "loss": 0.8203,
+      "step": 8260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175700080681864e-05,
+      "loss": 0.8867,
+      "step": 8261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173836600167202e-05,
+      "loss": 0.9102,
+      "step": 8262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171973078319946e-05,
+      "loss": 0.832,
+      "step": 8263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170109515205703e-05,
+      "loss": 0.7383,
+      "step": 8264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1168245910890101e-05,
+      "loss": 0.8477,
+      "step": 8265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1166382265438753e-05,
+      "loss": 0.9766,
+      "step": 8266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1164518578917287e-05,
+      "loss": 0.8359,
+      "step": 8267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162654851391323e-05,
+      "loss": 0.7578,
+      "step": 8268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160791082926488e-05,
+      "loss": 0.9453,
+      "step": 8269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158927273588404e-05,
+      "loss": 0.9062,
+      "step": 8270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157063423442705e-05,
+      "loss": 0.8516,
+      "step": 8271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155199532555011e-05,
+      "loss": 0.9258,
+      "step": 8272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153335600990965e-05,
+      "loss": 1.0,
+      "step": 8273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151471628816186e-05,
+      "loss": 0.8438,
+      "step": 8274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149607616096316e-05,
+      "loss": 0.8516,
+      "step": 8275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147743562896984e-05,
+      "loss": 0.9297,
+      "step": 8276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114587946928383e-05,
+      "loss": 0.8555,
+      "step": 8277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1144015335322495e-05,
+      "loss": 0.793,
+      "step": 8278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114215116107861e-05,
+      "loss": 0.9453,
+      "step": 8279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114028694661782e-05,
+      "loss": 0.8789,
+      "step": 8280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138422692005764e-05,
+      "loss": 0.8438,
+      "step": 8281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136558397308088e-05,
+      "loss": 0.9297,
+      "step": 8282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1134694062590436e-05,
+      "loss": 0.8984,
+      "step": 8283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132829687918456e-05,
+      "loss": 0.8438,
+      "step": 8284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130965273357793e-05,
+      "loss": 0.9414,
+      "step": 8285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129100818974095e-05,
+      "loss": 0.8594,
+      "step": 8286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127236324833013e-05,
+      "loss": 0.8438,
+      "step": 8287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1125371791000204e-05,
+      "loss": 0.8008,
+      "step": 8288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123507217541312e-05,
+      "loss": 0.9219,
+      "step": 8289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121642604521999e-05,
+      "loss": 0.9219,
+      "step": 8290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119777952007918e-05,
+      "loss": 0.9688,
+      "step": 8291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1117913260064724e-05,
+      "loss": 0.9219,
+      "step": 8292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111604852875808e-05,
+      "loss": 0.8984,
+      "step": 8293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114183758153642e-05,
+      "loss": 0.9453,
+      "step": 8294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1112318948317077e-05,
+      "loss": 0.8828,
+      "step": 8295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110454099314043e-05,
+      "loss": 1.0625,
+      "step": 8296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108589211210205e-05,
+      "loss": 0.9492,
+      "step": 8297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1106724284071229e-05,
+      "loss": 0.9453,
+      "step": 8298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104859317962784e-05,
+      "loss": 0.8867,
+      "step": 8299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102994312950534e-05,
+      "loss": 0.7148,
+      "step": 8300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1101129269100152e-05,
+      "loss": 0.8594,
+      "step": 8301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1099264186477309e-05,
+      "loss": 0.8359,
+      "step": 8302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097399065147678e-05,
+      "loss": 0.7617,
+      "step": 8303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109553390517693e-05,
+      "loss": 0.8086,
+      "step": 8304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1093668706630742e-05,
+      "loss": 0.9297,
+      "step": 8305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091803469574789e-05,
+      "loss": 0.8867,
+      "step": 8306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089938194074751e-05,
+      "loss": 0.8164,
+      "step": 8307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088072880196306e-05,
+      "loss": 0.9648,
+      "step": 8308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086207528005135e-05,
+      "loss": 0.8984,
+      "step": 8309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1084342137566919e-05,
+      "loss": 0.9688,
+      "step": 8310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082476708947345e-05,
+      "loss": 0.9219,
+      "step": 8311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1080611242212092e-05,
+      "loss": 0.9375,
+      "step": 8312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1078745737426851e-05,
+      "loss": 0.9023,
+      "step": 8313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076880194657305e-05,
+      "loss": 0.8789,
+      "step": 8314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1075014613969147e-05,
+      "loss": 0.9453,
+      "step": 8315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107314899542806e-05,
+      "loss": 0.957,
+      "step": 8316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071283339099742e-05,
+      "loss": 0.793,
+      "step": 8317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1069417645049885e-05,
+      "loss": 0.9141,
+      "step": 8318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1067551913344183e-05,
+      "loss": 0.9688,
+      "step": 8319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065686144048327e-05,
+      "loss": 0.8281,
+      "step": 8320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1063820337228018e-05,
+      "loss": 0.9102,
+      "step": 8321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106195449294895e-05,
+      "loss": 0.8086,
+      "step": 8322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060088611276827e-05,
+      "loss": 0.8945,
+      "step": 8323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058222692277345e-05,
+      "loss": 0.8086,
+      "step": 8324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105635673601621e-05,
+      "loss": 0.9062,
+      "step": 8325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1054490742559123e-05,
+      "loss": 0.8516,
+      "step": 8326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1052624711971791e-05,
+      "loss": 0.7617,
+      "step": 8327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050758644319917e-05,
+      "loss": 0.8594,
+      "step": 8328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048892539669206e-05,
+      "loss": 0.793,
+      "step": 8329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104702639808537e-05,
+      "loss": 1.0156,
+      "step": 8330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045160219634116e-05,
+      "loss": 0.8789,
+      "step": 8331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043294004381163e-05,
+      "loss": 0.9102,
+      "step": 8332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104142775239221e-05,
+      "loss": 0.8984,
+      "step": 8333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039561463732983e-05,
+      "loss": 0.7422,
+      "step": 8334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103769513846919e-05,
+      "loss": 0.9023,
+      "step": 8335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035828776666551e-05,
+      "loss": 0.957,
+      "step": 8336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033962378390777e-05,
+      "loss": 0.9297,
+      "step": 8337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1032095943707594e-05,
+      "loss": 0.8672,
+      "step": 8338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030229472682719e-05,
+      "loss": 1.0156,
+      "step": 8339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028362965381875e-05,
+      "loss": 0.8008,
+      "step": 8340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026496421870781e-05,
+      "loss": 0.9531,
+      "step": 8341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024629842215165e-05,
+      "loss": 0.8281,
+      "step": 8342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022763226480748e-05,
+      "loss": 0.793,
+      "step": 8343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020896574733261e-05,
+      "loss": 0.8828,
+      "step": 8344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1019029887038429e-05,
+      "loss": 0.875,
+      "step": 8345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017163163461979e-05,
+      "loss": 0.9531,
+      "step": 8346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015296404069644e-05,
+      "loss": 1.0234,
+      "step": 8347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013429608927157e-05,
+      "loss": 0.8945,
+      "step": 8348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011562778100246e-05,
+      "loss": 0.8633,
+      "step": 8349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009695911654651e-05,
+      "loss": 0.8086,
+      "step": 8350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007829009656102e-05,
+      "loss": 0.8984,
+      "step": 8351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100596207217034e-05,
+      "loss": 0.8555,
+      "step": 8352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1004095099263096e-05,
+      "loss": 0.8906,
+      "step": 8353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002228091000118e-05,
+      "loss": 0.8008,
+      "step": 8354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000361047447137e-05,
+      "loss": 0.8164,
+      "step": 8355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0998493968669901e-05,
+      "loss": 0.8828,
+      "step": 8356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0996626854734149e-05,
+      "loss": 0.8047,
+      "step": 8357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0994759705705632e-05,
+      "loss": 0.8906,
+      "step": 8358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0992892521650083e-05,
+      "loss": 0.8711,
+      "step": 8359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099102530263326e-05,
+      "loss": 0.8594,
+      "step": 8360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0989158048720903e-05,
+      "loss": 0.8164,
+      "step": 8361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987290759978767e-05,
+      "loss": 0.8086,
+      "step": 8362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0985423436472596e-05,
+      "loss": 0.8359,
+      "step": 8363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983556078268146e-05,
+      "loss": 1.0938,
+      "step": 8364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0981688685431166e-05,
+      "loss": 0.8125,
+      "step": 8365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979821258027415e-05,
+      "loss": 0.8398,
+      "step": 8366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097795379612264e-05,
+      "loss": 0.8633,
+      "step": 8367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0976086299782607e-05,
+      "loss": 0.8203,
+      "step": 8368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0974218769073065e-05,
+      "loss": 0.8906,
+      "step": 8369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972351204059777e-05,
+      "loss": 0.9883,
+      "step": 8370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.09704836048085e-05,
+      "loss": 0.7109,
+      "step": 8371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968615971385e-05,
+      "loss": 0.9453,
+      "step": 8372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966748303855032e-05,
+      "loss": 0.8281,
+      "step": 8373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964880602284369e-05,
+      "loss": 0.8945,
+      "step": 8374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963012866738767e-05,
+      "loss": 0.8555,
+      "step": 8375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961145097283996e-05,
+      "loss": 0.8945,
+      "step": 8376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095927729398582e-05,
+      "loss": 0.9336,
+      "step": 8377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957409456910011e-05,
+      "loss": 0.8672,
+      "step": 8378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955541586122336e-05,
+      "loss": 0.8984,
+      "step": 8379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953673681688568e-05,
+      "loss": 0.8555,
+      "step": 8380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0951805743674475e-05,
+      "loss": 0.7969,
+      "step": 8381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949937772145836e-05,
+      "loss": 0.8984,
+      "step": 8382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0948069767168416e-05,
+      "loss": 0.8711,
+      "step": 8383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946201728808001e-05,
+      "loss": 0.9219,
+      "step": 8384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094433365713036e-05,
+      "loss": 0.8672,
+      "step": 8385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942465552201271e-05,
+      "loss": 0.7656,
+      "step": 8386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940597414086519e-05,
+      "loss": 0.9102,
+      "step": 8387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938729242851876e-05,
+      "loss": 0.9844,
+      "step": 8388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0936861038563128e-05,
+      "loss": 0.7773,
+      "step": 8389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934992801286056e-05,
+      "loss": 0.832,
+      "step": 8390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933124531086445e-05,
+      "loss": 0.8477,
+      "step": 8391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931256228030076e-05,
+      "loss": 0.8164,
+      "step": 8392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0929387892182741e-05,
+      "loss": 0.7852,
+      "step": 8393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092751952361022e-05,
+      "loss": 0.8984,
+      "step": 8394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925651122378308e-05,
+      "loss": 0.8398,
+      "step": 8395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923782688552786e-05,
+      "loss": 0.832,
+      "step": 8396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921914222199454e-05,
+      "loss": 0.8828,
+      "step": 8397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920045723384096e-05,
+      "loss": 0.8633,
+      "step": 8398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0918177192172508e-05,
+      "loss": 0.9297,
+      "step": 8399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916308628630483e-05,
+      "loss": 0.75,
+      "step": 8400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0914440032823819e-05,
+      "loss": 0.7969,
+      "step": 8401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912571404818306e-05,
+      "loss": 0.7969,
+      "step": 8402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910702744679747e-05,
+      "loss": 1.0547,
+      "step": 8403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908834052473938e-05,
+      "loss": 0.8828,
+      "step": 8404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0906965328266679e-05,
+      "loss": 0.8203,
+      "step": 8405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905096572123768e-05,
+      "loss": 0.8086,
+      "step": 8406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903227784111013e-05,
+      "loss": 0.9375,
+      "step": 8407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090135896429421e-05,
+      "loss": 0.8477,
+      "step": 8408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0899490112739168e-05,
+      "loss": 0.8359,
+      "step": 8409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897621229511688e-05,
+      "loss": 0.7969,
+      "step": 8410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895752314677584e-05,
+      "loss": 0.793,
+      "step": 8411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089388336830265e-05,
+      "loss": 0.8438,
+      "step": 8412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089201439045271e-05,
+      "loss": 0.9453,
+      "step": 8413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890145381193562e-05,
+      "loss": 1.0703,
+      "step": 8414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0888276340591021e-05,
+      "loss": 1.0547,
+      "step": 8415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08864072687109e-05,
+      "loss": 1.0391,
+      "step": 8416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088453816561901e-05,
+      "loss": 0.8789,
+      "step": 8417
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882669031381163e-05,
+      "loss": 0.8984,
+      "step": 8418
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088079986606318e-05,
+      "loss": 0.7383,
+      "step": 8419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878930669730871e-05,
+      "loss": 0.8906,
+      "step": 8420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877061442450058e-05,
+      "loss": 0.9453,
+      "step": 8421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875192184286556e-05,
+      "loss": 0.8359,
+      "step": 8422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873322895306188e-05,
+      "loss": 0.9023,
+      "step": 8423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087145357557477e-05,
+      "loss": 0.8906,
+      "step": 8424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086958422515813e-05,
+      "loss": 0.8672,
+      "step": 8425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0867714844122082e-05,
+      "loss": 0.7422,
+      "step": 8426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0865845432532458e-05,
+      "loss": 0.8867,
+      "step": 8427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863975990455078e-05,
+      "loss": 1.0078,
+      "step": 8428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086210651795577e-05,
+      "loss": 0.8633,
+      "step": 8429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860237015100359e-05,
+      "loss": 0.875,
+      "step": 8430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0858367481954675e-05,
+      "loss": 0.8086,
+      "step": 8431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0856497918584548e-05,
+      "loss": 0.8242,
+      "step": 8432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854628325055805e-05,
+      "loss": 0.8984,
+      "step": 8433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852758701434281e-05,
+      "loss": 0.9766,
+      "step": 8434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0850889047785804e-05,
+      "loss": 0.8359,
+      "step": 8435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0849019364176214e-05,
+      "loss": 0.9492,
+      "step": 8436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847149650671337e-05,
+      "loss": 0.8555,
+      "step": 8437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845279907337016e-05,
+      "loss": 0.8945,
+      "step": 8438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084341013423908e-05,
+      "loss": 0.8672,
+      "step": 8439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841540331443376e-05,
+      "loss": 0.8672,
+      "step": 8440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0839670499015736e-05,
+      "loss": 0.8125,
+      "step": 8441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0837800637022e-05,
+      "loss": 0.9141,
+      "step": 8442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083593074552801e-05,
+      "loss": 0.9297,
+      "step": 8443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083406082459961e-05,
+      "loss": 0.918,
+      "step": 8444
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0832190874302636e-05,
+      "loss": 0.9844,
+      "step": 8445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830320894702942e-05,
+      "loss": 0.8438,
+      "step": 8446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828450885866363e-05,
+      "loss": 0.8516,
+      "step": 8447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826580847858755e-05,
+      "loss": 0.9336,
+      "step": 8448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0824710780745954e-05,
+      "loss": 0.9375,
+      "step": 8449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0822840684593814e-05,
+      "loss": 0.9336,
+      "step": 8450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820970559468185e-05,
+      "loss": 0.9062,
+      "step": 8451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081910040543491e-05,
+      "loss": 0.75,
+      "step": 8452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0817230222559849e-05,
+      "loss": 0.957,
+      "step": 8453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815360010908848e-05,
+      "loss": 0.8867,
+      "step": 8454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813489770547765e-05,
+      "loss": 0.918,
+      "step": 8455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081161950154245e-05,
+      "loss": 0.8906,
+      "step": 8456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809749203958759e-05,
+      "loss": 0.7383,
+      "step": 8457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0807878877862548e-05,
+      "loss": 1.0312,
+      "step": 8458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0806008523319674e-05,
+      "loss": 0.9336,
+      "step": 8459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804138140395995e-05,
+      "loss": 0.8281,
+      "step": 8460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0802267729157372e-05,
+      "loss": 0.7617,
+      "step": 8461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0800397289669661e-05,
+      "loss": 0.7617,
+      "step": 8462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798526821998729e-05,
+      "loss": 0.8086,
+      "step": 8463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796656326210436e-05,
+      "loss": 0.7695,
+      "step": 8464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794785802370637e-05,
+      "loss": 0.8633,
+      "step": 8465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0792915250545208e-05,
+      "loss": 0.8477,
+      "step": 8466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791044670800008e-05,
+      "loss": 0.9258,
+      "step": 8467
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0789174063200905e-05,
+      "loss": 0.8945,
+      "step": 8468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078730342781376e-05,
+      "loss": 0.8633,
+      "step": 8469
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785432764704451e-05,
+      "loss": 1.0391,
+      "step": 8470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783562073938838e-05,
+      "loss": 0.8281,
+      "step": 8471
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0781691355582797e-05,
+      "loss": 0.8984,
+      "step": 8472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779820609702194e-05,
+      "loss": 0.9609,
+      "step": 8473
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0777949836362905e-05,
+      "loss": 0.8203,
+      "step": 8474
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07760790356308e-05,
+      "loss": 0.8672,
+      "step": 8475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774208207571754e-05,
+      "loss": 0.9375,
+      "step": 8476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772337352251643e-05,
+      "loss": 1.0312,
+      "step": 8477
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077046646973634e-05,
+      "loss": 1.0234,
+      "step": 8478
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768595560091724e-05,
+      "loss": 1.0,
+      "step": 8479
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766724623383671e-05,
+      "loss": 1.0156,
+      "step": 8480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076485365967806e-05,
+      "loss": 0.8672,
+      "step": 8481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762982669040772e-05,
+      "loss": 0.8242,
+      "step": 8482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761111651537683e-05,
+      "loss": 0.8281,
+      "step": 8483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759240607234682e-05,
+      "loss": 0.8594,
+      "step": 8484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0757369536197644e-05,
+      "loss": 0.8984,
+      "step": 8485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755498438492458e-05,
+      "loss": 0.8359,
+      "step": 8486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753627314185002e-05,
+      "loss": 1.0156,
+      "step": 8487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751756163341167e-05,
+      "loss": 1.0156,
+      "step": 8488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0749884986026835e-05,
+      "loss": 0.875,
+      "step": 8489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748013782307895e-05,
+      "loss": 0.8594,
+      "step": 8490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746142552250235e-05,
+      "loss": 0.9062,
+      "step": 8491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744271295919746e-05,
+      "loss": 0.8359,
+      "step": 8492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742400013382313e-05,
+      "loss": 0.9766,
+      "step": 8493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740528704703833e-05,
+      "loss": 0.7539,
+      "step": 8494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073865736995019e-05,
+      "loss": 0.8945,
+      "step": 8495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073678600918728e-05,
+      "loss": 0.8359,
+      "step": 8496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734914622481e-05,
+      "loss": 0.7656,
+      "step": 8497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073304320989724e-05,
+      "loss": 0.8203,
+      "step": 8498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731171771501898e-05,
+      "loss": 0.9102,
+      "step": 8499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729300307360866e-05,
+      "loss": 0.793,
+      "step": 8500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0727428817540046e-05,
+      "loss": 0.9492,
+      "step": 8501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0725557302105334e-05,
+      "loss": 0.8906,
+      "step": 8502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723685761122632e-05,
+      "loss": 0.8594,
+      "step": 8503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0721814194657832e-05,
+      "loss": 0.8516,
+      "step": 8504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719942602776843e-05,
+      "loss": 1.0391,
+      "step": 8505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718070985545561e-05,
+      "loss": 0.7656,
+      "step": 8506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0716199343029893e-05,
+      "loss": 0.9453,
+      "step": 8507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714327675295739e-05,
+      "loss": 0.8789,
+      "step": 8508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712455982409007e-05,
+      "loss": 1.0,
+      "step": 8509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0710584264435595e-05,
+      "loss": 0.9531,
+      "step": 8510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0708712521441418e-05,
+      "loss": 0.918,
+      "step": 8511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0706840753492378e-05,
+      "loss": 0.793,
+      "step": 8512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704968960654383e-05,
+      "loss": 0.7031,
+      "step": 8513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703097142993342e-05,
+      "loss": 0.8359,
+      "step": 8514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701225300575166e-05,
+      "loss": 0.7656,
+      "step": 8515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0699353433465764e-05,
+      "loss": 0.8164,
+      "step": 8516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069748154173105e-05,
+      "loss": 0.875,
+      "step": 8517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695609625436932e-05,
+      "loss": 0.9336,
+      "step": 8518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693737684649325e-05,
+      "loss": 0.9062,
+      "step": 8519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691865719434143e-05,
+      "loss": 0.8906,
+      "step": 8520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689993729857301e-05,
+      "loss": 0.7969,
+      "step": 8521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688121715984713e-05,
+      "loss": 0.9023,
+      "step": 8522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06862496778823e-05,
+      "loss": 0.8047,
+      "step": 8523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0684377615615975e-05,
+      "loss": 0.7773,
+      "step": 8524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068250552925166e-05,
+      "loss": 1.0234,
+      "step": 8525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0680633418855266e-05,
+      "loss": 0.8789,
+      "step": 8526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678761284492722e-05,
+      "loss": 0.8281,
+      "step": 8527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676889126229943e-05,
+      "loss": 0.8633,
+      "step": 8528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0675016944132854e-05,
+      "loss": 0.9062,
+      "step": 8529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673144738267376e-05,
+      "loss": 1.0312,
+      "step": 8530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0671272508699434e-05,
+      "loss": 0.9414,
+      "step": 8531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0669400255494949e-05,
+      "loss": 0.875,
+      "step": 8532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667527978719848e-05,
+      "loss": 0.8672,
+      "step": 8533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0665655678440054e-05,
+      "loss": 0.8984,
+      "step": 8534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663783354721499e-05,
+      "loss": 0.8125,
+      "step": 8535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0661911007630105e-05,
+      "loss": 1.0781,
+      "step": 8536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0660038637231804e-05,
+      "loss": 0.9062,
+      "step": 8537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658166243592523e-05,
+      "loss": 0.7188,
+      "step": 8538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656293826778193e-05,
+      "loss": 0.8359,
+      "step": 8539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0654421386854741e-05,
+      "loss": 0.8672,
+      "step": 8540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652548923888104e-05,
+      "loss": 0.8906,
+      "step": 8541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0650676437944212e-05,
+      "loss": 0.8672,
+      "step": 8542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0648803929088998e-05,
+      "loss": 0.9297,
+      "step": 8543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646931397388393e-05,
+      "loss": 0.8516,
+      "step": 8544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064505884290834e-05,
+      "loss": 0.9492,
+      "step": 8545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643186265714764e-05,
+      "loss": 1.0391,
+      "step": 8546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641313665873611e-05,
+      "loss": 0.8984,
+      "step": 8547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0639441043450809e-05,
+      "loss": 1.0312,
+      "step": 8548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637568398512305e-05,
+      "loss": 0.8281,
+      "step": 8549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635695731124029e-05,
+      "loss": 0.8672,
+      "step": 8550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063382304135193e-05,
+      "loss": 1.0625,
+      "step": 8551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063195032926194e-05,
+      "loss": 0.8672,
+      "step": 8552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0630077594920003e-05,
+      "loss": 0.8945,
+      "step": 8553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0628204838392063e-05,
+      "loss": 0.9961,
+      "step": 8554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626332059744057e-05,
+      "loss": 0.9219,
+      "step": 8555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0624459259041934e-05,
+      "loss": 0.832,
+      "step": 8556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622586436351636e-05,
+      "loss": 0.9258,
+      "step": 8557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0620713591739111e-05,
+      "loss": 0.918,
+      "step": 8558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.06188407252703e-05,
+      "loss": 0.8555,
+      "step": 8559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616967837011151e-05,
+      "loss": 0.7383,
+      "step": 8560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0615094927027611e-05,
+      "loss": 0.9375,
+      "step": 8561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613221995385633e-05,
+      "loss": 0.7656,
+      "step": 8562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611349042151155e-05,
+      "loss": 0.875,
+      "step": 8563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060947606739014e-05,
+      "loss": 1.0625,
+      "step": 8564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607603071168526e-05,
+      "loss": 0.8359,
+      "step": 8565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0605730053552275e-05,
+      "loss": 0.8594,
+      "step": 8566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060385701460733e-05,
+      "loss": 0.8398,
+      "step": 8567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060198395439965e-05,
+      "loss": 0.9492,
+      "step": 8568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0600110872995182e-05,
+      "loss": 0.9922,
+      "step": 8569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0598237770459887e-05,
+      "loss": 0.8398,
+      "step": 8570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0596364646859714e-05,
+      "loss": 0.8281,
+      "step": 8571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594491502260623e-05,
+      "loss": 0.9492,
+      "step": 8572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592618336728567e-05,
+      "loss": 0.8359,
+      "step": 8573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0590745150329505e-05,
+      "loss": 0.8633,
+      "step": 8574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0588871943129393e-05,
+      "loss": 0.8711,
+      "step": 8575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586998715194192e-05,
+      "loss": 1.0156,
+      "step": 8576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585125466589859e-05,
+      "loss": 1.0078,
+      "step": 8577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0583252197382354e-05,
+      "loss": 0.9062,
+      "step": 8578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0581378907637641e-05,
+      "loss": 0.875,
+      "step": 8579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579505597421678e-05,
+      "loss": 0.8555,
+      "step": 8580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577632266800426e-05,
+      "loss": 0.9922,
+      "step": 8581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575758915839854e-05,
+      "loss": 0.7969,
+      "step": 8582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573885544605918e-05,
+      "loss": 1.0234,
+      "step": 8583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0572012153164588e-05,
+      "loss": 0.8945,
+      "step": 8584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0570138741581826e-05,
+      "loss": 0.8438,
+      "step": 8585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05682653099236e-05,
+      "loss": 0.9375,
+      "step": 8586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0566391858255873e-05,
+      "loss": 1.0469,
+      "step": 8587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0564518386644616e-05,
+      "loss": 0.8516,
+      "step": 8588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562644895155794e-05,
+      "loss": 0.9297,
+      "step": 8589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560771383855377e-05,
+      "loss": 0.8359,
+      "step": 8590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055889785280933e-05,
+      "loss": 0.9766,
+      "step": 8591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557024302083635e-05,
+      "loss": 0.9531,
+      "step": 8592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555150731744247e-05,
+      "loss": 0.8867,
+      "step": 8593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553277141857148e-05,
+      "loss": 0.9922,
+      "step": 8594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0551403532488304e-05,
+      "loss": 0.7891,
+      "step": 8595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549529903703692e-05,
+      "loss": 0.8945,
+      "step": 8596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547656255569283e-05,
+      "loss": 0.9102,
+      "step": 8597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545782588151053e-05,
+      "loss": 0.957,
+      "step": 8598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543908901514975e-05,
+      "loss": 0.8359,
+      "step": 8599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542035195727024e-05,
+      "loss": 0.8164,
+      "step": 8600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540161470853177e-05,
+      "loss": 0.8789,
+      "step": 8601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0538287726959411e-05,
+      "loss": 0.9336,
+      "step": 8602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536413964111703e-05,
+      "loss": 0.9453,
+      "step": 8603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053454018237603e-05,
+      "loss": 0.8516,
+      "step": 8604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0532666381818373e-05,
+      "loss": 0.8945,
+      "step": 8605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053079256250471e-05,
+      "loss": 0.8281,
+      "step": 8606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052891872450102e-05,
+      "loss": 0.8789,
+      "step": 8607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527044867873288e-05,
+      "loss": 0.9531,
+      "step": 8608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0525170992687493e-05,
+      "loss": 0.9453,
+      "step": 8609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0523297099009613e-05,
+      "loss": 0.9688,
+      "step": 8610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0521423186905635e-05,
+      "loss": 0.9688,
+      "step": 8611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519549256441541e-05,
+      "loss": 0.8789,
+      "step": 8612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517675307683316e-05,
+      "loss": 0.7969,
+      "step": 8613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0515801340696946e-05,
+      "loss": 0.9062,
+      "step": 8614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0513927355548414e-05,
+      "loss": 0.9375,
+      "step": 8615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512053352303704e-05,
+      "loss": 0.8125,
+      "step": 8616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510179331028806e-05,
+      "loss": 0.9492,
+      "step": 8617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0508305291789707e-05,
+      "loss": 0.9375,
+      "step": 8618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506431234652392e-05,
+      "loss": 0.9297,
+      "step": 8619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504557159682853e-05,
+      "loss": 0.9492,
+      "step": 8620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0502683066947077e-05,
+      "loss": 0.8711,
+      "step": 8621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500808956511052e-05,
+      "loss": 0.8281,
+      "step": 8622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498934828440774e-05,
+      "loss": 0.7461,
+      "step": 8623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049706068280223e-05,
+      "loss": 0.7969,
+      "step": 8624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0495186519661409e-05,
+      "loss": 0.7812,
+      "step": 8625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049331233908431e-05,
+      "loss": 0.8555,
+      "step": 8626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0491438141136915e-05,
+      "loss": 0.8281,
+      "step": 8627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048956392588523e-05,
+      "loss": 0.7852,
+      "step": 8628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487689693395243e-05,
+      "loss": 0.8984,
+      "step": 8629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048581544373295e-05,
+      "loss": 0.8281,
+      "step": 8630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0483941176964341e-05,
+      "loss": 0.8867,
+      "step": 8631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482066893155419e-05,
+      "loss": 0.8242,
+      "step": 8632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0480192592372173e-05,
+      "loss": 0.9102,
+      "step": 8633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047831827468061e-05,
+      "loss": 0.9492,
+      "step": 8634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047644394014672e-05,
+      "loss": 0.8672,
+      "step": 8635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0474569588836503e-05,
+      "loss": 0.8398,
+      "step": 8636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0472695220815958e-05,
+      "loss": 0.9766,
+      "step": 8637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0470820836151086e-05,
+      "loss": 0.8906,
+      "step": 8638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468946434907882e-05,
+      "loss": 0.9883,
+      "step": 8639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467072017152353e-05,
+      "loss": 1.0781,
+      "step": 8640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0465197582950496e-05,
+      "loss": 0.9219,
+      "step": 8641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0463323132368312e-05,
+      "loss": 0.8984,
+      "step": 8642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046144866547181e-05,
+      "loss": 0.8711,
+      "step": 8643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0459574182326984e-05,
+      "loss": 0.957,
+      "step": 8644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457699682999842e-05,
+      "loss": 0.8242,
+      "step": 8645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455825167556387e-05,
+      "loss": 0.9141,
+      "step": 8646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453950636062626e-05,
+      "loss": 0.8242,
+      "step": 8647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045207608858456e-05,
+      "loss": 0.8281,
+      "step": 8648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04502015251882e-05,
+      "loss": 0.9375,
+      "step": 8649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448326945939547e-05,
+      "loss": 0.9531,
+      "step": 8650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446452350904611e-05,
+      "loss": 0.832,
+      "step": 8651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444577740149395e-05,
+      "loss": 0.9414,
+      "step": 8652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442703113739914e-05,
+      "loss": 0.8438,
+      "step": 8653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440828471742172e-05,
+      "loss": 0.8594,
+      "step": 8654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438953814222178e-05,
+      "loss": 0.8984,
+      "step": 8655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437079141245942e-05,
+      "loss": 0.793,
+      "step": 8656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043520445287948e-05,
+      "loss": 0.9062,
+      "step": 8657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043332974918879e-05,
+      "loss": 0.8867,
+      "step": 8658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0431455030239895e-05,
+      "loss": 0.9883,
+      "step": 8659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04295802960988e-05,
+      "loss": 1.0156,
+      "step": 8660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427705546831522e-05,
+      "loss": 0.9336,
+      "step": 8661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0425830782504068e-05,
+      "loss": 0.957,
+      "step": 8662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423956003182455e-05,
+      "loss": 0.8477,
+      "step": 8663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422081208932697e-05,
+      "loss": 0.9102,
+      "step": 8664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420206399820812e-05,
+      "loss": 0.8789,
+      "step": 8665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0418331575912808e-05,
+      "loss": 0.8555,
+      "step": 8666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04164567372747e-05,
+      "loss": 0.793,
+      "step": 8667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414581883972511e-05,
+      "loss": 0.8281,
+      "step": 8668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0412707016072254e-05,
+      "loss": 1.0078,
+      "step": 8669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0410832133639945e-05,
+      "loss": 0.8984,
+      "step": 8670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408957236741601e-05,
+      "loss": 0.9258,
+      "step": 8671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407082325443243e-05,
+      "loss": 0.7773,
+      "step": 8672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405207399810887e-05,
+      "loss": 0.793,
+      "step": 8673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403332459910554e-05,
+      "loss": 0.8008,
+      "step": 8674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040145750580826e-05,
+      "loss": 1.0469,
+      "step": 8675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039958253757003e-05,
+      "loss": 0.9766,
+      "step": 8676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397707555261877e-05,
+      "loss": 0.9688,
+      "step": 8677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395832558949832e-05,
+      "loss": 0.9297,
+      "step": 8678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393957548699908e-05,
+      "loss": 0.9219,
+      "step": 8679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0392082524578135e-05,
+      "loss": 0.8164,
+      "step": 8680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390207486650525e-05,
+      "loss": 0.9023,
+      "step": 8681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388332434983112e-05,
+      "loss": 0.8125,
+      "step": 8682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038645736964191e-05,
+      "loss": 0.9102,
+      "step": 8683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384582290692952e-05,
+      "loss": 0.832,
+      "step": 8684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382707198202256e-05,
+      "loss": 0.9062,
+      "step": 8685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380832092235848e-05,
+      "loss": 0.8789,
+      "step": 8686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378956972859752e-05,
+      "loss": 0.9414,
+      "step": 8687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037708184014e-05,
+      "loss": 0.793,
+      "step": 8688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0375206694142613e-05,
+      "loss": 0.9297,
+      "step": 8689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0373331534933618e-05,
+      "loss": 0.9141,
+      "step": 8690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371456362579041e-05,
+      "loss": 0.8594,
+      "step": 8691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369581177144916e-05,
+      "loss": 0.9102,
+      "step": 8692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367705978697264e-05,
+      "loss": 0.9062,
+      "step": 8693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036583076730212e-05,
+      "loss": 0.8867,
+      "step": 8694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363955543025507e-05,
+      "loss": 0.8438,
+      "step": 8695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036208030593346e-05,
+      "loss": 0.793,
+      "step": 8696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0360205056092006e-05,
+      "loss": 0.9102,
+      "step": 8697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358329793567174e-05,
+      "loss": 0.8828,
+      "step": 8698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0356454518424996e-05,
+      "loss": 0.8906,
+      "step": 8699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0354579230731507e-05,
+      "loss": 0.8672,
+      "step": 8700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352703930552732e-05,
+      "loss": 0.7227,
+      "step": 8701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350828617954712e-05,
+      "loss": 0.8633,
+      "step": 8702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034895329300347e-05,
+      "loss": 0.8203,
+      "step": 8703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0347077955765049e-05,
+      "loss": 0.9062,
+      "step": 8704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345202606305471e-05,
+      "loss": 0.9297,
+      "step": 8705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034332724469078e-05,
+      "loss": 0.8672,
+      "step": 8706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341451870987005e-05,
+      "loss": 0.9062,
+      "step": 8707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339576485260186e-05,
+      "loss": 0.8594,
+      "step": 8708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0337701087576351e-05,
+      "loss": 0.8438,
+      "step": 8709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335825678001541e-05,
+      "loss": 0.8281,
+      "step": 8710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0333950256601789e-05,
+      "loss": 0.9453,
+      "step": 8711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0332074823443132e-05,
+      "loss": 0.9688,
+      "step": 8712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033019937859161e-05,
+      "loss": 1.0,
+      "step": 8713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0328323922113257e-05,
+      "loss": 0.8555,
+      "step": 8714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0326448454074112e-05,
+      "loss": 0.9492,
+      "step": 8715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0324572974540215e-05,
+      "loss": 0.9844,
+      "step": 8716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0322697483577597e-05,
+      "loss": 0.9102,
+      "step": 8717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320821981252308e-05,
+      "loss": 0.9688,
+      "step": 8718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0318946467630378e-05,
+      "loss": 0.8086,
+      "step": 8719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0317070942777853e-05,
+      "loss": 0.8711,
+      "step": 8720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315195406760772e-05,
+      "loss": 1.1484,
+      "step": 8721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031331985964517e-05,
+      "loss": 0.9102,
+      "step": 8722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311444301497093e-05,
+      "loss": 0.8984,
+      "step": 8723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309568732382582e-05,
+      "loss": 0.8984,
+      "step": 8724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307693152367677e-05,
+      "loss": 0.8008,
+      "step": 8725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030581756151842e-05,
+      "loss": 0.9141,
+      "step": 8726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0303941959900858e-05,
+      "loss": 1.1094,
+      "step": 8727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302066347581028e-05,
+      "loss": 0.8594,
+      "step": 8728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0300190724624976e-05,
+      "loss": 0.918,
+      "step": 8729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298315091098744e-05,
+      "loss": 0.8711,
+      "step": 8730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029643944706838e-05,
+      "loss": 0.7852,
+      "step": 8731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294563792599923e-05,
+      "loss": 0.8438,
+      "step": 8732
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0292688127759422e-05,
+      "loss": 0.8125,
+      "step": 8733
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290812452612917e-05,
+      "loss": 0.793,
+      "step": 8734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0288936767226461e-05,
+      "loss": 0.9102,
+      "step": 8735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028706107166609e-05,
+      "loss": 0.875,
+      "step": 8736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0285185365997858e-05,
+      "loss": 0.9219,
+      "step": 8737
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283309650287809e-05,
+      "loss": 0.8594,
+      "step": 8738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281433924601992e-05,
+      "loss": 0.9883,
+      "step": 8739
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0279558189006446e-05,
+      "loss": 0.8281,
+      "step": 8740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0277682443567232e-05,
+      "loss": 0.8125,
+      "step": 8741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275806688350385e-05,
+      "loss": 0.918,
+      "step": 8742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027393092342196e-05,
+      "loss": 0.7969,
+      "step": 8743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0272055148848003e-05,
+      "loss": 0.9727,
+      "step": 8744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0270179364694568e-05,
+      "loss": 0.8789,
+      "step": 8745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268303571027696e-05,
+      "loss": 0.8906,
+      "step": 8746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0266427767913446e-05,
+      "loss": 0.9375,
+      "step": 8747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264551955417858e-05,
+      "loss": 0.8516,
+      "step": 8748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026267613360699e-05,
+      "loss": 0.9375,
+      "step": 8749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260800302546887e-05,
+      "loss": 0.7344,
+      "step": 8750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0258924462303608e-05,
+      "loss": 0.7891,
+      "step": 8751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257048612943193e-05,
+      "loss": 0.8359,
+      "step": 8752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255172754531704e-05,
+      "loss": 0.8477,
+      "step": 8753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253296887135185e-05,
+      "loss": 0.8672,
+      "step": 8754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251421010819693e-05,
+      "loss": 0.8359,
+      "step": 8755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249545125651279e-05,
+      "loss": 0.9688,
+      "step": 8756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247669231695995e-05,
+      "loss": 0.7891,
+      "step": 8757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0245793329019896e-05,
+      "loss": 1.1094,
+      "step": 8758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243917417689036e-05,
+      "loss": 0.8594,
+      "step": 8759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242041497769464e-05,
+      "loss": 0.9648,
+      "step": 8760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0240165569327239e-05,
+      "loss": 0.9453,
+      "step": 8761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023828963242841e-05,
+      "loss": 0.8008,
+      "step": 8762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236413687139039e-05,
+      "loss": 0.8086,
+      "step": 8763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0234537733525177e-05,
+      "loss": 0.9102,
+      "step": 8764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023266177165288e-05,
+      "loss": 0.9336,
+      "step": 8765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02307858015882e-05,
+      "loss": 0.8945,
+      "step": 8766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228909823397195e-05,
+      "loss": 0.9023,
+      "step": 8767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0227033837145924e-05,
+      "loss": 0.8438,
+      "step": 8768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022515784290044e-05,
+      "loss": 0.8203,
+      "step": 8769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02232818407268e-05,
+      "loss": 0.8945,
+      "step": 8770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0221405830691065e-05,
+      "loss": 0.9414,
+      "step": 8771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219529812859286e-05,
+      "loss": 0.9766,
+      "step": 8772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0217653787297523e-05,
+      "loss": 0.8906,
+      "step": 8773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215777754071834e-05,
+      "loss": 0.9453,
+      "step": 8774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021390171324828e-05,
+      "loss": 0.8984,
+      "step": 8775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0212025664892913e-05,
+      "loss": 0.9102,
+      "step": 8776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210149609071798e-05,
+      "loss": 0.7812,
+      "step": 8777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020827354585099e-05,
+      "loss": 1.0391,
+      "step": 8778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0206397475296548e-05,
+      "loss": 0.8359,
+      "step": 8779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204521397474533e-05,
+      "loss": 0.7578,
+      "step": 8780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202645312451002e-05,
+      "loss": 0.8828,
+      "step": 8781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020076922029202e-05,
+      "loss": 0.7969,
+      "step": 8782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019889312106364e-05,
+      "loss": 0.8516,
+      "step": 8783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197017014831929e-05,
+      "loss": 1.0,
+      "step": 8784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0195140901662941e-05,
+      "loss": 1.0469,
+      "step": 8785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193264781622745e-05,
+      "loss": 0.8164,
+      "step": 8786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191388654777394e-05,
+      "loss": 0.8242,
+      "step": 8787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189512521192956e-05,
+      "loss": 0.9805,
+      "step": 8788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187636380935486e-05,
+      "loss": 0.875,
+      "step": 8789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0185760234071054e-05,
+      "loss": 0.8633,
+      "step": 8790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0183884080665713e-05,
+      "loss": 0.8594,
+      "step": 8791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0182007920785533e-05,
+      "loss": 0.8867,
+      "step": 8792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180131754496569e-05,
+      "loss": 0.7812,
+      "step": 8793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0178255581864893e-05,
+      "loss": 0.9961,
+      "step": 8794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0176379402956557e-05,
+      "loss": 0.7383,
+      "step": 8795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174503217837635e-05,
+      "loss": 0.9258,
+      "step": 8796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172627026574185e-05,
+      "loss": 0.9102,
+      "step": 8797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0170750829232268e-05,
+      "loss": 0.8438,
+      "step": 8798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168874625877953e-05,
+      "loss": 0.8984,
+      "step": 8799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166998416577299e-05,
+      "loss": 0.8516,
+      "step": 8800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0165122201396377e-05,
+      "loss": 0.8906,
+      "step": 8801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163245980401242e-05,
+      "loss": 0.9766,
+      "step": 8802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161369753657967e-05,
+      "loss": 0.9062,
+      "step": 8803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0159493521232613e-05,
+      "loss": 0.8672,
+      "step": 8804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157617283191248e-05,
+      "loss": 0.793,
+      "step": 8805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155741039599931e-05,
+      "loss": 0.9375,
+      "step": 8806
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0153864790524736e-05,
+      "loss": 0.9453,
+      "step": 8807
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015198853603172e-05,
+      "loss": 0.7969,
+      "step": 8808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150112276186957e-05,
+      "loss": 0.7695,
+      "step": 8809
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0148236011056506e-05,
+      "loss": 0.8984,
+      "step": 8810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146359740706437e-05,
+      "loss": 0.8906,
+      "step": 8811
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144483465202818e-05,
+      "loss": 0.8633,
+      "step": 8812
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142607184611711e-05,
+      "loss": 0.9023,
+      "step": 8813
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0140730898999184e-05,
+      "loss": 0.8125,
+      "step": 8814
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0138854608431306e-05,
+      "loss": 0.9375,
+      "step": 8815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0136978312974142e-05,
+      "loss": 0.8516,
+      "step": 8816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013510201269376e-05,
+      "loss": 0.9336,
+      "step": 8817
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0133225707656229e-05,
+      "loss": 0.9531,
+      "step": 8818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131349397927613e-05,
+      "loss": 0.9297,
+      "step": 8819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129473083573984e-05,
+      "loss": 1.0234,
+      "step": 8820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0127596764661406e-05,
+      "loss": 0.8125,
+      "step": 8821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012572044125595e-05,
+      "loss": 0.8984,
+      "step": 8822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123844113423681e-05,
+      "loss": 0.9492,
+      "step": 8823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0121967781230675e-05,
+      "loss": 0.8477,
+      "step": 8824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0120091444742989e-05,
+      "loss": 0.9219,
+      "step": 8825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118215104026703e-05,
+      "loss": 0.9219,
+      "step": 8826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116338759147877e-05,
+      "loss": 0.8477,
+      "step": 8827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0114462410172585e-05,
+      "loss": 0.9062,
+      "step": 8828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112586057166893e-05,
+      "loss": 0.957,
+      "step": 8829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110709700196875e-05,
+      "loss": 0.9453,
+      "step": 8830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108833339328594e-05,
+      "loss": 0.8281,
+      "step": 8831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106956974628127e-05,
+      "loss": 0.875,
+      "step": 8832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105080606161539e-05,
+      "loss": 0.9492,
+      "step": 8833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01032042339949e-05,
+      "loss": 0.7812,
+      "step": 8834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0101327858194282e-05,
+      "loss": 0.9336,
+      "step": 8835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099451478825748e-05,
+      "loss": 0.9492,
+      "step": 8836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0097575095955378e-05,
+      "loss": 0.8633,
+      "step": 8837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095698709649238e-05,
+      "loss": 0.8281,
+      "step": 8838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093822319973401e-05,
+      "loss": 0.8242,
+      "step": 8839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0091945926993932e-05,
+      "loss": 0.8867,
+      "step": 8840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090069530776905e-05,
+      "loss": 1.0078,
+      "step": 8841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008819313138839e-05,
+      "loss": 0.9141,
+      "step": 8842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0086316728894463e-05,
+      "loss": 0.8516,
+      "step": 8843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084440323361186e-05,
+      "loss": 1.0156,
+      "step": 8844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082563914854638e-05,
+      "loss": 0.8555,
+      "step": 8845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080687503440884e-05,
+      "loss": 0.8828,
+      "step": 8846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078811089186002e-05,
+      "loss": 0.8477,
+      "step": 8847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076934672156058e-05,
+      "loss": 0.9492,
+      "step": 8848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0075058252417124e-05,
+      "loss": 0.8594,
+      "step": 8849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0073181830035275e-05,
+      "loss": 0.9219,
+      "step": 8850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071305405076582e-05,
+      "loss": 0.9453,
+      "step": 8851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069428977607112e-05,
+      "loss": 0.7812,
+      "step": 8852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067552547692944e-05,
+      "loss": 1.0234,
+      "step": 8853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065676115400145e-05,
+      "loss": 0.9648,
+      "step": 8854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063799680794789e-05,
+      "loss": 0.8711,
+      "step": 8855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061923243942947e-05,
+      "loss": 0.9648,
+      "step": 8856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060046804910693e-05,
+      "loss": 0.9414,
+      "step": 8857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058170363764096e-05,
+      "loss": 0.7578,
+      "step": 8858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056293920569234e-05,
+      "loss": 0.9805,
+      "step": 8859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054417475392171e-05,
+      "loss": 0.7539,
+      "step": 8860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052541028298989e-05,
+      "loss": 0.793,
+      "step": 8861
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050664579355755e-05,
+      "loss": 0.793,
+      "step": 8862
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048788128628545e-05,
+      "loss": 0.7344,
+      "step": 8863
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046911676183426e-05,
+      "loss": 0.957,
+      "step": 8864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045035222086477e-05,
+      "loss": 1.0,
+      "step": 8865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043158766403767e-05,
+      "loss": 0.8125,
+      "step": 8866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041282309201372e-05,
+      "loss": 0.793,
+      "step": 8867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003940585054536e-05,
+      "loss": 0.9219,
+      "step": 8868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003752939050181e-05,
+      "loss": 0.7891,
+      "step": 8869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035652929136793e-05,
+      "loss": 0.9531,
+      "step": 8870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033776466516379e-05,
+      "loss": 1.0156,
+      "step": 8871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031900002706643e-05,
+      "loss": 0.8906,
+      "step": 8872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030023537773661e-05,
+      "loss": 0.8867,
+      "step": 8873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028147071783501e-05,
+      "loss": 0.9023,
+      "step": 8874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026270604802242e-05,
+      "loss": 0.8203,
+      "step": 8875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024394136895953e-05,
+      "loss": 0.9688,
+      "step": 8876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002251766813071e-05,
+      "loss": 0.8984,
+      "step": 8877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020641198572582e-05,
+      "loss": 0.8711,
+      "step": 8878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018764728287649e-05,
+      "loss": 0.8281,
+      "step": 8879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0016888257341978e-05,
+      "loss": 0.8672,
+      "step": 8880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0015011785801649e-05,
+      "loss": 0.9336,
+      "step": 8881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001313531373273e-05,
+      "loss": 0.7891,
+      "step": 8882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011258841201297e-05,
+      "loss": 0.8945,
+      "step": 8883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0009382368273419e-05,
+      "loss": 0.9219,
+      "step": 8884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0007505895015179e-05,
+      "loss": 0.8203,
+      "step": 8885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000562942149264e-05,
+      "loss": 0.8359,
+      "step": 8886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0003752947771883e-05,
+      "loss": 0.875,
+      "step": 8887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0001876473918979e-05,
+      "loss": 0.957,
+      "step": 8888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.8203,
+      "step": 8889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998123526081023e-06,
+      "loss": 0.9375,
+      "step": 8890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99624705222812e-06,
+      "loss": 0.7695,
+      "step": 8891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994370578507362e-06,
+      "loss": 0.875,
+      "step": 8892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992494104984824e-06,
+      "loss": 0.957,
+      "step": 8893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99061763172658e-06,
+      "loss": 0.9688,
+      "step": 8894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98874115879871e-06,
+      "loss": 0.9609,
+      "step": 8895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986864686267274e-06,
+      "loss": 0.9219,
+      "step": 8896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984988214198353e-06,
+      "loss": 0.8828,
+      "step": 8897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983111742658022e-06,
+      "loss": 0.7852,
+      "step": 8898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981235271712355e-06,
+      "loss": 0.8281,
+      "step": 8899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97935880142742e-06,
+      "loss": 0.8359,
+      "step": 8900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.977482331869294e-06,
+      "loss": 0.8164,
+      "step": 8901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975605863104047e-06,
+      "loss": 0.8984,
+      "step": 8902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973729395197763e-06,
+      "loss": 1.0234,
+      "step": 8903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9718529282165e-06,
+      "loss": 0.8828,
+      "step": 8904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969976462226342e-06,
+      "loss": 0.8438,
+      "step": 8905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968099997293357e-06,
+      "loss": 0.7734,
+      "step": 8906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966223533483626e-06,
+      "loss": 0.9336,
+      "step": 8907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964347070863212e-06,
+      "loss": 0.9258,
+      "step": 8908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962470609498191e-06,
+      "loss": 0.8086,
+      "step": 8909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96059414945464e-06,
+      "loss": 0.9102,
+      "step": 8910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958717690798633e-06,
+      "loss": 0.7969,
+      "step": 8911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956841233596237e-06,
+      "loss": 0.832,
+      "step": 8912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954964777913524e-06,
+      "loss": 0.8828,
+      "step": 8913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953088323816574e-06,
+      "loss": 0.8203,
+      "step": 8914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95121187137146e-06,
+      "loss": 0.9453,
+      "step": 8915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949335420644246e-06,
+      "loss": 0.8555,
+      "step": 8916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.947458971701012e-06,
+      "loss": 0.9219,
+      "step": 8917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945582524607829e-06,
+      "loss": 0.8359,
+      "step": 8918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943706079430773e-06,
+      "loss": 0.7891,
+      "step": 8919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941829636235908e-06,
+      "loss": 0.9453,
+      "step": 8920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93995319508931e-06,
+      "loss": 0.8359,
+      "step": 8921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938076756057055e-06,
+      "loss": 0.9414,
+      "step": 8922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936200319205216e-06,
+      "loss": 0.8359,
+      "step": 8923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934323884599858e-06,
+      "loss": 0.875,
+      "step": 8924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93244745230706e-06,
+      "loss": 0.8477,
+      "step": 8925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930571022392888e-06,
+      "loss": 0.8086,
+      "step": 8926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928694594923423e-06,
+      "loss": 0.8555,
+      "step": 8927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.926818169964729e-06,
+      "loss": 0.918,
+      "step": 8928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.924941747582878e-06,
+      "loss": 0.8398,
+      "step": 8929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923065327843944e-06,
+      "loss": 0.8242,
+      "step": 8930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.921188910814003e-06,
+      "loss": 0.8164,
+      "step": 8931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919312496559118e-06,
+      "loss": 0.8555,
+      "step": 8932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917436085145364e-06,
+      "loss": 0.8711,
+      "step": 8933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915559676638814e-06,
+      "loss": 0.8125,
+      "step": 8934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.913683271105542e-06,
+      "loss": 0.8867,
+      "step": 8935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.911806868611612e-06,
+      "loss": 0.9023,
+      "step": 8936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909930469223097e-06,
+      "loss": 0.7695,
+      "step": 8937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90805407300607e-06,
+      "loss": 0.875,
+      "step": 8938
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906177680026606e-06,
+      "loss": 0.8477,
+      "step": 8939
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904301290350765e-06,
+      "loss": 0.9375,
+      "step": 8940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.902424904044623e-06,
+      "loss": 0.7617,
+      "step": 8941
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900548521174255e-06,
+      "loss": 0.8984,
+      "step": 8942
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.898672141805725e-06,
+      "loss": 0.9062,
+      "step": 8943
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896795766005104e-06,
+      "loss": 0.7539,
+      "step": 8944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894919393838463e-06,
+      "loss": 0.8906,
+      "step": 8945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893043025371878e-06,
+      "loss": 0.8984,
+      "step": 8946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89116666067141e-06,
+      "loss": 0.8008,
+      "step": 8947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889290299803129e-06,
+      "loss": 0.8398,
+      "step": 8948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887413942833107e-06,
+      "loss": 0.8438,
+      "step": 8949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88553758982742e-06,
+      "loss": 0.8867,
+      "step": 8950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883661240852128e-06,
+      "loss": 0.9883,
+      "step": 8951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881784895973302e-06,
+      "loss": 0.9766,
+      "step": 8952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87990855525701e-06,
+      "loss": 0.9062,
+      "step": 8953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.878032218769332e-06,
+      "loss": 0.9375,
+      "step": 8954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87615588657632e-06,
+      "loss": 0.9414,
+      "step": 8955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.874279558744053e-06,
+      "loss": 0.8008,
+      "step": 8956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872403235338596e-06,
+      "loss": 0.8047,
+      "step": 8957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87052691642602e-06,
+      "loss": 0.8828,
+      "step": 8958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868650602072389e-06,
+      "loss": 0.8984,
+      "step": 8959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866774292343773e-06,
+      "loss": 0.8867,
+      "step": 8960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864897987306243e-06,
+      "loss": 0.9453,
+      "step": 8961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863021687025861e-06,
+      "loss": 0.9922,
+      "step": 8962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.861145391568697e-06,
+      "loss": 0.7969,
+      "step": 8963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859269101000818e-06,
+      "loss": 0.7656,
+      "step": 8964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857392815388292e-06,
+      "loss": 0.8438,
+      "step": 8965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855516534797187e-06,
+      "loss": 0.8477,
+      "step": 8966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853640259293564e-06,
+      "loss": 0.7617,
+      "step": 8967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851763988943495e-06,
+      "loss": 0.7812,
+      "step": 8968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.849887723813047e-06,
+      "loss": 0.9219,
+      "step": 8969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848011463968282e-06,
+      "loss": 0.8477,
+      "step": 8970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.846135209475267e-06,
+      "loss": 0.8711,
+      "step": 8971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84425896040007e-06,
+      "loss": 0.7812,
+      "step": 8972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842382716808757e-06,
+      "loss": 0.8633,
+      "step": 8973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.840506478767389e-06,
+      "loss": 0.8281,
+      "step": 8974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838630246342034e-06,
+      "loss": 0.7617,
+      "step": 8975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.836754019598761e-06,
+      "loss": 0.8047,
+      "step": 8976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834877798603627e-06,
+      "loss": 0.9141,
+      "step": 8977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833001583422704e-06,
+      "loss": 0.8789,
+      "step": 8978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83112537412205e-06,
+      "loss": 0.8242,
+      "step": 8979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829249170767736e-06,
+      "loss": 0.8594,
+      "step": 8980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827372973425818e-06,
+      "loss": 0.8984,
+      "step": 8981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825496782162367e-06,
+      "loss": 0.9531,
+      "step": 8982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823620597043445e-06,
+      "loss": 0.8086,
+      "step": 8983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82174441813511e-06,
+      "loss": 0.9062,
+      "step": 8984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81986824550343e-06,
+      "loss": 0.9297,
+      "step": 8985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81799207921447e-06,
+      "loss": 0.8281,
+      "step": 8986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816115919334289e-06,
+      "loss": 0.8125,
+      "step": 8987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81423976592895e-06,
+      "loss": 0.8984,
+      "step": 8988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812363619064514e-06,
+      "loss": 0.7578,
+      "step": 8989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810487478807047e-06,
+      "loss": 0.8438,
+      "step": 8990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808611345222609e-06,
+      "loss": 0.9219,
+      "step": 8991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806735218377258e-06,
+      "loss": 0.8711,
+      "step": 8992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80485909833706e-06,
+      "loss": 0.8516,
+      "step": 8993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802982985168075e-06,
+      "loss": 0.9648,
+      "step": 8994
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.801106878936363e-06,
+      "loss": 0.8984,
+      "step": 8995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.799230779707985e-06,
+      "loss": 0.8555,
+      "step": 8996
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797354687549e-06,
+      "loss": 0.8867,
+      "step": 8997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.795478602525472e-06,
+      "loss": 1.0156,
+      "step": 8998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.793602524703456e-06,
+      "loss": 0.9141,
+      "step": 8999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791726454149013e-06,
+      "loss": 0.8828,
+      "step": 9000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789850390928203e-06,
+      "loss": 0.8867,
+      "step": 9001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78797433510709e-06,
+      "loss": 0.8008,
+      "step": 9002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786098286751722e-06,
+      "loss": 0.9727,
+      "step": 9003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784222245928165e-06,
+      "loss": 0.8477,
+      "step": 9004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.782346212702479e-06,
+      "loss": 0.918,
+      "step": 9005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780470187140719e-06,
+      "loss": 0.8828,
+      "step": 9006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778594169308939e-06,
+      "loss": 0.7734,
+      "step": 9007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776718159273199e-06,
+      "loss": 0.9609,
+      "step": 9008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774842157099561e-06,
+      "loss": 0.8438,
+      "step": 9009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77296616285408e-06,
+      "loss": 0.832,
+      "step": 9010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771090176602807e-06,
+      "loss": 1.0703,
+      "step": 9011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769214198411802e-06,
+      "loss": 0.9062,
+      "step": 9012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767338228347127e-06,
+      "loss": 0.8633,
+      "step": 9013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.765462266474826e-06,
+      "loss": 0.8281,
+      "step": 9014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763586312860963e-06,
+      "loss": 1.1016,
+      "step": 9015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76171036757159e-06,
+      "loss": 0.8281,
+      "step": 9016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759834430672766e-06,
+      "loss": 0.8242,
+      "step": 9017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75795850223054e-06,
+      "loss": 0.9609,
+      "step": 9018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756082582310969e-06,
+      "loss": 0.9883,
+      "step": 9019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754206670980106e-06,
+      "loss": 0.8984,
+      "step": 9020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75233076830401e-06,
+      "loss": 0.7812,
+      "step": 9021
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750454874348725e-06,
+      "loss": 0.8359,
+      "step": 9022
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748578989180309e-06,
+      "loss": 0.9336,
+      "step": 9023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746703112864815e-06,
+      "loss": 0.9297,
+      "step": 9024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7448272454683e-06,
+      "loss": 0.875,
+      "step": 9025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742951387056808e-06,
+      "loss": 0.832,
+      "step": 9026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741075537696397e-06,
+      "loss": 0.875,
+      "step": 9027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.739199697453113e-06,
+      "loss": 0.8789,
+      "step": 9028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737323866393016e-06,
+      "loss": 0.9023,
+      "step": 9029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735448044582146e-06,
+      "loss": 0.8672,
+      "step": 9030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733572232086558e-06,
+      "loss": 0.9023,
+      "step": 9031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731696428972304e-06,
+      "loss": 0.793,
+      "step": 9032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.729820635305437e-06,
+      "loss": 0.9258,
+      "step": 9033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.727944851152e-06,
+      "loss": 0.918,
+      "step": 9034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.726069076578042e-06,
+      "loss": 0.8633,
+      "step": 9035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724193311649617e-06,
+      "loss": 0.7734,
+      "step": 9036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.722317556432774e-06,
+      "loss": 0.9375,
+      "step": 9037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720441810993555e-06,
+      "loss": 0.8867,
+      "step": 9038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718566075398012e-06,
+      "loss": 0.9336,
+      "step": 9039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716690349712191e-06,
+      "loss": 0.8359,
+      "step": 9040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714814634002146e-06,
+      "loss": 0.9297,
+      "step": 9041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.712938928333913e-06,
+      "loss": 0.8047,
+      "step": 9042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.711063232773544e-06,
+      "loss": 0.9492,
+      "step": 9043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.709187547387083e-06,
+      "loss": 0.9297,
+      "step": 9044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.707311872240583e-06,
+      "loss": 0.8555,
+      "step": 9045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70543620740008e-06,
+      "loss": 0.9688,
+      "step": 9046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703560552931623e-06,
+      "loss": 0.7969,
+      "step": 9047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701684908901255e-06,
+      "loss": 0.8672,
+      "step": 9048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.699809275375027e-06,
+      "loss": 0.7812,
+      "step": 9049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697933652418975e-06,
+      "loss": 0.8906,
+      "step": 9050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.696058040099145e-06,
+      "loss": 0.8633,
+      "step": 9051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69418243848158e-06,
+      "loss": 0.9492,
+      "step": 9052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692306847632327e-06,
+      "loss": 0.8359,
+      "step": 9053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690431267617421e-06,
+      "loss": 0.8281,
+      "step": 9054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688555698502909e-06,
+      "loss": 0.8594,
+      "step": 9055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686680140354832e-06,
+      "loss": 0.8477,
+      "step": 9056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684804593239233e-06,
+      "loss": 0.8008,
+      "step": 9057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68292905722215e-06,
+      "loss": 0.875,
+      "step": 9058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.681053532369622e-06,
+      "loss": 0.9453,
+      "step": 9059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679178018747697e-06,
+      "loss": 0.918,
+      "step": 9060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677302516422405e-06,
+      "loss": 0.9297,
+      "step": 9061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675427025459789e-06,
+      "loss": 0.8438,
+      "step": 9062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.673551545925888e-06,
+      "loss": 0.7852,
+      "step": 9063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671676077886747e-06,
+      "loss": 0.9648,
+      "step": 9064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669800621408393e-06,
+      "loss": 0.8125,
+      "step": 9065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66792517655687e-06,
+      "loss": 1.1094,
+      "step": 9066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666049743398213e-06,
+      "loss": 0.8398,
+      "step": 9067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664174321998466e-06,
+      "loss": 1.0078,
+      "step": 9068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662298912423654e-06,
+      "loss": 0.9453,
+      "step": 9069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66042351473982e-06,
+      "loss": 0.7539,
+      "step": 9070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658548129012996e-06,
+      "loss": 0.8164,
+      "step": 9071
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656672755309224e-06,
+      "loss": 0.8828,
+      "step": 9072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654797393694532e-06,
+      "loss": 0.8359,
+      "step": 9073
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652922044234956e-06,
+      "loss": 0.7578,
+      "step": 9074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651046706996531e-06,
+      "loss": 0.8516,
+      "step": 9075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649171382045293e-06,
+      "loss": 0.8516,
+      "step": 9076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64729606944727e-06,
+      "loss": 0.9453,
+      "step": 9077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.645420769268496e-06,
+      "loss": 0.9375,
+      "step": 9078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.643545481575003e-06,
+      "loss": 0.9219,
+      "step": 9079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641670206432831e-06,
+      "loss": 0.9297,
+      "step": 9080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.639794943907998e-06,
+      "loss": 0.8906,
+      "step": 9081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637919694066542e-06,
+      "loss": 0.8984,
+      "step": 9082
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636044456974493e-06,
+      "loss": 0.9609,
+      "step": 9083
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634169232697884e-06,
+      "loss": 0.9883,
+      "step": 9084
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632294021302738e-06,
+      "loss": 0.8906,
+      "step": 9085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630418822855086e-06,
+      "loss": 0.9219,
+      "step": 9086
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628543637420958e-06,
+      "loss": 0.8438,
+      "step": 9087
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626668465066387e-06,
+      "loss": 0.7695,
+      "step": 9088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62479330585739e-06,
+      "loss": 0.9141,
+      "step": 9089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622918159860001e-06,
+      "loss": 0.9141,
+      "step": 9090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621043027140246e-06,
+      "loss": 0.918,
+      "step": 9091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619167907764157e-06,
+      "loss": 0.9414,
+      "step": 9092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617292801797747e-06,
+      "loss": 0.8945,
+      "step": 9093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61541770930705e-06,
+      "loss": 0.918,
+      "step": 9094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61354263035809e-06,
+      "loss": 0.9297,
+      "step": 9095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611667565016893e-06,
+      "loss": 0.8516,
+      "step": 9096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609792513349477e-06,
+      "loss": 0.9766,
+      "step": 9097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607917475421868e-06,
+      "loss": 0.7852,
+      "step": 9098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606042451300092e-06,
+      "loss": 1.0078,
+      "step": 9099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604167441050173e-06,
+      "loss": 0.8242,
+      "step": 9100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602292444738124e-06,
+      "loss": 1.0078,
+      "step": 9101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.600417462429974e-06,
+      "loss": 0.7969,
+      "step": 9102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.598542494191742e-06,
+      "loss": 0.8359,
+      "step": 9103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596667540089451e-06,
+      "loss": 0.8633,
+      "step": 9104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594792600189116e-06,
+      "loss": 0.75,
+      "step": 9105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59291767455676e-06,
+      "loss": 0.9219,
+      "step": 9106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591042763258399e-06,
+      "loss": 0.8594,
+      "step": 9107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58916786636006e-06,
+      "loss": 0.9336,
+      "step": 9108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58729298392775e-06,
+      "loss": 0.8555,
+      "step": 9109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58541811602749e-06,
+      "loss": 1.0234,
+      "step": 9110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583543262725298e-06,
+      "loss": 0.8047,
+      "step": 9111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581668424087197e-06,
+      "loss": 0.9648,
+      "step": 9112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579793600179191e-06,
+      "loss": 0.8242,
+      "step": 9113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577918791067301e-06,
+      "loss": 0.8008,
+      "step": 9114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57604399681755e-06,
+      "loss": 0.8242,
+      "step": 9115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574169217495935e-06,
+      "loss": 0.9023,
+      "step": 9116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572294453168483e-06,
+      "loss": 0.9609,
+      "step": 9117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570419703901202e-06,
+      "loss": 1.0938,
+      "step": 9118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56854496976011e-06,
+      "loss": 0.8008,
+      "step": 9119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566670250811213e-06,
+      "loss": 0.8203,
+      "step": 9120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564795547120524e-06,
+      "loss": 0.9102,
+      "step": 9121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562920858754056e-06,
+      "loss": 0.8281,
+      "step": 9122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561046185777825e-06,
+      "loss": 0.8789,
+      "step": 9123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55917152825783e-06,
+      "loss": 0.8672,
+      "step": 9124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557296886260088e-06,
+      "loss": 0.7891,
+      "step": 9125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555422259850604e-06,
+      "loss": 0.7461,
+      "step": 9126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553547649095394e-06,
+      "loss": 0.9766,
+      "step": 9127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551673054060458e-06,
+      "loss": 0.9727,
+      "step": 9128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549798474811804e-06,
+      "loss": 0.9336,
+      "step": 9129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547923911415442e-06,
+      "loss": 0.9531,
+      "step": 9130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54604936393738e-06,
+      "loss": 0.7734,
+      "step": 9131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544174832443616e-06,
+      "loss": 0.9141,
+      "step": 9132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542300317000161e-06,
+      "loss": 0.7969,
+      "step": 9133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54042581767302e-06,
+      "loss": 0.9062,
+      "step": 9134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538551334528195e-06,
+      "loss": 0.9844,
+      "step": 9135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53667686763169e-06,
+      "loss": 0.7852,
+      "step": 9136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534802417049507e-06,
+      "loss": 0.9219,
+      "step": 9137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53292798284765e-06,
+      "loss": 0.9883,
+      "step": 9138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531053565092121e-06,
+      "loss": 0.7891,
+      "step": 9139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529179163848918e-06,
+      "loss": 0.8672,
+      "step": 9140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.527304779184043e-06,
+      "loss": 0.8711,
+      "step": 9141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5254304111635e-06,
+      "loss": 0.7891,
+      "step": 9142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523556059853284e-06,
+      "loss": 0.9297,
+      "step": 9143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521681725319392e-06,
+      "loss": 0.8984,
+      "step": 9144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.519807407627829e-06,
+      "loss": 0.8594,
+      "step": 9145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517933106844585e-06,
+      "loss": 0.8125,
+      "step": 9146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.516058823035662e-06,
+      "loss": 0.8281,
+      "step": 9147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514184556267054e-06,
+      "loss": 0.8672,
+      "step": 9148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51231030660476e-06,
+      "loss": 0.7422,
+      "step": 9149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.510436074114771e-06,
+      "loss": 0.9219,
+      "step": 9150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.508561858863086e-06,
+      "loss": 0.8086,
+      "step": 9151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.506687660915695e-06,
+      "loss": 0.9961,
+      "step": 9152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504813480338594e-06,
+      "loss": 0.9062,
+      "step": 9153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.502939317197774e-06,
+      "loss": 0.793,
+      "step": 9154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501065171559228e-06,
+      "loss": 0.8203,
+      "step": 9155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499191043488951e-06,
+      "loss": 1.0078,
+      "step": 9156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497316933052926e-06,
+      "loss": 0.7656,
+      "step": 9157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495442840317149e-06,
+      "loss": 0.8516,
+      "step": 9158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49356876534761e-06,
+      "loss": 0.7812,
+      "step": 9159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491694708210298e-06,
+      "loss": 0.918,
+      "step": 9160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489820668971195e-06,
+      "loss": 0.8633,
+      "step": 9161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487946647696296e-06,
+      "loss": 0.8047,
+      "step": 9162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48607264445159e-06,
+      "loss": 0.8125,
+      "step": 9163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484198659303057e-06,
+      "loss": 0.8359,
+      "step": 9164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482324692316685e-06,
+      "loss": 0.8203,
+      "step": 9165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480450743558459e-06,
+      "loss": 0.8359,
+      "step": 9166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478576813094369e-06,
+      "loss": 0.7305,
+      "step": 9167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476702900990392e-06,
+      "loss": 0.8164,
+      "step": 9168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474829007312512e-06,
+      "loss": 0.8398,
+      "step": 9169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472955132126715e-06,
+      "loss": 0.9297,
+      "step": 9170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.471081275498982e-06,
+      "loss": 0.875,
+      "step": 9171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469207437495292e-06,
+      "loss": 0.875,
+      "step": 9172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.467333618181628e-06,
+      "loss": 0.8242,
+      "step": 9173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465459817623972e-06,
+      "loss": 0.8125,
+      "step": 9174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463586035888302e-06,
+      "loss": 0.9219,
+      "step": 9175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.461712273040592e-06,
+      "loss": 0.9492,
+      "step": 9176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459838529146825e-06,
+      "loss": 0.8438,
+      "step": 9177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457964804272979e-06,
+      "loss": 0.8789,
+      "step": 9178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45609109848503e-06,
+      "loss": 0.8984,
+      "step": 9179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454217411848949e-06,
+      "loss": 0.8125,
+      "step": 9180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452343744430718e-06,
+      "loss": 0.8438,
+      "step": 9181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450470096296311e-06,
+      "loss": 0.875,
+      "step": 9182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448596467511699e-06,
+      "loss": 0.9219,
+      "step": 9183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.446722858142856e-06,
+      "loss": 0.875,
+      "step": 9184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444849268255755e-06,
+      "loss": 0.9297,
+      "step": 9185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442975697916372e-06,
+      "loss": 0.9688,
+      "step": 9186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441102147190671e-06,
+      "loss": 0.8047,
+      "step": 9187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439228616144626e-06,
+      "loss": 0.8828,
+      "step": 9188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437355104844207e-06,
+      "loss": 0.8633,
+      "step": 9189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43548161335539e-06,
+      "loss": 0.6719,
+      "step": 9190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43360814174413e-06,
+      "loss": 0.9844,
+      "step": 9191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431734690076403e-06,
+      "loss": 0.9375,
+      "step": 9192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.429861258418174e-06,
+      "loss": 0.8867,
+      "step": 9193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427987846835417e-06,
+      "loss": 0.9258,
+      "step": 9194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426114455394085e-06,
+      "loss": 1.0,
+      "step": 9195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42424108416015e-06,
+      "loss": 0.8594,
+      "step": 9196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422367733199574e-06,
+      "loss": 0.9141,
+      "step": 9197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420494402578327e-06,
+      "loss": 0.8516,
+      "step": 9198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418621092362364e-06,
+      "loss": 1.0234,
+      "step": 9199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41674780261765e-06,
+      "loss": 0.8477,
+      "step": 9200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414874533410143e-06,
+      "loss": 0.8125,
+      "step": 9201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413001284805813e-06,
+      "loss": 0.8789,
+      "step": 9202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41112805687061e-06,
+      "loss": 0.8242,
+      "step": 9203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409254849670498e-06,
+      "loss": 0.8516,
+      "step": 9204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407381663271435e-06,
+      "loss": 0.7891,
+      "step": 9205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.405508497739382e-06,
+      "loss": 0.8164,
+      "step": 9206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.403635353140289e-06,
+      "loss": 0.8984,
+      "step": 9207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401762229540116e-06,
+      "loss": 0.8633,
+      "step": 9208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.39988912700482e-06,
+      "loss": 0.8438,
+      "step": 9209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398016045600355e-06,
+      "loss": 0.9141,
+      "step": 9210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.396142985392674e-06,
+      "loss": 0.7695,
+      "step": 9211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.394269946447728e-06,
+      "loss": 0.8906,
+      "step": 9212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392396928831474e-06,
+      "loss": 0.9453,
+      "step": 9213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.390523932609866e-06,
+      "loss": 0.8906,
+      "step": 9214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388650957848847e-06,
+      "loss": 0.8867,
+      "step": 9215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386778004614372e-06,
+      "loss": 0.8398,
+      "step": 9216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38490507297239e-06,
+      "loss": 0.7305,
+      "step": 9217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383032162988854e-06,
+      "loss": 1.125,
+      "step": 9218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381159274729704e-06,
+      "loss": 0.9336,
+      "step": 9219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379286408260892e-06,
+      "loss": 0.9102,
+      "step": 9220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377413563648364e-06,
+      "loss": 0.9258,
+      "step": 9221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37554074095807e-06,
+      "loss": 0.9219,
+      "step": 9222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373667940255946e-06,
+      "loss": 0.8906,
+      "step": 9223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37179516160794e-06,
+      "loss": 0.8594,
+      "step": 9224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369922405079999e-06,
+      "loss": 0.9102,
+      "step": 9225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368049670738064e-06,
+      "loss": 0.7578,
+      "step": 9226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366176958648074e-06,
+      "loss": 0.8945,
+      "step": 9227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364304268875971e-06,
+      "loss": 0.8516,
+      "step": 9228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.3624316014877e-06,
+      "loss": 0.8633,
+      "step": 9229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360558956549193e-06,
+      "loss": 0.8945,
+      "step": 9230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358686334126392e-06,
+      "loss": 0.7695,
+      "step": 9231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356813734285236e-06,
+      "loss": 0.9141,
+      "step": 9232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.354941157091665e-06,
+      "loss": 0.9258,
+      "step": 9233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.353068602611608e-06,
+      "loss": 0.8008,
+      "step": 9234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.351196070911005e-06,
+      "loss": 0.8672,
+      "step": 9235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34932356205579e-06,
+      "loss": 0.8672,
+      "step": 9236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.3474510761119e-06,
+      "loss": 0.7734,
+      "step": 9237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345578613145262e-06,
+      "loss": 1.0391,
+      "step": 9238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34370617322181e-06,
+      "loss": 0.8438,
+      "step": 9239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.341833756407479e-06,
+      "loss": 0.9336,
+      "step": 9240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339961362768201e-06,
+      "loss": 0.875,
+      "step": 9241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338088992369898e-06,
+      "loss": 0.8008,
+      "step": 9242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336216645278504e-06,
+      "loss": 0.9219,
+      "step": 9243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334344321559946e-06,
+      "loss": 0.9648,
+      "step": 9244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332472021280157e-06,
+      "loss": 0.9375,
+      "step": 9245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330599744505054e-06,
+      "loss": 0.8125,
+      "step": 9246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.328727491300568e-06,
+      "loss": 0.8867,
+      "step": 9247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326855261732624e-06,
+      "loss": 0.8242,
+      "step": 9248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.324983055867149e-06,
+      "loss": 0.918,
+      "step": 9249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32311087377006e-06,
+      "loss": 0.8086,
+      "step": 9250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321238715507281e-06,
+      "loss": 0.8398,
+      "step": 9251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.319366581144736e-06,
+      "loss": 0.9609,
+      "step": 9252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317494470748347e-06,
+      "loss": 0.9336,
+      "step": 9253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.315622384384029e-06,
+      "loss": 1.0078,
+      "step": 9254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313750322117703e-06,
+      "loss": 1.0625,
+      "step": 9255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311878284015287e-06,
+      "loss": 0.875,
+      "step": 9256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310006270142704e-06,
+      "loss": 0.9102,
+      "step": 9257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30813428056586e-06,
+      "loss": 0.8008,
+      "step": 9258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306262315350677e-06,
+      "loss": 0.793,
+      "step": 9259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30439037456307e-06,
+      "loss": 0.8555,
+      "step": 9260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302518458268955e-06,
+      "loss": 0.8828,
+      "step": 9261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300646566534239e-06,
+      "loss": 0.9297,
+      "step": 9262
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298774699424836e-06,
+      "loss": 0.9141,
+      "step": 9263
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296902857006658e-06,
+      "loss": 0.8867,
+      "step": 9264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29503103934562e-06,
+      "loss": 0.8906,
+      "step": 9265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293159246507625e-06,
+      "loss": 0.8711,
+      "step": 9266
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291287478558583e-06,
+      "loss": 0.9258,
+      "step": 9267
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289415735564405e-06,
+      "loss": 0.8086,
+      "step": 9268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287544017591e-06,
+      "loss": 0.8828,
+      "step": 9269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285672324704265e-06,
+      "loss": 0.9375,
+      "step": 9270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28380065697011e-06,
+      "loss": 0.9648,
+      "step": 9271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28192901445444e-06,
+      "loss": 0.8828,
+      "step": 9272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280057397223162e-06,
+      "loss": 0.8555,
+      "step": 9273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.278185805342171e-06,
+      "loss": 0.8438,
+      "step": 9274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276314238877373e-06,
+      "loss": 0.8203,
+      "step": 9275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274442697894667e-06,
+      "loss": 0.918,
+      "step": 9276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272571182459959e-06,
+      "loss": 1.0156,
+      "step": 9277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270699692639137e-06,
+      "loss": 0.9102,
+      "step": 9278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268828228498105e-06,
+      "loss": 0.8008,
+      "step": 9279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266956790102761e-06,
+      "loss": 0.7422,
+      "step": 9280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265085377519004e-06,
+      "loss": 0.8906,
+      "step": 9281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263213990812721e-06,
+      "loss": 0.8477,
+      "step": 9282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261342630049812e-06,
+      "loss": 0.8125,
+      "step": 9283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259471295296174e-06,
+      "loss": 1.0547,
+      "step": 9284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257599986617688e-06,
+      "loss": 0.957,
+      "step": 9285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255728704080257e-06,
+      "loss": 0.7578,
+      "step": 9286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253857447749765e-06,
+      "loss": 0.9219,
+      "step": 9287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.251986217692108e-06,
+      "loss": 0.8789,
+      "step": 9288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25011501397317e-06,
+      "loss": 0.9141,
+      "step": 9289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248243836658835e-06,
+      "loss": 0.9375,
+      "step": 9290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246372685815e-06,
+      "loss": 0.8906,
+      "step": 9291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244501561507548e-06,
+      "loss": 0.8867,
+      "step": 9292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242630463802359e-06,
+      "loss": 0.8984,
+      "step": 9293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240759392765321e-06,
+      "loss": 0.9922,
+      "step": 9294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238888348462317e-06,
+      "loss": 0.8359,
+      "step": 9295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237017330959233e-06,
+      "loss": 0.8242,
+      "step": 9296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.235146340321943e-06,
+      "loss": 0.793,
+      "step": 9297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23327537661633e-06,
+      "loss": 0.7812,
+      "step": 9298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231404439908278e-06,
+      "loss": 0.8867,
+      "step": 9299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229533530263664e-06,
+      "loss": 0.8828,
+      "step": 9300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22766264774836e-06,
+      "loss": 0.8594,
+      "step": 9301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225791792428248e-06,
+      "loss": 0.8789,
+      "step": 9302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223920964369202e-06,
+      "loss": 0.8672,
+      "step": 9303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2220501636371e-06,
+      "loss": 0.9258,
+      "step": 9304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22017939029781e-06,
+      "loss": 0.8945,
+      "step": 9305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218308644417206e-06,
+      "loss": 0.8633,
+      "step": 9306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216437926061165e-06,
+      "loss": 0.8047,
+      "step": 9307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214567235295556e-06,
+      "loss": 0.832,
+      "step": 9308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212696572186243e-06,
+      "loss": 0.8164,
+      "step": 9309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210825936799099e-06,
+      "loss": 0.9297,
+      "step": 9310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208955329199995e-06,
+      "loss": 0.7852,
+      "step": 9311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207084749454796e-06,
+      "loss": 0.9688,
+      "step": 9312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205214197629364e-06,
+      "loss": 0.7656,
+      "step": 9313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203343673789568e-06,
+      "loss": 0.8633,
+      "step": 9314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201473178001273e-06,
+      "loss": 0.8516,
+      "step": 9315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19960271033034e-06,
+      "loss": 0.9336,
+      "step": 9316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197732270842631e-06,
+      "loss": 0.8672,
+      "step": 9317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195861859604008e-06,
+      "loss": 0.875,
+      "step": 9318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193991476680328e-06,
+      "loss": 0.957,
+      "step": 9319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192121122137455e-06,
+      "loss": 0.9414,
+      "step": 9320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190250796041244e-06,
+      "loss": 0.9375,
+      "step": 9321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188380498457555e-06,
+      "loss": 0.8789,
+      "step": 9322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.186510229452237e-06,
+      "loss": 1.0078,
+      "step": 9323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184639989091154e-06,
+      "loss": 0.8906,
+      "step": 9324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.182769777440153e-06,
+      "loss": 0.918,
+      "step": 9325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.180899594565092e-06,
+      "loss": 0.8594,
+      "step": 9326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17902944053182e-06,
+      "loss": 0.9219,
+      "step": 9327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17715931540619e-06,
+      "loss": 0.9414,
+      "step": 9328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175289219254051e-06,
+      "loss": 0.9766,
+      "step": 9329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17341915214125e-06,
+      "loss": 0.8906,
+      "step": 9330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171549114133637e-06,
+      "loss": 0.9023,
+      "step": 9331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16967910529706e-06,
+      "loss": 0.8281,
+      "step": 9332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.167809125697365e-06,
+      "loss": 0.7812,
+      "step": 9333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165939175400394e-06,
+      "loss": 0.8281,
+      "step": 9334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16406925447199e-06,
+      "loss": 0.8828,
+      "step": 9335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162199362978001e-06,
+      "loss": 1.0547,
+      "step": 9336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.160329500984268e-06,
+      "loss": 0.9102,
+      "step": 9337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158459668556625e-06,
+      "loss": 0.7617,
+      "step": 9338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156589865760922e-06,
+      "loss": 0.793,
+      "step": 9339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154720092662986e-06,
+      "loss": 0.7969,
+      "step": 9340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152850349328666e-06,
+      "loss": 0.9766,
+      "step": 9341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150980635823789e-06,
+      "loss": 0.8398,
+      "step": 9342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.149110952214198e-06,
+      "loss": 0.7812,
+      "step": 9343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147241298565724e-06,
+      "loss": 0.8398,
+      "step": 9344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145371674944197e-06,
+      "loss": 0.7773,
+      "step": 9345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.143502081415454e-06,
+      "loss": 0.8672,
+      "step": 9346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.141632518045327e-06,
+      "loss": 0.8633,
+      "step": 9347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139762984899644e-06,
+      "loss": 0.8828,
+      "step": 9348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137893482044233e-06,
+      "loss": 0.7969,
+      "step": 9349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136024009544924e-06,
+      "loss": 0.8672,
+      "step": 9350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134154567467545e-06,
+      "loss": 0.9258,
+      "step": 9351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132285155877921e-06,
+      "loss": 0.7656,
+      "step": 9352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.130415774841875e-06,
+      "loss": 0.8242,
+      "step": 9353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12854642442523e-06,
+      "loss": 0.8477,
+      "step": 9354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126677104693813e-06,
+      "loss": 0.8086,
+      "step": 9355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.124807815713445e-06,
+      "loss": 0.9219,
+      "step": 9356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122938557549943e-06,
+      "loss": 0.9648,
+      "step": 9357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121069330269129e-06,
+      "loss": 0.7656,
+      "step": 9358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119200133936825e-06,
+      "loss": 0.707,
+      "step": 9359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11733096861884e-06,
+      "loss": 0.8984,
+      "step": 9360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115461834380993e-06,
+      "loss": 1.0312,
+      "step": 9361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113592731289102e-06,
+      "loss": 0.8438,
+      "step": 9362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.111723659408984e-06,
+      "loss": 0.8398,
+      "step": 9363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109854618806442e-06,
+      "loss": 0.7383,
+      "step": 9364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107985609547293e-06,
+      "loss": 0.8945,
+      "step": 9365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10611663169735e-06,
+      "loss": 0.8516,
+      "step": 9366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104247685322423e-06,
+      "loss": 0.8594,
+      "step": 9367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102378770488314e-06,
+      "loss": 0.9531,
+      "step": 9368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100509887260834e-06,
+      "loss": 0.9023,
+      "step": 9369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098641035705791e-06,
+      "loss": 1.0,
+      "step": 9370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096772215888992e-06,
+      "loss": 0.7461,
+      "step": 9371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094903427876235e-06,
+      "loss": 0.8438,
+      "step": 9372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093034671733325e-06,
+      "loss": 0.8516,
+      "step": 9373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091165947526064e-06,
+      "loss": 1.0469,
+      "step": 9374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.089297255320257e-06,
+      "loss": 0.8984,
+      "step": 9375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087428595181697e-06,
+      "loss": 0.8281,
+      "step": 9376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085559967176185e-06,
+      "loss": 0.918,
+      "step": 9377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083691371369518e-06,
+      "loss": 0.7461,
+      "step": 9378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081822807827496e-06,
+      "loss": 0.7422,
+      "step": 9379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079954276615907e-06,
+      "loss": 0.8086,
+      "step": 9380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07808577780055e-06,
+      "loss": 0.9727,
+      "step": 9381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076217311447214e-06,
+      "loss": 0.8906,
+      "step": 9382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074348877621697e-06,
+      "loss": 0.9766,
+      "step": 9383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072480476389782e-06,
+      "loss": 0.8828,
+      "step": 9384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070612107817262e-06,
+      "loss": 0.7383,
+      "step": 9385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068743771969925e-06,
+      "loss": 0.7969,
+      "step": 9386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.066875468913561e-06,
+      "loss": 0.7695,
+      "step": 9387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065007198713948e-06,
+      "loss": 0.9648,
+      "step": 9388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.063138961436875e-06,
+      "loss": 0.9453,
+      "step": 9389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061270757148125e-06,
+      "loss": 0.9219,
+      "step": 9390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059402585913486e-06,
+      "loss": 1.0312,
+      "step": 9391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.057534447798732e-06,
+      "loss": 0.9609,
+      "step": 9392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055666342869642e-06,
+      "loss": 0.8633,
+      "step": 9393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053798271192e-06,
+      "loss": 0.8906,
+      "step": 9394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051930232831586e-06,
+      "loss": 0.7773,
+      "step": 9395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050062227854168e-06,
+      "loss": 0.9727,
+      "step": 9396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048194256325524e-06,
+      "loss": 0.9883,
+      "step": 9397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046326318311437e-06,
+      "loss": 0.9453,
+      "step": 9398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044458413877665e-06,
+      "loss": 0.8984,
+      "step": 9399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04259054308999e-06,
+      "loss": 0.8828,
+      "step": 9400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04072270601418e-06,
+      "loss": 0.9688,
+      "step": 9401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03885490271601e-06,
+      "loss": 0.8516,
+      "step": 9402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036987133261237e-06,
+      "loss": 0.8242,
+      "step": 9403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035119397715634e-06,
+      "loss": 0.7969,
+      "step": 9404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033251696144966e-06,
+      "loss": 0.8984,
+      "step": 9405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.031384028615005e-06,
+      "loss": 0.8945,
+      "step": 9406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029516395191501e-06,
+      "loss": 0.8477,
+      "step": 9407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027648795940226e-06,
+      "loss": 0.9375,
+      "step": 9408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025781230926937e-06,
+      "loss": 0.9023,
+      "step": 9409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.0239137002174e-06,
+      "loss": 0.9453,
+      "step": 9410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022046203877361e-06,
+      "loss": 0.9805,
+      "step": 9411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020178741972588e-06,
+      "loss": 0.8438,
+      "step": 9412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.018311314568836e-06,
+      "loss": 0.7812,
+      "step": 9413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01644392173186e-06,
+      "loss": 0.8164,
+      "step": 9414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014576563527407e-06,
+      "loss": 0.8828,
+      "step": 9415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012709240021237e-06,
+      "loss": 1.0703,
+      "step": 9416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010841951279099e-06,
+      "loss": 0.8516,
+      "step": 9417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008974697366745e-06,
+      "loss": 0.8438,
+      "step": 9418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007107478349919e-06,
+      "loss": 0.8984,
+      "step": 9419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005240294294373e-06,
+      "loss": 1.0234,
+      "step": 9420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003373145265851e-06,
+      "loss": 0.9609,
+      "step": 9421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.001506031330104e-06,
+      "loss": 0.9453,
+      "step": 9422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999638952552866e-06,
+      "loss": 0.8984,
+      "step": 9423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997771908999887e-06,
+      "loss": 0.8086,
+      "step": 9424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995904900736905e-06,
+      "loss": 0.8789,
+      "step": 9425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994037927829667e-06,
+      "loss": 0.9766,
+      "step": 9426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9921709903439e-06,
+      "loss": 0.8906,
+      "step": 9427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99030408834535e-06,
+      "loss": 0.8398,
+      "step": 9428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988437221899754e-06,
+      "loss": 0.9141,
+      "step": 9429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986570391072848e-06,
+      "loss": 0.832,
+      "step": 9430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984703595930359e-06,
+      "loss": 0.9375,
+      "step": 9431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982836836538023e-06,
+      "loss": 0.9922,
+      "step": 9432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.980970112961573e-06,
+      "loss": 0.8984,
+      "step": 9433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979103425266744e-06,
+      "loss": 0.7422,
+      "step": 9434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.977236773519254e-06,
+      "loss": 0.7734,
+      "step": 9435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.975370157784837e-06,
+      "loss": 0.7891,
+      "step": 9436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97350357812922e-06,
+      "loss": 0.8125,
+      "step": 9437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97163703461813e-06,
+      "loss": 0.9219,
+      "step": 9438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969770527317283e-06,
+      "loss": 0.8398,
+      "step": 9439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.967904056292407e-06,
+      "loss": 0.7734,
+      "step": 9440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966037621609223e-06,
+      "loss": 0.9453,
+      "step": 9441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964171223333456e-06,
+      "loss": 0.8398,
+      "step": 9442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962304861530812e-06,
+      "loss": 0.7969,
+      "step": 9443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960438536267018e-06,
+      "loss": 0.9492,
+      "step": 9444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95857224760779e-06,
+      "loss": 0.7773,
+      "step": 9445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956705995618842e-06,
+      "loss": 0.8125,
+      "step": 9446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954839780365885e-06,
+      "loss": 0.7969,
+      "step": 9447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952973601914631e-06,
+      "loss": 0.9844,
+      "step": 9448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951107460330796e-06,
+      "loss": 0.8047,
+      "step": 9449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94924135568009e-06,
+      "loss": 1.0312,
+      "step": 9450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947375288028212e-06,
+      "loss": 0.8711,
+      "step": 9451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945509257440877e-06,
+      "loss": 0.8438,
+      "step": 9452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.943643263983794e-06,
+      "loss": 0.8477,
+      "step": 9453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941777307722657e-06,
+      "loss": 0.8828,
+      "step": 9454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939911388723175e-06,
+      "loss": 0.8906,
+      "step": 9455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93804550705105e-06,
+      "loss": 0.8906,
+      "step": 9456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936179662771987e-06,
+      "loss": 0.8477,
+      "step": 9457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934313855951676e-06,
+      "loss": 0.8516,
+      "step": 9458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93244808665582e-06,
+      "loss": 0.9453,
+      "step": 9459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930582354950115e-06,
+      "loss": 0.7812,
+      "step": 9460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928716660900261e-06,
+      "loss": 0.8438,
+      "step": 9461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926851004571942e-06,
+      "loss": 0.7969,
+      "step": 9462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924985386030858e-06,
+      "loss": 0.8594,
+      "step": 9463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923119805342697e-06,
+      "loss": 0.9531,
+      "step": 9464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921254262573154e-06,
+      "loss": 0.9922,
+      "step": 9465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919388757787911e-06,
+      "loss": 1.0156,
+      "step": 9466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917523291052657e-06,
+      "loss": 0.8945,
+      "step": 9467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915657862433081e-06,
+      "loss": 0.7812,
+      "step": 9468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913792471994869e-06,
+      "loss": 0.8242,
+      "step": 9469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911927119803697e-06,
+      "loss": 0.9336,
+      "step": 9470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910061805925252e-06,
+      "loss": 0.9375,
+      "step": 9471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908196530425213e-06,
+      "loss": 0.8906,
+      "step": 9472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.906331293369263e-06,
+      "loss": 0.8086,
+      "step": 9473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904466094823075e-06,
+      "loss": 0.8672,
+      "step": 9474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902600934852326e-06,
+      "loss": 0.8398,
+      "step": 9475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.900735813522694e-06,
+      "loss": 0.8633,
+      "step": 9476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.898870730899851e-06,
+      "loss": 0.9375,
+      "step": 9477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897005687049469e-06,
+      "loss": 0.9141,
+      "step": 9478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89514068203722e-06,
+      "loss": 0.8516,
+      "step": 9479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893275715928774e-06,
+      "loss": 0.8086,
+      "step": 9480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8914107887898e-06,
+      "loss": 0.9492,
+      "step": 9481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88954590068596e-06,
+      "loss": 0.8828,
+      "step": 9482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.887681051682924e-06,
+      "loss": 0.9414,
+      "step": 9483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88581624184636e-06,
+      "loss": 1.0078,
+      "step": 9484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883951471241924e-06,
+      "loss": 0.8008,
+      "step": 9485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882086739935278e-06,
+      "loss": 0.8789,
+      "step": 9486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880222047992086e-06,
+      "loss": 0.7969,
+      "step": 9487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.878357395478005e-06,
+      "loss": 0.8516,
+      "step": 9488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876492782458691e-06,
+      "loss": 0.9102,
+      "step": 9489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8746282089998e-06,
+      "loss": 0.8672,
+      "step": 9490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872763675166989e-06,
+      "loss": 0.8203,
+      "step": 9491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870899181025909e-06,
+      "loss": 0.8906,
+      "step": 9492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869034726642212e-06,
+      "loss": 0.7852,
+      "step": 9493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867170312081546e-06,
+      "loss": 0.9062,
+      "step": 9494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865305937409567e-06,
+      "loss": 0.9961,
+      "step": 9495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863441602691914e-06,
+      "loss": 0.9102,
+      "step": 9496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86157730799424e-06,
+      "loss": 0.9922,
+      "step": 9497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859713053382185e-06,
+      "loss": 0.9102,
+      "step": 9498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857848838921395e-06,
+      "loss": 0.7539,
+      "step": 9499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855984664677508e-06,
+      "loss": 0.9453,
+      "step": 9500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85412053071617e-06,
+      "loss": 0.9219,
+      "step": 9501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852256437103019e-06,
+      "loss": 0.8359,
+      "step": 9502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850392383903688e-06,
+      "loss": 0.8125,
+      "step": 9503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848528371183815e-06,
+      "loss": 0.8828,
+      "step": 9504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84666439900904e-06,
+      "loss": 0.9297,
+      "step": 9505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84480046744499e-06,
+      "loss": 0.875,
+      "step": 9506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842936576557299e-06,
+      "loss": 0.8242,
+      "step": 9507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841072726411597e-06,
+      "loss": 0.9375,
+      "step": 9508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839208917073516e-06,
+      "loss": 1.0078,
+      "step": 9509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837345148608679e-06,
+      "loss": 0.7578,
+      "step": 9510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.835481421082714e-06,
+      "loss": 0.9414,
+      "step": 9511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.833617734561249e-06,
+      "loss": 0.9922,
+      "step": 9512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831754089109902e-06,
+      "loss": 0.9102,
+      "step": 9513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829890484794299e-06,
+      "loss": 0.9414,
+      "step": 9514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828026921680057e-06,
+      "loss": 0.7461,
+      "step": 9515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8261633998328e-06,
+      "loss": 0.7383,
+      "step": 9516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82429991931814e-06,
+      "loss": 0.8281,
+      "step": 9517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822436480201693e-06,
+      "loss": 0.8203,
+      "step": 9518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820573082549076e-06,
+      "loss": 0.8086,
+      "step": 9519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.818709726425905e-06,
+      "loss": 0.9375,
+      "step": 9520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816846411897788e-06,
+      "loss": 0.7695,
+      "step": 9521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814983139030332e-06,
+      "loss": 0.7578,
+      "step": 9522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.813119907889152e-06,
+      "loss": 0.8477,
+      "step": 9523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811256718539853e-06,
+      "loss": 1.0156,
+      "step": 9524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80939357104804e-06,
+      "loss": 0.8594,
+      "step": 9525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807530465479316e-06,
+      "loss": 0.9492,
+      "step": 9526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805667401899286e-06,
+      "loss": 0.8633,
+      "step": 9527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.803804380373553e-06,
+      "loss": 0.8672,
+      "step": 9528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801941400967715e-06,
+      "loss": 0.8945,
+      "step": 9529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800078463747366e-06,
+      "loss": 0.8398,
+      "step": 9530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798215568778109e-06,
+      "loss": 0.832,
+      "step": 9531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79635271612554e-06,
+      "loss": 0.9102,
+      "step": 9532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.794489905855247e-06,
+      "loss": 0.8594,
+      "step": 9533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792627138032826e-06,
+      "loss": 0.9805,
+      "step": 9534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790764412723865e-06,
+      "loss": 0.8359,
+      "step": 9535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788901729993964e-06,
+      "loss": 0.9062,
+      "step": 9536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787039089908696e-06,
+      "loss": 0.9023,
+      "step": 9537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785176492533656e-06,
+      "loss": 0.8867,
+      "step": 9538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.783313937934428e-06,
+      "loss": 0.9219,
+      "step": 9539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781451426176597e-06,
+      "loss": 0.8047,
+      "step": 9540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779588957325741e-06,
+      "loss": 0.9414,
+      "step": 9541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77772653144744e-06,
+      "loss": 0.8438,
+      "step": 9542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775864148607275e-06,
+      "loss": 0.9336,
+      "step": 9543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774001808870829e-06,
+      "loss": 0.8906,
+      "step": 9544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.772139512303669e-06,
+      "loss": 0.7773,
+      "step": 9545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770277258971369e-06,
+      "loss": 0.9023,
+      "step": 9546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768415048939509e-06,
+      "loss": 0.7109,
+      "step": 9547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.766552882273659e-06,
+      "loss": 0.8203,
+      "step": 9548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.764690759039382e-06,
+      "loss": 0.8867,
+      "step": 9549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.762828679302252e-06,
+      "loss": 0.7891,
+      "step": 9550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760966643127832e-06,
+      "loss": 0.793,
+      "step": 9551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759104650581695e-06,
+      "loss": 1.0156,
+      "step": 9552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757242701729397e-06,
+      "loss": 0.9883,
+      "step": 9553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7553807966365e-06,
+      "loss": 0.9141,
+      "step": 9554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753518935368568e-06,
+      "loss": 0.8867,
+      "step": 9555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751657117991163e-06,
+      "loss": 0.8242,
+      "step": 9556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749795344569832e-06,
+      "loss": 0.7969,
+      "step": 9557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74793361517014e-06,
+      "loss": 0.8633,
+      "step": 9558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746071929857636e-06,
+      "loss": 0.8828,
+      "step": 9559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744210288697882e-06,
+      "loss": 0.9883,
+      "step": 9560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742348691756416e-06,
+      "loss": 0.8828,
+      "step": 9561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740487139098798e-06,
+      "loss": 0.8477,
+      "step": 9562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73862563079057e-06,
+      "loss": 0.9297,
+      "step": 9563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736764166897285e-06,
+      "loss": 0.9883,
+      "step": 9564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734902747484482e-06,
+      "loss": 0.9297,
+      "step": 9565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733041372617705e-06,
+      "loss": 0.9297,
+      "step": 9566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731180042362502e-06,
+      "loss": 0.9062,
+      "step": 9567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.729318756784404e-06,
+      "loss": 0.7695,
+      "step": 9568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727457515948956e-06,
+      "loss": 0.918,
+      "step": 9569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725596319921692e-06,
+      "loss": 0.8203,
+      "step": 9570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723735168768154e-06,
+      "loss": 0.8789,
+      "step": 9571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721874062553867e-06,
+      "loss": 0.8633,
+      "step": 9572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.720013001344367e-06,
+      "loss": 0.7578,
+      "step": 9573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718151985205187e-06,
+      "loss": 0.9453,
+      "step": 9574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.716291014201858e-06,
+      "loss": 1.0078,
+      "step": 9575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7144300883999e-06,
+      "loss": 0.8047,
+      "step": 9576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712569207864845e-06,
+      "loss": 0.9023,
+      "step": 9577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.710708372662214e-06,
+      "loss": 0.9961,
+      "step": 9578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.708847582857537e-06,
+      "loss": 0.8164,
+      "step": 9579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706986838516327e-06,
+      "loss": 0.8242,
+      "step": 9580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705126139704105e-06,
+      "loss": 0.9219,
+      "step": 9581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703265486486391e-06,
+      "loss": 0.8594,
+      "step": 9582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701404878928707e-06,
+      "loss": 0.9883,
+      "step": 9583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699544317096558e-06,
+      "loss": 0.918,
+      "step": 9584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69768380105546e-06,
+      "loss": 0.9141,
+      "step": 9585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.695823330870927e-06,
+      "loss": 0.8086,
+      "step": 9586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693962906608474e-06,
+      "loss": 0.8203,
+      "step": 9587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692102528333595e-06,
+      "loss": 0.832,
+      "step": 9588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690242196111809e-06,
+      "loss": 0.9219,
+      "step": 9589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688381910008616e-06,
+      "loss": 0.8984,
+      "step": 9590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686521670089528e-06,
+      "loss": 0.8398,
+      "step": 9591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.684661476420032e-06,
+      "loss": 0.8555,
+      "step": 9592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682801329065638e-06,
+      "loss": 0.9492,
+      "step": 9593
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680941228091843e-06,
+      "loss": 0.918,
+      "step": 9594
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679081173564148e-06,
+      "loss": 0.9414,
+      "step": 9595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67722116554804e-06,
+      "loss": 0.9805,
+      "step": 9596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675361204109018e-06,
+      "loss": 0.8125,
+      "step": 9597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673501289312572e-06,
+      "loss": 0.9297,
+      "step": 9598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6716414212242e-06,
+      "loss": 0.8828,
+      "step": 9599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669781599909377e-06,
+      "loss": 0.9609,
+      "step": 9600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6679218254336e-06,
+      "loss": 0.8672,
+      "step": 9601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666062097862352e-06,
+      "loss": 0.7891,
+      "step": 9602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664202417261121e-06,
+      "loss": 0.8125,
+      "step": 9603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.662342783695381e-06,
+      "loss": 0.8555,
+      "step": 9604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660483197230619e-06,
+      "loss": 0.8203,
+      "step": 9605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65862365793231e-06,
+      "loss": 0.8594,
+      "step": 9606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656764165865936e-06,
+      "loss": 0.7305,
+      "step": 9607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654904721096968e-06,
+      "loss": 0.9453,
+      "step": 9608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65304532369088e-06,
+      "loss": 0.9141,
+      "step": 9609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.651185973713145e-06,
+      "loss": 0.9414,
+      "step": 9610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649326671229241e-06,
+      "loss": 0.9141,
+      "step": 9611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647467416304625e-06,
+      "loss": 0.7695,
+      "step": 9612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645608209004769e-06,
+      "loss": 0.8359,
+      "step": 9613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64374904939514e-06,
+      "loss": 0.9375,
+      "step": 9614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641889937541205e-06,
+      "loss": 0.8203,
+      "step": 9615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640030873508418e-06,
+      "loss": 0.8594,
+      "step": 9616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638171857362243e-06,
+      "loss": 0.8828,
+      "step": 9617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63631288916814e-06,
+      "loss": 0.8711,
+      "step": 9618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634453968991571e-06,
+      "loss": 0.8086,
+      "step": 9619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632595096897983e-06,
+      "loss": 0.9102,
+      "step": 9620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.630736272952829e-06,
+      "loss": 0.875,
+      "step": 9621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628877497221572e-06,
+      "loss": 0.8047,
+      "step": 9622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627018769769651e-06,
+      "loss": 0.8086,
+      "step": 9623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.625160090662518e-06,
+      "loss": 1.0156,
+      "step": 9624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623301459965621e-06,
+      "loss": 0.8711,
+      "step": 9625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.621442877744409e-06,
+      "loss": 0.8203,
+      "step": 9626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619584344064318e-06,
+      "loss": 0.8633,
+      "step": 9627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.617725858990795e-06,
+      "loss": 0.8086,
+      "step": 9628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615867422589277e-06,
+      "loss": 0.9922,
+      "step": 9629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614009034925209e-06,
+      "loss": 0.9297,
+      "step": 9630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.612150696064019e-06,
+      "loss": 0.9258,
+      "step": 9631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610292406071145e-06,
+      "loss": 0.8203,
+      "step": 9632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60843416501202e-06,
+      "loss": 0.9219,
+      "step": 9633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606575972952084e-06,
+      "loss": 0.8164,
+      "step": 9634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604717829956754e-06,
+      "loss": 0.875,
+      "step": 9635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602859736091462e-06,
+      "loss": 0.9258,
+      "step": 9636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.601001691421638e-06,
+      "loss": 0.8672,
+      "step": 9637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599143696012707e-06,
+      "loss": 0.9297,
+      "step": 9638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597285749930085e-06,
+      "loss": 0.9023,
+      "step": 9639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5954278532392e-06,
+      "loss": 0.875,
+      "step": 9640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.593570006005467e-06,
+      "loss": 0.9648,
+      "step": 9641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59171220829431e-06,
+      "loss": 0.8867,
+      "step": 9642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589854460171136e-06,
+      "loss": 0.8906,
+      "step": 9643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587996761701363e-06,
+      "loss": 0.8906,
+      "step": 9644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586139112950408e-06,
+      "loss": 0.8555,
+      "step": 9645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584281513983678e-06,
+      "loss": 0.8711,
+      "step": 9646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58242396486658e-06,
+      "loss": 0.8945,
+      "step": 9647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58056646566452e-06,
+      "loss": 0.8672,
+      "step": 9648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57870901644291e-06,
+      "loss": 0.8633,
+      "step": 9649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 1.0391,
+      "step": 9650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57499426820264e-06,
+      "loss": 0.8594,
+      "step": 9651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573136969314782e-06,
+      "loss": 0.8516,
+      "step": 9652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571279720668977e-06,
+      "loss": 0.8047,
+      "step": 9653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569422522330619e-06,
+      "loss": 0.8125,
+      "step": 9654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5675653743651e-06,
+      "loss": 0.9336,
+      "step": 9655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565708276837816e-06,
+      "loss": 1.0312,
+      "step": 9656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56385122981416e-06,
+      "loss": 0.8828,
+      "step": 9657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56199423335952e-06,
+      "loss": 0.7734,
+      "step": 9658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56013728753928e-06,
+      "loss": 0.9141,
+      "step": 9659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558280392418832e-06,
+      "loss": 0.9219,
+      "step": 9660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556423548063557e-06,
+      "loss": 0.7734,
+      "step": 9661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55456675453884e-06,
+      "loss": 0.7188,
+      "step": 9662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552710011910055e-06,
+      "loss": 0.8398,
+      "step": 9663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550853320242588e-06,
+      "loss": 0.793,
+      "step": 9664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548996679601811e-06,
+      "loss": 0.8047,
+      "step": 9665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547140090053103e-06,
+      "loss": 0.8594,
+      "step": 9666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545283551661835e-06,
+      "loss": 0.832,
+      "step": 9667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543427064493379e-06,
+      "loss": 0.8086,
+      "step": 9668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541570628613104e-06,
+      "loss": 0.8008,
+      "step": 9669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539714244086381e-06,
+      "loss": 0.918,
+      "step": 9670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.537857910978574e-06,
+      "loss": 0.918,
+      "step": 9671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536001629355047e-06,
+      "loss": 0.9609,
+      "step": 9672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53414539928116e-06,
+      "loss": 0.9609,
+      "step": 9673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532289220822282e-06,
+      "loss": 0.8906,
+      "step": 9674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.530433094043765e-06,
+      "loss": 0.8516,
+      "step": 9675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528577019010965e-06,
+      "loss": 0.8594,
+      "step": 9676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526720995789243e-06,
+      "loss": 0.9258,
+      "step": 9677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.524865024443947e-06,
+      "loss": 0.8516,
+      "step": 9678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523009105040435e-06,
+      "loss": 0.8359,
+      "step": 9679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521153237644049e-06,
+      "loss": 0.8477,
+      "step": 9680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.519297422320143e-06,
+      "loss": 0.9062,
+      "step": 9681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51744165913406e-06,
+      "loss": 0.8828,
+      "step": 9682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515585948151147e-06,
+      "loss": 0.7695,
+      "step": 9683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513730289436742e-06,
+      "loss": 1.0781,
+      "step": 9684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511874683056189e-06,
+      "loss": 0.918,
+      "step": 9685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510019129074826e-06,
+      "loss": 0.8203,
+      "step": 9686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508163627557993e-06,
+      "loss": 0.8555,
+      "step": 9687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506308178571017e-06,
+      "loss": 0.8789,
+      "step": 9688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50445278217924e-06,
+      "loss": 0.8125,
+      "step": 9689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50259743844799e-06,
+      "loss": 0.8242,
+      "step": 9690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500742147442593e-06,
+      "loss": 0.7812,
+      "step": 9691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49888690922838e-06,
+      "loss": 0.8516,
+      "step": 9692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49703172387068e-06,
+      "loss": 0.7734,
+      "step": 9693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.495176591434813e-06,
+      "loss": 0.9062,
+      "step": 9694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4933215119861e-06,
+      "loss": 0.9336,
+      "step": 9695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491466485589862e-06,
+      "loss": 0.8516,
+      "step": 9696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489611512311421e-06,
+      "loss": 0.8125,
+      "step": 9697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48775659221609e-06,
+      "loss": 0.8516,
+      "step": 9698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485901725369182e-06,
+      "loss": 1.0,
+      "step": 9699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484046911836012e-06,
+      "loss": 0.875,
+      "step": 9700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482192151681894e-06,
+      "loss": 0.8594,
+      "step": 9701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480337444972134e-06,
+      "loss": 0.9062,
+      "step": 9702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478482791772036e-06,
+      "loss": 0.793,
+      "step": 9703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476628192146908e-06,
+      "loss": 0.8359,
+      "step": 9704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474773646162057e-06,
+      "loss": 0.8555,
+      "step": 9705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472919153882778e-06,
+      "loss": 0.8555,
+      "step": 9706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.471064715374373e-06,
+      "loss": 0.9688,
+      "step": 9707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46921033070214e-06,
+      "loss": 0.9336,
+      "step": 9708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46735599993138e-06,
+      "loss": 0.8672,
+      "step": 9709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465501723127375e-06,
+      "loss": 0.8516,
+      "step": 9710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463647500355425e-06,
+      "loss": 0.8086,
+      "step": 9711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46179333168082e-06,
+      "loss": 0.6992,
+      "step": 9712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45993921716885e-06,
+      "loss": 0.6992,
+      "step": 9713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458085156884791e-06,
+      "loss": 0.9258,
+      "step": 9714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456231150893938e-06,
+      "loss": 0.9414,
+      "step": 9715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454377199261567e-06,
+      "loss": 0.957,
+      "step": 9716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.452523302052968e-06,
+      "loss": 0.9141,
+      "step": 9717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450669459333406e-06,
+      "loss": 0.7969,
+      "step": 9718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448815671168167e-06,
+      "loss": 1.0469,
+      "step": 9719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.446961937622522e-06,
+      "loss": 0.9258,
+      "step": 9720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44510825876175e-06,
+      "loss": 0.8828,
+      "step": 9721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443254634651113e-06,
+      "loss": 0.8555,
+      "step": 9722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441401065355883e-06,
+      "loss": 0.8867,
+      "step": 9723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439547550941327e-06,
+      "loss": 0.8477,
+      "step": 9724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437694091472717e-06,
+      "loss": 0.918,
+      "step": 9725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.435840687015306e-06,
+      "loss": 0.8359,
+      "step": 9726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433987337634358e-06,
+      "loss": 0.8008,
+      "step": 9727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432134043395134e-06,
+      "loss": 0.832,
+      "step": 9728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430280804362897e-06,
+      "loss": 0.9258,
+      "step": 9729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.428427620602891e-06,
+      "loss": 1.0547,
+      "step": 9730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426574492180377e-06,
+      "loss": 0.9492,
+      "step": 9731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424721419160607e-06,
+      "loss": 0.8281,
+      "step": 9732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422868401608823e-06,
+      "loss": 1.0625,
+      "step": 9733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421015439590278e-06,
+      "loss": 0.8242,
+      "step": 9734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419162533170218e-06,
+      "loss": 0.8164,
+      "step": 9735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41730968241389e-06,
+      "loss": 0.7812,
+      "step": 9736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415456887386528e-06,
+      "loss": 0.9414,
+      "step": 9737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413604148153375e-06,
+      "loss": 0.9609,
+      "step": 9738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411751464779669e-06,
+      "loss": 0.8594,
+      "step": 9739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40989883733065e-06,
+      "loss": 0.8633,
+      "step": 9740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408046265871545e-06,
+      "loss": 0.8945,
+      "step": 9741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40619375046759e-06,
+      "loss": 0.793,
+      "step": 9742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404341291184012e-06,
+      "loss": 0.7031,
+      "step": 9743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.402488888086046e-06,
+      "loss": 0.9102,
+      "step": 9744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.400636541238907e-06,
+      "loss": 0.8711,
+      "step": 9745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398784250707827e-06,
+      "loss": 0.8828,
+      "step": 9746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396932016558024e-06,
+      "loss": 0.9062,
+      "step": 9747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.395079838854725e-06,
+      "loss": 0.7891,
+      "step": 9748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393227717663139e-06,
+      "loss": 0.8477,
+      "step": 9749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391375653048485e-06,
+      "loss": 0.8711,
+      "step": 9750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.389523645075977e-06,
+      "loss": 0.8555,
+      "step": 9751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.387671693810833e-06,
+      "loss": 0.8906,
+      "step": 9752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385819799318255e-06,
+      "loss": 0.8594,
+      "step": 9753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383967961663454e-06,
+      "loss": 0.875,
+      "step": 9754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382116180911636e-06,
+      "loss": 0.8867,
+      "step": 9755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38026445712801e-06,
+      "loss": 0.8906,
+      "step": 9756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378412790377768e-06,
+      "loss": 0.8477,
+      "step": 9757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.376561180726115e-06,
+      "loss": 0.8906,
+      "step": 9758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37470962823825e-06,
+      "loss": 0.8594,
+      "step": 9759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.372858132979373e-06,
+      "loss": 0.8594,
+      "step": 9760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37100669501467e-06,
+      "loss": 0.8672,
+      "step": 9761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369155314409334e-06,
+      "loss": 0.9844,
+      "step": 9762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367303991228558e-06,
+      "loss": 0.8516,
+      "step": 9763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365452725537537e-06,
+      "loss": 1.1016,
+      "step": 9764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363601517401441e-06,
+      "loss": 0.8359,
+      "step": 9765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361750366885464e-06,
+      "loss": 0.8281,
+      "step": 9766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359899274054786e-06,
+      "loss": 0.8203,
+      "step": 9767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35804823897459e-06,
+      "loss": 0.7578,
+      "step": 9768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356197261710048e-06,
+      "loss": 0.8242,
+      "step": 9769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.354346342326338e-06,
+      "loss": 0.7852,
+      "step": 9770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.352495480888632e-06,
+      "loss": 0.8867,
+      "step": 9771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350644677462111e-06,
+      "loss": 0.8359,
+      "step": 9772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348793932111931e-06,
+      "loss": 0.9766,
+      "step": 9773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.346943244903269e-06,
+      "loss": 0.8008,
+      "step": 9774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345092615901285e-06,
+      "loss": 0.8164,
+      "step": 9775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34324204517115e-06,
+      "loss": 0.9883,
+      "step": 9776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.341391532778016e-06,
+      "loss": 0.9102,
+      "step": 9777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339541078787048e-06,
+      "loss": 0.8477,
+      "step": 9778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337690683263404e-06,
+      "loss": 0.8633,
+      "step": 9779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33584034627224e-06,
+      "loss": 0.8125,
+      "step": 9780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333990067878702e-06,
+      "loss": 1.0859,
+      "step": 9781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.332139848147947e-06,
+      "loss": 0.8125,
+      "step": 9782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330289687145121e-06,
+      "loss": 0.9258,
+      "step": 9783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328439584935381e-06,
+      "loss": 0.9219,
+      "step": 9784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326589541583857e-06,
+      "loss": 0.8516,
+      "step": 9785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3247395571557e-06,
+      "loss": 0.9102,
+      "step": 9786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32288963171605e-06,
+      "loss": 0.8281,
+      "step": 9787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32103976533005e-06,
+      "loss": 0.7695,
+      "step": 9788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319189958062827e-06,
+      "loss": 0.9258,
+      "step": 9789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31734020997952e-06,
+      "loss": 0.9023,
+      "step": 9790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.315490521145269e-06,
+      "loss": 0.9922,
+      "step": 9791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313640891625192e-06,
+      "loss": 0.9492,
+      "step": 9792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311791321484422e-06,
+      "loss": 0.8867,
+      "step": 9793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309941810788088e-06,
+      "loss": 0.9023,
+      "step": 9794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308092359601315e-06,
+      "loss": 0.7109,
+      "step": 9795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30624296798922e-06,
+      "loss": 0.8594,
+      "step": 9796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304393636016924e-06,
+      "loss": 0.9141,
+      "step": 9797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302544363749545e-06,
+      "loss": 0.9062,
+      "step": 9798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300695151252205e-06,
+      "loss": 1.0078,
+      "step": 9799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298845998590009e-06,
+      "loss": 0.9453,
+      "step": 9800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29699690582807e-06,
+      "loss": 0.9453,
+      "step": 9801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.295147873031502e-06,
+      "loss": 0.9375,
+      "step": 9802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293298900265412e-06,
+      "loss": 0.9609,
+      "step": 9803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2914499875949e-06,
+      "loss": 0.8359,
+      "step": 9804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28960113508507e-06,
+      "loss": 0.8359,
+      "step": 9805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287752342801026e-06,
+      "loss": 1.0938,
+      "step": 9806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28590361080787e-06,
+      "loss": 0.9531,
+      "step": 9807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28405493917069e-06,
+      "loss": 0.8359,
+      "step": 9808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.282206327954586e-06,
+      "loss": 0.7773,
+      "step": 9809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280357777224648e-06,
+      "loss": 0.8867,
+      "step": 9810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.278509287045973e-06,
+      "loss": 0.8242,
+      "step": 9811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276660857483639e-06,
+      "loss": 0.7617,
+      "step": 9812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274812488602738e-06,
+      "loss": 0.7812,
+      "step": 9813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272964180468352e-06,
+      "loss": 0.8906,
+      "step": 9814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.271115933145568e-06,
+      "loss": 0.8711,
+      "step": 9815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269267746699459e-06,
+      "loss": 0.875,
+      "step": 9816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267419621195104e-06,
+      "loss": 0.9297,
+      "step": 9817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.265571556697576e-06,
+      "loss": 0.9844,
+      "step": 9818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26372355327196e-06,
+      "loss": 0.8203,
+      "step": 9819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261875610983313e-06,
+      "loss": 0.8906,
+      "step": 9820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260027729896708e-06,
+      "loss": 0.7266,
+      "step": 9821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258179910077216e-06,
+      "loss": 0.8203,
+      "step": 9822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256332151589899e-06,
+      "loss": 1.0781,
+      "step": 9823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254484454499817e-06,
+      "loss": 0.8398,
+      "step": 9824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252636818872032e-06,
+      "loss": 0.9336,
+      "step": 9825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250789244771606e-06,
+      "loss": 0.8633,
+      "step": 9826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24894173226359e-06,
+      "loss": 0.8828,
+      "step": 9827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247094281413037e-06,
+      "loss": 0.7617,
+      "step": 9828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245246892285001e-06,
+      "loss": 0.9883,
+      "step": 9829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243399564944533e-06,
+      "loss": 0.8633,
+      "step": 9830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24155229945668e-06,
+      "loss": 0.8477,
+      "step": 9831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239705095886483e-06,
+      "loss": 0.8828,
+      "step": 9832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.237857954298987e-06,
+      "loss": 0.7461,
+      "step": 9833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236010874759237e-06,
+      "loss": 0.8359,
+      "step": 9834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234163857332265e-06,
+      "loss": 0.7891,
+      "step": 9835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23231690208311e-06,
+      "loss": 0.8125,
+      "step": 9836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.230470009076808e-06,
+      "loss": 0.8242,
+      "step": 9837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.228623178378385e-06,
+      "loss": 0.6992,
+      "step": 9838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22677641005288e-06,
+      "loss": 0.8242,
+      "step": 9839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22492970416531e-06,
+      "loss": 0.9102,
+      "step": 9840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22308306078071e-06,
+      "loss": 0.9375,
+      "step": 9841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221236479964098e-06,
+      "loss": 0.9297,
+      "step": 9842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.219389961780497e-06,
+      "loss": 0.8281,
+      "step": 9843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.217543506294921e-06,
+      "loss": 0.7812,
+      "step": 9844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215697113572393e-06,
+      "loss": 0.8594,
+      "step": 9845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213850783677926e-06,
+      "loss": 0.7812,
+      "step": 9846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.212004516676528e-06,
+      "loss": 0.8984,
+      "step": 9847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210158312633215e-06,
+      "loss": 0.9102,
+      "step": 9848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208312171612987e-06,
+      "loss": 0.8242,
+      "step": 9849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.206466093680857e-06,
+      "loss": 0.7891,
+      "step": 9850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204620078901821e-06,
+      "loss": 0.875,
+      "step": 9851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202774127340888e-06,
+      "loss": 0.8125,
+      "step": 9852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20092823906305e-06,
+      "loss": 0.8672,
+      "step": 9853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19908241413331e-06,
+      "loss": 0.75,
+      "step": 9854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197236652616657e-06,
+      "loss": 0.9062,
+      "step": 9855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195390954578083e-06,
+      "loss": 0.8438,
+      "step": 9856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193545320082582e-06,
+      "loss": 0.8555,
+      "step": 9857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19169974919514e-06,
+      "loss": 0.8594,
+      "step": 9858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189854241980738e-06,
+      "loss": 0.7461,
+      "step": 9859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188008798504368e-06,
+      "loss": 0.7891,
+      "step": 9860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186163418831004e-06,
+      "loss": 0.8594,
+      "step": 9861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184318103025627e-06,
+      "loss": 0.8594,
+      "step": 9862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182472851153214e-06,
+      "loss": 0.8906,
+      "step": 9863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180627663278736e-06,
+      "loss": 0.7734,
+      "step": 9864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178782539467166e-06,
+      "loss": 0.8281,
+      "step": 9865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176937479783477e-06,
+      "loss": 0.9062,
+      "step": 9866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175092484292633e-06,
+      "loss": 0.8984,
+      "step": 9867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1732475530596e-06,
+      "loss": 0.9961,
+      "step": 9868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17140268614934e-06,
+      "loss": 0.8633,
+      "step": 9869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16955788362682e-06,
+      "loss": 0.8125,
+      "step": 9870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16771314555699e-06,
+      "loss": 0.8203,
+      "step": 9871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165868472004807e-06,
+      "loss": 0.7969,
+      "step": 9872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164023863035229e-06,
+      "loss": 0.7773,
+      "step": 9873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.162179318713206e-06,
+      "loss": 0.8164,
+      "step": 9874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.160334839103687e-06,
+      "loss": 0.8242,
+      "step": 9875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158490424271618e-06,
+      "loss": 0.8047,
+      "step": 9876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156646074281943e-06,
+      "loss": 0.8125,
+      "step": 9877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15480178919961e-06,
+      "loss": 0.8398,
+      "step": 9878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152957569089552e-06,
+      "loss": 0.7539,
+      "step": 9879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.151113414016709e-06,
+      "loss": 0.8594,
+      "step": 9880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.149269324046017e-06,
+      "loss": 0.9102,
+      "step": 9881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147425299242415e-06,
+      "loss": 0.9922,
+      "step": 9882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145581339670824e-06,
+      "loss": 0.8906,
+      "step": 9883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143737445396176e-06,
+      "loss": 0.8789,
+      "step": 9884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1418936164834e-06,
+      "loss": 0.7812,
+      "step": 9885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140049852997422e-06,
+      "loss": 0.9102,
+      "step": 9886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138206155003156e-06,
+      "loss": 0.9531,
+      "step": 9887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136362522565527e-06,
+      "loss": 0.7578,
+      "step": 9888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13451895574945e-06,
+      "loss": 0.9844,
+      "step": 9889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132675454619846e-06,
+      "loss": 0.8828,
+      "step": 9890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130832019241618e-06,
+      "loss": 0.9102,
+      "step": 9891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12898864967968e-06,
+      "loss": 1.0469,
+      "step": 9892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12714534599894e-06,
+      "loss": 0.7305,
+      "step": 9893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12530210826431e-06,
+      "loss": 0.8672,
+      "step": 9894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123458936540682e-06,
+      "loss": 0.8281,
+      "step": 9895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121615830892963e-06,
+      "loss": 0.9375,
+      "step": 9896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11977279138605e-06,
+      "loss": 0.8281,
+      "step": 9897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117929818084844e-06,
+      "loss": 0.9414,
+      "step": 9898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116086911054232e-06,
+      "loss": 0.8047,
+      "step": 9899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114244070359108e-06,
+      "loss": 0.9609,
+      "step": 9900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112401296064364e-06,
+      "loss": 0.8594,
+      "step": 9901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.110558588234885e-06,
+      "loss": 0.8594,
+      "step": 9902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108715946935552e-06,
+      "loss": 0.918,
+      "step": 9903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106873372231253e-06,
+      "loss": 0.8555,
+      "step": 9904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105030864186867e-06,
+      "loss": 0.9414,
+      "step": 9905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103188422867268e-06,
+      "loss": 0.8555,
+      "step": 9906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.101346048337332e-06,
+      "loss": 0.8672,
+      "step": 9907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099503740661934e-06,
+      "loss": 0.8516,
+      "step": 9908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097661499905947e-06,
+      "loss": 0.8516,
+      "step": 9909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.095819326134232e-06,
+      "loss": 0.7891,
+      "step": 9910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09397721941166e-06,
+      "loss": 0.8203,
+      "step": 9911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092135179803091e-06,
+      "loss": 0.8789,
+      "step": 9912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090293207373393e-06,
+      "loss": 0.8398,
+      "step": 9913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088451302187415e-06,
+      "loss": 0.7969,
+      "step": 9914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086609464310019e-06,
+      "loss": 0.7734,
+      "step": 9915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084767693806057e-06,
+      "loss": 0.8125,
+      "step": 9916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082925990740386e-06,
+      "loss": 0.8008,
+      "step": 9917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081084355177848e-06,
+      "loss": 0.8516,
+      "step": 9918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079242787183292e-06,
+      "loss": 0.8789,
+      "step": 9919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077401286821562e-06,
+      "loss": 0.8438,
+      "step": 9920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075559854157506e-06,
+      "loss": 0.7891,
+      "step": 9921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.073718489255955e-06,
+      "loss": 0.9102,
+      "step": 9922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07187719218175e-06,
+      "loss": 0.9766,
+      "step": 9923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.070035962999726e-06,
+      "loss": 0.8516,
+      "step": 9924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068194801774722e-06,
+      "loss": 0.7812,
+      "step": 9925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.066353708571556e-06,
+      "loss": 0.8164,
+      "step": 9926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064512683455059e-06,
+      "loss": 0.8984,
+      "step": 9927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062671726490061e-06,
+      "loss": 0.9258,
+      "step": 9928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060830837741387e-06,
+      "loss": 0.8477,
+      "step": 9929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058990017273849e-06,
+      "loss": 0.8672,
+      "step": 9930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057149265152268e-06,
+      "loss": 0.7344,
+      "step": 9931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055308581441463e-06,
+      "loss": 0.8086,
+      "step": 9932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05346796620625e-06,
+      "loss": 0.832,
+      "step": 9933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051627419511429e-06,
+      "loss": 0.8438,
+      "step": 9934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049786941421815e-06,
+      "loss": 0.8828,
+      "step": 9935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047946532002216e-06,
+      "loss": 0.8711,
+      "step": 9936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046106191317435e-06,
+      "loss": 0.8828,
+      "step": 9937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04426591943227e-06,
+      "loss": 0.8906,
+      "step": 9938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04242571641152e-06,
+      "loss": 0.8945,
+      "step": 9939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040585582319984e-06,
+      "loss": 0.8047,
+      "step": 9940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03874551722246e-06,
+      "loss": 0.8359,
+      "step": 9941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03690552118373e-06,
+      "loss": 0.8633,
+      "step": 9942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035065594268585e-06,
+      "loss": 0.7969,
+      "step": 9943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033225736541817e-06,
+      "loss": 0.8008,
+      "step": 9944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.031385948068211e-06,
+      "loss": 0.8984,
+      "step": 9945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029546228912542e-06,
+      "loss": 0.8984,
+      "step": 9946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02770657913959e-06,
+      "loss": 0.8633,
+      "step": 9947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.025866998814137e-06,
+      "loss": 0.8789,
+      "step": 9948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.024027488000958e-06,
+      "loss": 0.8906,
+      "step": 9949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02218804676482e-06,
+      "loss": 0.7656,
+      "step": 9950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02034867517049e-06,
+      "loss": 0.8125,
+      "step": 9951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018509373282743e-06,
+      "loss": 0.7734,
+      "step": 9952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016670141166345e-06,
+      "loss": 0.9102,
+      "step": 9953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014830978886048e-06,
+      "loss": 0.75,
+      "step": 9954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012991886506619e-06,
+      "loss": 0.9883,
+      "step": 9955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011152864092812e-06,
+      "loss": 0.875,
+      "step": 9956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.009313911709388e-06,
+      "loss": 0.9219,
+      "step": 9957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00747502942109e-06,
+      "loss": 0.9336,
+      "step": 9958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005636217292674e-06,
+      "loss": 0.9219,
+      "step": 9959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00379747538889e-06,
+      "loss": 0.9375,
+      "step": 9960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.001958803774474e-06,
+      "loss": 0.9336,
+      "step": 9961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000120202514174e-06,
+      "loss": 0.8086,
+      "step": 9962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99828167167273e-06,
+      "loss": 0.8359,
+      "step": 9963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996443211314882e-06,
+      "loss": 0.8125,
+      "step": 9964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994604821505359e-06,
+      "loss": 0.9219,
+      "step": 9965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992766502308896e-06,
+      "loss": 1.0234,
+      "step": 9966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990928253790224e-06,
+      "loss": 0.8398,
+      "step": 9967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989090076014074e-06,
+      "loss": 0.8555,
+      "step": 9968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987251969045162e-06,
+      "loss": 0.8516,
+      "step": 9969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.985413932948218e-06,
+      "loss": 0.8516,
+      "step": 9970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983575967787957e-06,
+      "loss": 0.8438,
+      "step": 9971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981738073629105e-06,
+      "loss": 0.9219,
+      "step": 9972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97990025053637e-06,
+      "loss": 0.7695,
+      "step": 9973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978062498574464e-06,
+      "loss": 0.9336,
+      "step": 9974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976224817808099e-06,
+      "loss": 0.8516,
+      "step": 9975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974387208301987e-06,
+      "loss": 0.8281,
+      "step": 9976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972549670120825e-06,
+      "loss": 0.832,
+      "step": 9977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97071220332932e-06,
+      "loss": 0.8281,
+      "step": 9978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968874807992172e-06,
+      "loss": 0.8086,
+      "step": 9979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967037484174082e-06,
+      "loss": 0.9102,
+      "step": 9980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965200231939737e-06,
+      "loss": 0.8398,
+      "step": 9981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963363051353832e-06,
+      "loss": 0.8164,
+      "step": 9982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961525942481058e-06,
+      "loss": 0.8711,
+      "step": 9983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959688905386109e-06,
+      "loss": 1.1094,
+      "step": 9984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95785194013366e-06,
+      "loss": 0.8594,
+      "step": 9985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.956015046788397e-06,
+      "loss": 0.9141,
+      "step": 9986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954178225414999e-06,
+      "loss": 0.75,
+      "step": 9987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95234147607815e-06,
+      "loss": 0.8828,
+      "step": 9988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950504798842513e-06,
+      "loss": 0.8945,
+      "step": 9989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948668193772768e-06,
+      "loss": 0.8516,
+      "step": 9990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946831660933584e-06,
+      "loss": 0.9922,
+      "step": 9991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.94499520038963e-06,
+      "loss": 1.0547,
+      "step": 9992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943158812205564e-06,
+      "loss": 0.7852,
+      "step": 9993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941322496446053e-06,
+      "loss": 0.8359,
+      "step": 9994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.939486253175755e-06,
+      "loss": 0.8711,
+      "step": 9995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93765008245933e-06,
+      "loss": 0.8477,
+      "step": 9996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935813984361427e-06,
+      "loss": 0.832,
+      "step": 9997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9339779589467e-06,
+      "loss": 0.832,
+      "step": 9998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932142006279802e-06,
+      "loss": 0.8477,
+      "step": 9999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.930306126425377e-06,
+      "loss": 0.8477,
+      "step": 10000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928470319448064e-06,
+      "loss": 0.8672,
+      "step": 10001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926634585412514e-06,
+      "loss": 0.8633,
+      "step": 10002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.924798924383361e-06,
+      "loss": 0.8789,
+      "step": 10003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922963336425243e-06,
+      "loss": 0.793,
+      "step": 10004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.921127821602789e-06,
+      "loss": 0.9844,
+      "step": 10005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919292379980636e-06,
+      "loss": 0.7969,
+      "step": 10006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91745701162341e-06,
+      "loss": 0.8789,
+      "step": 10007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915621716595742e-06,
+      "loss": 0.9141,
+      "step": 10008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913786494962248e-06,
+      "loss": 0.9375,
+      "step": 10009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911951346787554e-06,
+      "loss": 0.7539,
+      "step": 10010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.910116272136276e-06,
+      "loss": 0.918,
+      "step": 10011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908281271073032e-06,
+      "loss": 0.7617,
+      "step": 10012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906446343662431e-06,
+      "loss": 0.8555,
+      "step": 10013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.904611489969092e-06,
+      "loss": 0.8359,
+      "step": 10014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902776710057614e-06,
+      "loss": 0.8672,
+      "step": 10015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900942003992606e-06,
+      "loss": 0.9961,
+      "step": 10016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899107371838671e-06,
+      "loss": 0.8164,
+      "step": 10017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897272813660411e-06,
+      "loss": 0.918,
+      "step": 10018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895438329522423e-06,
+      "loss": 0.8438,
+      "step": 10019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893603919489298e-06,
+      "loss": 0.7773,
+      "step": 10020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891769583625633e-06,
+      "loss": 0.7969,
+      "step": 10021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889935321996014e-06,
+      "loss": 0.8242,
+      "step": 10022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.888101134665031e-06,
+      "loss": 0.9648,
+      "step": 10023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886267021697268e-06,
+      "loss": 0.8633,
+      "step": 10024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884432983157308e-06,
+      "loss": 0.7734,
+      "step": 10025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882599019109726e-06,
+      "loss": 0.9219,
+      "step": 10026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880765129619106e-06,
+      "loss": 0.8047,
+      "step": 10027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878931314750015e-06,
+      "loss": 0.8281,
+      "step": 10028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.877097574567029e-06,
+      "loss": 0.9141,
+      "step": 10029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875263909134712e-06,
+      "loss": 0.8359,
+      "step": 10030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873430318517636e-06,
+      "loss": 0.8359,
+      "step": 10031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871596802780361e-06,
+      "loss": 0.8984,
+      "step": 10032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86976336198745e-06,
+      "loss": 0.8242,
+      "step": 10033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.867929996203458e-06,
+      "loss": 0.9609,
+      "step": 10034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866096705492945e-06,
+      "loss": 0.875,
+      "step": 10035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864263489920462e-06,
+      "loss": 0.957,
+      "step": 10036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862430349550558e-06,
+      "loss": 0.9883,
+      "step": 10037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86059728444778e-06,
+      "loss": 0.8047,
+      "step": 10038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858764294676677e-06,
+      "loss": 0.875,
+      "step": 10039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85693138030179e-06,
+      "loss": 0.8672,
+      "step": 10040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.855098541387656e-06,
+      "loss": 0.8516,
+      "step": 10041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853265777998814e-06,
+      "loss": 0.8281,
+      "step": 10042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.851433090199801e-06,
+      "loss": 0.9375,
+      "step": 10043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849600478055145e-06,
+      "loss": 0.8594,
+      "step": 10044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847767941629374e-06,
+      "loss": 0.8711,
+      "step": 10045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.845935480987018e-06,
+      "loss": 0.8984,
+      "step": 10046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844103096192602e-06,
+      "loss": 0.8555,
+      "step": 10047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.842270787310644e-06,
+      "loss": 0.8008,
+      "step": 10048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840438554405662e-06,
+      "loss": 0.9453,
+      "step": 10049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83860639754217e-06,
+      "loss": 0.957,
+      "step": 10050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836774316784693e-06,
+      "loss": 0.8516,
+      "step": 10051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834942312197725e-06,
+      "loss": 0.7852,
+      "step": 10052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83311038384578e-06,
+      "loss": 0.8359,
+      "step": 10053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831278531793365e-06,
+      "loss": 0.7305,
+      "step": 10054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829446756104986e-06,
+      "loss": 0.9062,
+      "step": 10055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.827615056845134e-06,
+      "loss": 0.8086,
+      "step": 10056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82578343407831e-06,
+      "loss": 0.8203,
+      "step": 10057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823951887869008e-06,
+      "loss": 0.8984,
+      "step": 10058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822120418281724e-06,
+      "loss": 0.8203,
+      "step": 10059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.820289025380938e-06,
+      "loss": 0.9414,
+      "step": 10060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818457709231141e-06,
+      "loss": 0.793,
+      "step": 10061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816626469896818e-06,
+      "loss": 0.9688,
+      "step": 10062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814795307442451e-06,
+      "loss": 0.9219,
+      "step": 10063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812964221932511e-06,
+      "loss": 0.918,
+      "step": 10064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.811133213431477e-06,
+      "loss": 0.8828,
+      "step": 10065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809302282003823e-06,
+      "loss": 0.8086,
+      "step": 10066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807471427714022e-06,
+      "loss": 0.8828,
+      "step": 10067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805640650626532e-06,
+      "loss": 0.9414,
+      "step": 10068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.803809950805824e-06,
+      "loss": 0.8281,
+      "step": 10069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801979328316364e-06,
+      "loss": 0.8906,
+      "step": 10070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800148783222599e-06,
+      "loss": 0.8633,
+      "step": 10071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798318315588993e-06,
+      "loss": 0.8594,
+      "step": 10072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796487925479999e-06,
+      "loss": 0.7852,
+      "step": 10073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79465761296007e-06,
+      "loss": 0.8945,
+      "step": 10074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792827378093649e-06,
+      "loss": 0.9453,
+      "step": 10075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790997220945184e-06,
+      "loss": 0.8359,
+      "step": 10076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789167141579116e-06,
+      "loss": 0.9492,
+      "step": 10077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78733714005989e-06,
+      "loss": 0.6953,
+      "step": 10078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785507216451937e-06,
+      "loss": 0.8594,
+      "step": 10079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783677370819695e-06,
+      "loss": 0.8633,
+      "step": 10080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781847603227594e-06,
+      "loss": 0.8906,
+      "step": 10081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780017913740066e-06,
+      "loss": 0.8555,
+      "step": 10082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778188302421533e-06,
+      "loss": 1.0391,
+      "step": 10083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77635876933642e-06,
+      "loss": 0.8828,
+      "step": 10084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774529314549147e-06,
+      "loss": 0.9766,
+      "step": 10085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772699938124137e-06,
+      "loss": 0.9414,
+      "step": 10086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770870640125799e-06,
+      "loss": 0.8828,
+      "step": 10087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769041420618545e-06,
+      "loss": 0.8828,
+      "step": 10088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76721227966679e-06,
+      "loss": 0.918,
+      "step": 10089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76538321733494e-06,
+      "loss": 0.918,
+      "step": 10090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763554233687393e-06,
+      "loss": 0.9414,
+      "step": 10091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761725328788555e-06,
+      "loss": 0.75,
+      "step": 10092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759896502702823e-06,
+      "loss": 0.8672,
+      "step": 10093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7580677554946e-06,
+      "loss": 0.8789,
+      "step": 10094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756239087228268e-06,
+      "loss": 0.75,
+      "step": 10095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.754410497968222e-06,
+      "loss": 0.7617,
+      "step": 10096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75258198777885e-06,
+      "loss": 0.9531,
+      "step": 10097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.75075355672454e-06,
+      "loss": 0.8047,
+      "step": 10098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748925204869667e-06,
+      "loss": 0.7812,
+      "step": 10099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74709693227861e-06,
+      "loss": 0.9023,
+      "step": 10100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745268739015751e-06,
+      "loss": 0.8906,
+      "step": 10101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743440625145466e-06,
+      "loss": 1.0234,
+      "step": 10102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741612590732114e-06,
+      "loss": 0.9023,
+      "step": 10103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73978463584007e-06,
+      "loss": 1.0078,
+      "step": 10104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.7379567605337e-06,
+      "loss": 0.8242,
+      "step": 10105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73612896487737e-06,
+      "loss": 0.8594,
+      "step": 10106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734301248935429e-06,
+      "loss": 1.0312,
+      "step": 10107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.732473612772239e-06,
+      "loss": 0.9727,
+      "step": 10108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730646056452153e-06,
+      "loss": 0.8789,
+      "step": 10109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72881858003953e-06,
+      "loss": 0.8125,
+      "step": 10110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726991183598707e-06,
+      "loss": 0.875,
+      "step": 10111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725163867194033e-06,
+      "loss": 0.8984,
+      "step": 10112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72333663088985e-06,
+      "loss": 0.8359,
+      "step": 10113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721509474750507e-06,
+      "loss": 0.8438,
+      "step": 10114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719682398840328e-06,
+      "loss": 0.8125,
+      "step": 10115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.717855403223651e-06,
+      "loss": 0.8359,
+      "step": 10116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71602848796481e-06,
+      "loss": 0.8828,
+      "step": 10117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714201653128138e-06,
+      "loss": 0.9648,
+      "step": 10118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.712374898777948e-06,
+      "loss": 0.9375,
+      "step": 10119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710548224978571e-06,
+      "loss": 0.8867,
+      "step": 10120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708721631794327e-06,
+      "loss": 0.9258,
+      "step": 10121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706895119289534e-06,
+      "loss": 0.8516,
+      "step": 10122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705068687528503e-06,
+      "loss": 0.8594,
+      "step": 10123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703242336575544e-06,
+      "loss": 0.8711,
+      "step": 10124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701416066494976e-06,
+      "loss": 0.7969,
+      "step": 10125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69958987735109e-06,
+      "loss": 0.8477,
+      "step": 10126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697763769208199e-06,
+      "loss": 0.8828,
+      "step": 10127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695937742130598e-06,
+      "loss": 0.9102,
+      "step": 10128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694111796182591e-06,
+      "loss": 0.9141,
+      "step": 10129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692285931428466e-06,
+      "loss": 0.9883,
+      "step": 10130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690460147932515e-06,
+      "loss": 0.9492,
+      "step": 10131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688634445759026e-06,
+      "loss": 0.8945,
+      "step": 10132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686808824972294e-06,
+      "loss": 0.9492,
+      "step": 10133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68498328563659e-06,
+      "loss": 0.8164,
+      "step": 10134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6831578278162e-06,
+      "loss": 0.9062,
+      "step": 10135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681332451575397e-06,
+      "loss": 0.875,
+      "step": 10136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679507156978465e-06,
+      "loss": 0.8984,
+      "step": 10137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677681944089663e-06,
+      "loss": 0.8672,
+      "step": 10138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675856812973266e-06,
+      "loss": 0.7812,
+      "step": 10139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67403176369354e-06,
+      "loss": 0.7969,
+      "step": 10140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672206796314751e-06,
+      "loss": 0.875,
+      "step": 10141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67038191090115e-06,
+      "loss": 0.8633,
+      "step": 10142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668557107516998e-06,
+      "loss": 0.9219,
+      "step": 10143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666732386226553e-06,
+      "loss": 0.8594,
+      "step": 10144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664907747094063e-06,
+      "loss": 0.7656,
+      "step": 10145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663083190183775e-06,
+      "loss": 0.8789,
+      "step": 10146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.661258715559936e-06,
+      "loss": 0.957,
+      "step": 10147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659434323286788e-06,
+      "loss": 0.8008,
+      "step": 10148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657610013428577e-06,
+      "loss": 0.7695,
+      "step": 10149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655785786049529e-06,
+      "loss": 0.9414,
+      "step": 10150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653961641213882e-06,
+      "loss": 0.7344,
+      "step": 10151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652137578985868e-06,
+      "loss": 0.7344,
+      "step": 10152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65031359942972e-06,
+      "loss": 0.8477,
+      "step": 10153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648489702609653e-06,
+      "loss": 0.8047,
+      "step": 10154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646665888589897e-06,
+      "loss": 0.9766,
+      "step": 10155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.644842157434665e-06,
+      "loss": 0.7969,
+      "step": 10156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643018509208183e-06,
+      "loss": 0.75,
+      "step": 10157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641194943974654e-06,
+      "loss": 0.8242,
+      "step": 10158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.639371461798293e-06,
+      "loss": 0.7891,
+      "step": 10159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637548062743307e-06,
+      "loss": 0.8672,
+      "step": 10160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635724746873908e-06,
+      "loss": 1.0234,
+      "step": 10161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633901514254284e-06,
+      "loss": 0.9805,
+      "step": 10162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632078364948642e-06,
+      "loss": 0.8867,
+      "step": 10163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630255299021176e-06,
+      "loss": 0.8828,
+      "step": 10164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628432316536085e-06,
+      "loss": 0.9219,
+      "step": 10165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.62660941755755e-06,
+      "loss": 0.8203,
+      "step": 10166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.624786602149761e-06,
+      "loss": 0.8672,
+      "step": 10167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622963870376906e-06,
+      "loss": 0.9336,
+      "step": 10168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.621141222303162e-06,
+      "loss": 1.0234,
+      "step": 10169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619318657992707e-06,
+      "loss": 1.0391,
+      "step": 10170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6174961775097185e-06,
+      "loss": 0.7812,
+      "step": 10171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61567378091837e-06,
+      "loss": 0.9453,
+      "step": 10172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613851468282829e-06,
+      "loss": 0.8125,
+      "step": 10173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.612029239667259e-06,
+      "loss": 0.9922,
+      "step": 10174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610207095135827e-06,
+      "loss": 0.7461,
+      "step": 10175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608385034752696e-06,
+      "loss": 0.8281,
+      "step": 10176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606563058582019e-06,
+      "loss": 0.8594,
+      "step": 10177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604741166687952e-06,
+      "loss": 0.8594,
+      "step": 10178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.602919359134647e-06,
+      "loss": 0.8984,
+      "step": 10179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601097635986252e-06,
+      "loss": 0.9219,
+      "step": 10180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.599275997306914e-06,
+      "loss": 1.0,
+      "step": 10181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597454443160773e-06,
+      "loss": 0.7656,
+      "step": 10182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595632973611973e-06,
+      "loss": 0.9688,
+      "step": 10183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593811588724649e-06,
+      "loss": 0.7969,
+      "step": 10184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.59199028856293e-06,
+      "loss": 0.8516,
+      "step": 10185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.590169073190953e-06,
+      "loss": 0.8359,
+      "step": 10186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588347942672844e-06,
+      "loss": 0.8125,
+      "step": 10187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586526897072729e-06,
+      "loss": 0.8398,
+      "step": 10188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.584705936454726e-06,
+      "loss": 0.8047,
+      "step": 10189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582885060882955e-06,
+      "loss": 0.8516,
+      "step": 10190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5810642704215365e-06,
+      "loss": 0.7773,
+      "step": 10191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579243565134579e-06,
+      "loss": 0.8125,
+      "step": 10192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577422945086192e-06,
+      "loss": 0.8555,
+      "step": 10193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575602410340484e-06,
+      "loss": 0.8125,
+      "step": 10194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573781960961558e-06,
+      "loss": 1.0469,
+      "step": 10195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571961597013516e-06,
+      "loss": 0.7734,
+      "step": 10196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5701413185604545e-06,
+      "loss": 0.7656,
+      "step": 10197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56832112566647e-06,
+      "loss": 0.8203,
+      "step": 10198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.566501018395651e-06,
+      "loss": 0.8789,
+      "step": 10199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56468099681209e-06,
+      "loss": 0.8555,
+      "step": 10200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56286106097987e-06,
+      "loss": 0.7695,
+      "step": 10201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561041210963077e-06,
+      "loss": 0.7422,
+      "step": 10202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5592214468257865e-06,
+      "loss": 0.8867,
+      "step": 10203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55740176863208e-06,
+      "loss": 0.9414,
+      "step": 10204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555582176446025e-06,
+      "loss": 0.8477,
+      "step": 10205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5537626703317005e-06,
+      "loss": 0.7891,
+      "step": 10206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551943250353165e-06,
+      "loss": 0.8906,
+      "step": 10207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55012391657449e-06,
+      "loss": 0.7852,
+      "step": 10208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548304669059735e-06,
+      "loss": 0.8359,
+      "step": 10209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546485507872956e-06,
+      "loss": 0.8789,
+      "step": 10210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5446664330782096e-06,
+      "loss": 0.8477,
+      "step": 10211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.542847444739552e-06,
+      "loss": 0.8125,
+      "step": 10212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54102854292103e-06,
+      "loss": 0.832,
+      "step": 10213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539209727686686e-06,
+      "loss": 0.8555,
+      "step": 10214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5373909991005675e-06,
+      "loss": 0.8828,
+      "step": 10215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535572357226717e-06,
+      "loss": 0.8633,
+      "step": 10216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533753802129169e-06,
+      "loss": 0.7578,
+      "step": 10217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5319353338719535e-06,
+      "loss": 0.9453,
+      "step": 10218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530116952519107e-06,
+      "loss": 0.7891,
+      "step": 10219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528298658134658e-06,
+      "loss": 0.8203,
+      "step": 10220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52648045078263e-06,
+      "loss": 0.7812,
+      "step": 10221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524662330527043e-06,
+      "loss": 0.9023,
+      "step": 10222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522844297431916e-06,
+      "loss": 0.918,
+      "step": 10223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52102635156127e-06,
+      "loss": 0.8008,
+      "step": 10224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519208492979111e-06,
+      "loss": 0.7578,
+      "step": 10225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5173907217494504e-06,
+      "loss": 0.7969,
+      "step": 10226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515573037936294e-06,
+      "loss": 0.9297,
+      "step": 10227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513755441603654e-06,
+      "loss": 0.9023,
+      "step": 10228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.511937932815517e-06,
+      "loss": 0.9375,
+      "step": 10229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510120511635889e-06,
+      "loss": 0.832,
+      "step": 10230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508303178128761e-06,
+      "loss": 0.9141,
+      "step": 10231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50648593235813e-06,
+      "loss": 0.8281,
+      "step": 10232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504668774387974e-06,
+      "loss": 0.875,
+      "step": 10233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502851704282284e-06,
+      "loss": 0.8398,
+      "step": 10234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501034722105041e-06,
+      "loss": 0.8945,
+      "step": 10235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499217827920229e-06,
+      "loss": 0.8242,
+      "step": 10236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497401021791812e-06,
+      "loss": 0.9023,
+      "step": 10237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.495584303783772e-06,
+      "loss": 0.957,
+      "step": 10238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493767673960077e-06,
+      "loss": 0.8242,
+      "step": 10239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491951132384689e-06,
+      "loss": 0.9766,
+      "step": 10240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4901346791215736e-06,
+      "loss": 0.8945,
+      "step": 10241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48831831423469e-06,
+      "loss": 0.957,
+      "step": 10242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486502037788004e-06,
+      "loss": 0.9766,
+      "step": 10243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484685849845455e-06,
+      "loss": 0.8398,
+      "step": 10244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4828697504710025e-06,
+      "loss": 0.9062,
+      "step": 10245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481053739728592e-06,
+      "loss": 0.8438,
+      "step": 10246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.479237817682172e-06,
+      "loss": 1.0234,
+      "step": 10247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.477421984395679e-06,
+      "loss": 0.8047,
+      "step": 10248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4756062399330495e-06,
+      "loss": 0.9102,
+      "step": 10249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473790584358224e-06,
+      "loss": 0.7812,
+      "step": 10250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471975017735136e-06,
+      "loss": 1.1016,
+      "step": 10251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470159540127706e-06,
+      "loss": 0.75,
+      "step": 10252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468344151599865e-06,
+      "loss": 1.0078,
+      "step": 10253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466528852215536e-06,
+      "loss": 0.8906,
+      "step": 10254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464713642038642e-06,
+      "loss": 0.7617,
+      "step": 10255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46289852113309e-06,
+      "loss": 0.8398,
+      "step": 10256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461083489562799e-06,
+      "loss": 0.9141,
+      "step": 10257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4592685473916784e-06,
+      "loss": 0.957,
+      "step": 10258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.457453694683639e-06,
+      "loss": 0.8242,
+      "step": 10259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455638931502577e-06,
+      "loss": 0.8086,
+      "step": 10260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453824257912397e-06,
+      "loss": 0.8789,
+      "step": 10261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.452009673976994e-06,
+      "loss": 0.8398,
+      "step": 10262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450195179760271e-06,
+      "loss": 0.7812,
+      "step": 10263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.448380775326108e-06,
+      "loss": 0.7617,
+      "step": 10264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446566460738397e-06,
+      "loss": 0.9844,
+      "step": 10265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444752236061024e-06,
+      "loss": 0.8984,
+      "step": 10266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.442938101357873e-06,
+      "loss": 0.7969,
+      "step": 10267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441124056692817e-06,
+      "loss": 0.8477,
+      "step": 10268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.439310102129733e-06,
+      "loss": 0.9102,
+      "step": 10269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.437496237732493e-06,
+      "loss": 0.7891,
+      "step": 10270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435682463564973e-06,
+      "loss": 0.8594,
+      "step": 10271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.433868779691028e-06,
+      "loss": 0.875,
+      "step": 10272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432055186174525e-06,
+      "loss": 0.9883,
+      "step": 10273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.430241683079325e-06,
+      "loss": 0.8789,
+      "step": 10274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428428270469285e-06,
+      "loss": 0.7969,
+      "step": 10275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426614948408253e-06,
+      "loss": 0.9492,
+      "step": 10276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424801716960082e-06,
+      "loss": 0.8906,
+      "step": 10277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422988576188619e-06,
+      "loss": 0.7734,
+      "step": 10278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421175526157711e-06,
+      "loss": 0.8008,
+      "step": 10279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4193625669311895e-06,
+      "loss": 0.9531,
+      "step": 10280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4175496985728986e-06,
+      "loss": 0.9062,
+      "step": 10281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4157369211466676e-06,
+      "loss": 1.0234,
+      "step": 10282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4139242347163365e-06,
+      "loss": 0.8906,
+      "step": 10283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.41211163934572e-06,
+      "loss": 0.9922,
+      "step": 10284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410299135098649e-06,
+      "loss": 0.9727,
+      "step": 10285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408486722038943e-06,
+      "loss": 0.8359,
+      "step": 10286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406674400230426e-06,
+      "loss": 0.8984,
+      "step": 10287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404862169736903e-06,
+      "loss": 0.8438,
+      "step": 10288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4030500306221894e-06,
+      "loss": 0.8398,
+      "step": 10289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.401237982950094e-06,
+      "loss": 0.8477,
+      "step": 10290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399426026784426e-06,
+      "loss": 0.8359,
+      "step": 10291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3976141621889794e-06,
+      "loss": 0.8906,
+      "step": 10292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395802389227555e-06,
+      "loss": 0.8867,
+      "step": 10293
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.393990707963955e-06,
+      "loss": 0.8633,
+      "step": 10294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392179118461962e-06,
+      "loss": 0.8906,
+      "step": 10295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390367620785366e-06,
+      "loss": 0.8828,
+      "step": 10296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388556214997957e-06,
+      "loss": 1.0312,
+      "step": 10297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386744901163521e-06,
+      "loss": 0.7539,
+      "step": 10298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.384933679345826e-06,
+      "loss": 0.8203,
+      "step": 10299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383122549608654e-06,
+      "loss": 0.8359,
+      "step": 10300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.381311512015779e-06,
+      "loss": 0.9219,
+      "step": 10301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37950056663097e-06,
+      "loss": 0.8555,
+      "step": 10302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37768971351799e-06,
+      "loss": 0.9062,
+      "step": 10303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375878952740605e-06,
+      "loss": 0.6758,
+      "step": 10304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3740682843625724e-06,
+      "loss": 0.9062,
+      "step": 10305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372257708447654e-06,
+      "loss": 0.8125,
+      "step": 10306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370447225059595e-06,
+      "loss": 0.8594,
+      "step": 10307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3686368342621496e-06,
+      "loss": 0.9961,
+      "step": 10308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366826536119062e-06,
+      "loss": 0.8867,
+      "step": 10309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.365016330694085e-06,
+      "loss": 0.8359,
+      "step": 10310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363206218050945e-06,
+      "loss": 0.9609,
+      "step": 10311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.361396198253387e-06,
+      "loss": 0.9883,
+      "step": 10312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.359586271365142e-06,
+      "loss": 0.8672,
+      "step": 10313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357776437449945e-06,
+      "loss": 0.8242,
+      "step": 10314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355966696571515e-06,
+      "loss": 0.8281,
+      "step": 10315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35415704879358e-06,
+      "loss": 0.875,
+      "step": 10316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.352347494179861e-06,
+      "loss": 0.793,
+      "step": 10317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350538032794079e-06,
+      "loss": 0.8398,
+      "step": 10318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.348728664699939e-06,
+      "loss": 0.9219,
+      "step": 10319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346919389961156e-06,
+      "loss": 0.9023,
+      "step": 10320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3451102086414385e-06,
+      "loss": 0.8086,
+      "step": 10321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3433011208044936e-06,
+      "loss": 0.8398,
+      "step": 10322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341492126514014e-06,
+      "loss": 0.8164,
+      "step": 10323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3396832258337005e-06,
+      "loss": 0.8984,
+      "step": 10324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337874418827249e-06,
+      "loss": 0.9648,
+      "step": 10325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336065705558353e-06,
+      "loss": 0.8398,
+      "step": 10326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3342570860906924e-06,
+      "loss": 0.918,
+      "step": 10327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332448560487957e-06,
+      "loss": 1.0,
+      "step": 10328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3306401288138255e-06,
+      "loss": 0.9141,
+      "step": 10329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32883179113198e-06,
+      "loss": 0.9648,
+      "step": 10330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.327023547506089e-06,
+      "loss": 0.7617,
+      "step": 10331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3252153979998254e-06,
+      "loss": 0.9102,
+      "step": 10332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.323407342676856e-06,
+      "loss": 0.9023,
+      "step": 10333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321599381600853e-06,
+      "loss": 0.8555,
+      "step": 10334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319791514835466e-06,
+      "loss": 0.8125,
+      "step": 10335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.317983742444357e-06,
+      "loss": 0.875,
+      "step": 10336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3161760644911825e-06,
+      "loss": 0.832,
+      "step": 10337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314368481039597e-06,
+      "loss": 1.0,
+      "step": 10338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312560992153239e-06,
+      "loss": 0.9531,
+      "step": 10339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310753597895759e-06,
+      "loss": 0.7969,
+      "step": 10340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308946298330799e-06,
+      "loss": 0.9023,
+      "step": 10341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3071390935219935e-06,
+      "loss": 0.7656,
+      "step": 10342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3053319835329775e-06,
+      "loss": 0.8438,
+      "step": 10343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303524968427383e-06,
+      "loss": 0.8945,
+      "step": 10344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.301718048268837e-06,
+      "loss": 0.8438,
+      "step": 10345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.299911223120968e-06,
+      "loss": 0.9609,
+      "step": 10346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.298104493047391e-06,
+      "loss": 0.8477,
+      "step": 10347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296297858111725e-06,
+      "loss": 0.8398,
+      "step": 10348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29449131837759e-06,
+      "loss": 0.7891,
+      "step": 10349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.292684873908591e-06,
+      "loss": 0.9688,
+      "step": 10350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.290878524768336e-06,
+      "loss": 0.7852,
+      "step": 10351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289072271020434e-06,
+      "loss": 0.7578,
+      "step": 10352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2872661127284835e-06,
+      "loss": 0.7266,
+      "step": 10353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285460049956079e-06,
+      "loss": 0.7852,
+      "step": 10354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.283654082766817e-06,
+      "loss": 0.7188,
+      "step": 10355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.281848211224291e-06,
+      "loss": 0.9219,
+      "step": 10356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280042435392085e-06,
+      "loss": 0.8867,
+      "step": 10357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278236755333784e-06,
+      "loss": 0.875,
+      "step": 10358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27643117111297e-06,
+      "loss": 0.7617,
+      "step": 10359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27462568279322e-06,
+      "loss": 0.8047,
+      "step": 10360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272820290438107e-06,
+      "loss": 0.8242,
+      "step": 10361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.271014994111203e-06,
+      "loss": 0.9648,
+      "step": 10362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269209793876073e-06,
+      "loss": 0.9219,
+      "step": 10363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267404689796284e-06,
+      "loss": 0.8711,
+      "step": 10364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.265599681935396e-06,
+      "loss": 0.8359,
+      "step": 10365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263794770356962e-06,
+      "loss": 0.875,
+      "step": 10366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261989955124542e-06,
+      "loss": 0.9062,
+      "step": 10367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26018523630168e-06,
+      "loss": 0.9062,
+      "step": 10368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25838061395193e-06,
+      "loss": 0.9062,
+      "step": 10369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2565760881388285e-06,
+      "loss": 0.8047,
+      "step": 10370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2547716589259195e-06,
+      "loss": 0.8789,
+      "step": 10371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.252967326376738e-06,
+      "loss": 0.8594,
+      "step": 10372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251163090554822e-06,
+      "loss": 0.8906,
+      "step": 10373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249358951523693e-06,
+      "loss": 0.8359,
+      "step": 10374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247554909346886e-06,
+      "loss": 0.793,
+      "step": 10375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245750964087918e-06,
+      "loss": 0.8789,
+      "step": 10376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243947115810313e-06,
+      "loss": 0.9102,
+      "step": 10377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242143364577584e-06,
+      "loss": 0.8398,
+      "step": 10378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240339710453247e-06,
+      "loss": 0.9883,
+      "step": 10379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.238536153500807e-06,
+      "loss": 0.9414,
+      "step": 10380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236732693783774e-06,
+      "loss": 0.8516,
+      "step": 10381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234929331365652e-06,
+      "loss": 0.9609,
+      "step": 10382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.233126066309934e-06,
+      "loss": 0.9297,
+      "step": 10383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231322898680119e-06,
+      "loss": 0.9492,
+      "step": 10384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229519828539702e-06,
+      "loss": 0.8477,
+      "step": 10385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227716855952169e-06,
+      "loss": 0.7891,
+      "step": 10386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225913980981004e-06,
+      "loss": 0.8398,
+      "step": 10387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2241112036896924e-06,
+      "loss": 0.9648,
+      "step": 10388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.22230852414171e-06,
+      "loss": 0.8242,
+      "step": 10389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.220505942400537e-06,
+      "loss": 0.8047,
+      "step": 10390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218703458529637e-06,
+      "loss": 0.9375,
+      "step": 10391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2169010725924826e-06,
+      "loss": 0.7227,
+      "step": 10392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.215098784652541e-06,
+      "loss": 0.8398,
+      "step": 10393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213296594773271e-06,
+      "loss": 0.793,
+      "step": 10394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211494503018129e-06,
+      "loss": 0.832,
+      "step": 10395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2096925094505695e-06,
+      "loss": 0.9258,
+      "step": 10396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20789061413405e-06,
+      "loss": 0.9375,
+      "step": 10397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206088817132008e-06,
+      "loss": 0.8203,
+      "step": 10398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204287118507895e-06,
+      "loss": 0.8867,
+      "step": 10399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.202485518325147e-06,
+      "loss": 0.8711,
+      "step": 10400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200684016647208e-06,
+      "loss": 0.8672,
+      "step": 10401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1988826135375035e-06,
+      "loss": 0.9102,
+      "step": 10402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.197081309059466e-06,
+      "loss": 0.9844,
+      "step": 10403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195280103276524e-06,
+      "loss": 0.8047,
+      "step": 10404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1934789962521035e-06,
+      "loss": 0.7617,
+      "step": 10405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191677988049617e-06,
+      "loss": 0.793,
+      "step": 10406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.189877078732485e-06,
+      "loss": 0.7617,
+      "step": 10407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188076268364121e-06,
+      "loss": 0.9062,
+      "step": 10408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186275557007934e-06,
+      "loss": 0.9062,
+      "step": 10409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.184474944727326e-06,
+      "loss": 0.7773,
+      "step": 10410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.8711,
+      "step": 10411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1808740176464685e-06,
+      "loss": 0.918,
+      "step": 10412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.179073702973007e-06,
+      "loss": 0.8398,
+      "step": 10413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177273487628715e-06,
+      "loss": 0.8438,
+      "step": 10414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1754733716769845e-06,
+      "loss": 0.7188,
+      "step": 10415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1736733551812e-06,
+      "loss": 0.7148,
+      "step": 10416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1718734382047375e-06,
+      "loss": 0.7773,
+      "step": 10417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170073620810975e-06,
+      "loss": 0.9336,
+      "step": 10418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168273903063292e-06,
+      "loss": 0.8203,
+      "step": 10419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1664742850250626e-06,
+      "loss": 0.9062,
+      "step": 10420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1646747667596415e-06,
+      "loss": 0.8438,
+      "step": 10421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1628753483304e-06,
+      "loss": 0.8672,
+      "step": 10422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1610760298006995e-06,
+      "loss": 0.875,
+      "step": 10423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1592768112339e-06,
+      "loss": 1.0156,
+      "step": 10424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157477692693344e-06,
+      "loss": 1.0,
+      "step": 10425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155678674242388e-06,
+      "loss": 0.8242,
+      "step": 10426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153879755944377e-06,
+      "loss": 0.8164,
+      "step": 10427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1520809378626585e-06,
+      "loss": 0.8477,
+      "step": 10428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150282220060564e-06,
+      "loss": 0.9102,
+      "step": 10429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148483602601431e-06,
+      "loss": 0.7891,
+      "step": 10430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146685085548595e-06,
+      "loss": 0.7539,
+      "step": 10431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144886668965385e-06,
+      "loss": 0.9336,
+      "step": 10432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14308835291512e-06,
+      "loss": 0.8477,
+      "step": 10433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.141290137461125e-06,
+      "loss": 0.8906,
+      "step": 10434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13949202266672e-06,
+      "loss": 0.8359,
+      "step": 10435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13769400859522e-06,
+      "loss": 0.7109,
+      "step": 10436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135896095309929e-06,
+      "loss": 0.8086,
+      "step": 10437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13409828287416e-06,
+      "loss": 0.9141,
+      "step": 10438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132300571351214e-06,
+      "loss": 0.9297,
+      "step": 10439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130502960804397e-06,
+      "loss": 1.0391,
+      "step": 10440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128705451296998e-06,
+      "loss": 1.0625,
+      "step": 10441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1269080428923124e-06,
+      "loss": 0.9062,
+      "step": 10442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.125110735653631e-06,
+      "loss": 0.8984,
+      "step": 10443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1233135296442426e-06,
+      "loss": 0.8672,
+      "step": 10444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.121516424927424e-06,
+      "loss": 0.9766,
+      "step": 10445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119719421566456e-06,
+      "loss": 0.8594,
+      "step": 10446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117922519624612e-06,
+      "loss": 0.7344,
+      "step": 10447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1161257191651724e-06,
+      "loss": 0.8594,
+      "step": 10448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114329020251395e-06,
+      "loss": 0.8477,
+      "step": 10449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112532422946548e-06,
+      "loss": 0.9727,
+      "step": 10450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110735927313892e-06,
+      "loss": 0.8398,
+      "step": 10451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10893953341669e-06,
+      "loss": 0.8281,
+      "step": 10452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107143241318185e-06,
+      "loss": 0.8633,
+      "step": 10453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1053470510816346e-06,
+      "loss": 0.8555,
+      "step": 10454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.103550962770282e-06,
+      "loss": 0.7734,
+      "step": 10455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10175497644738e-06,
+      "loss": 0.8203,
+      "step": 10456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0999590921761516e-06,
+      "loss": 0.8594,
+      "step": 10457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0981633100198445e-06,
+      "loss": 0.8047,
+      "step": 10458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096367630041686e-06,
+      "loss": 0.8125,
+      "step": 10459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094572052304911e-06,
+      "loss": 0.7852,
+      "step": 10460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.092776576872736e-06,
+      "loss": 0.8242,
+      "step": 10461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090981203808387e-06,
+      "loss": 0.957,
+      "step": 10462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089185933175085e-06,
+      "loss": 0.6953,
+      "step": 10463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087390765036035e-06,
+      "loss": 0.8281,
+      "step": 10464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085595699454455e-06,
+      "loss": 0.9414,
+      "step": 10465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08380073649355e-06,
+      "loss": 0.8203,
+      "step": 10466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082005876216528e-06,
+      "loss": 0.8516,
+      "step": 10467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08021111868658e-06,
+      "loss": 0.9453,
+      "step": 10468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078416463966906e-06,
+      "loss": 0.8906,
+      "step": 10469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0766219121207005e-06,
+      "loss": 0.8438,
+      "step": 10470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.074827463211153e-06,
+      "loss": 0.8359,
+      "step": 10471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.073033117301444e-06,
+      "loss": 0.8164,
+      "step": 10472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07123887445476e-06,
+      "loss": 0.8359,
+      "step": 10473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069444734734275e-06,
+      "loss": 0.8438,
+      "step": 10474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.067650698203171e-06,
+      "loss": 0.8398,
+      "step": 10475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065856764924609e-06,
+      "loss": 0.7734,
+      "step": 10476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0640629349617604e-06,
+      "loss": 0.8281,
+      "step": 10477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.062269208377789e-06,
+      "loss": 0.8555,
+      "step": 10478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.06047558523586e-06,
+      "loss": 0.793,
+      "step": 10479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058682065599119e-06,
+      "loss": 0.8125,
+      "step": 10480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056888649530722e-06,
+      "loss": 0.832,
+      "step": 10481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055095337093822e-06,
+      "loss": 0.8477,
+      "step": 10482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053302128351567e-06,
+      "loss": 0.8555,
+      "step": 10483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051509023367087e-06,
+      "loss": 0.7695,
+      "step": 10484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049716022203529e-06,
+      "loss": 0.8398,
+      "step": 10485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0479231249240235e-06,
+      "loss": 0.8281,
+      "step": 10486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.046130331591707e-06,
+      "loss": 0.8477,
+      "step": 10487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0443376422697e-06,
+      "loss": 0.8867,
+      "step": 10488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042545057021126e-06,
+      "loss": 0.9141,
+      "step": 10489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040752575909109e-06,
+      "loss": 0.918,
+      "step": 10490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038960198996765e-06,
+      "loss": 0.8867,
+      "step": 10491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037167926347201e-06,
+      "loss": 0.832,
+      "step": 10492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03537575802353e-06,
+      "loss": 0.8242,
+      "step": 10493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033583694088855e-06,
+      "loss": 0.9141,
+      "step": 10494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.031791734606282e-06,
+      "loss": 0.918,
+      "step": 10495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0299998796389006e-06,
+      "loss": 0.7227,
+      "step": 10496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0282081292498096e-06,
+      "loss": 0.875,
+      "step": 10497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026416483502097e-06,
+      "loss": 0.875,
+      "step": 10498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.024624942458858e-06,
+      "loss": 0.8594,
+      "step": 10499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022833506183163e-06,
+      "loss": 0.8711,
+      "step": 10500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021042174738098e-06,
+      "loss": 0.918,
+      "step": 10501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019250948186738e-06,
+      "loss": 0.8047,
+      "step": 10502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017459826592156e-06,
+      "loss": 0.7539,
+      "step": 10503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.015668810017417e-06,
+      "loss": 0.8008,
+      "step": 10504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013877898525585e-06,
+      "loss": 0.8438,
+      "step": 10505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012087092179725e-06,
+      "loss": 0.8281,
+      "step": 10506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010296391042894e-06,
+      "loss": 0.8711,
+      "step": 10507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00850579517814e-06,
+      "loss": 0.9922,
+      "step": 10508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0067153046485145e-06,
+      "loss": 0.8594,
+      "step": 10509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004924919517067e-06,
+      "loss": 0.8359,
+      "step": 10510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0031346398468406e-06,
+      "loss": 0.9023,
+      "step": 10511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001344465700866e-06,
+      "loss": 0.7891,
+      "step": 10512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999554397142183e-06,
+      "loss": 0.8438,
+      "step": 10513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.997764434233825e-06,
+      "loss": 0.957,
+      "step": 10514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.995974577038815e-06,
+      "loss": 0.9062,
+      "step": 10515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994184825620178e-06,
+      "loss": 0.8203,
+      "step": 10516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992395180040934e-06,
+      "loss": 0.918,
+      "step": 10517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990605640364101e-06,
+      "loss": 0.8594,
+      "step": 10518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98881620665269e-06,
+      "loss": 0.8906,
+      "step": 10519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9870268789697075e-06,
+      "loss": 0.7969,
+      "step": 10520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985237657378161e-06,
+      "loss": 0.8945,
+      "step": 10521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983448541941056e-06,
+      "loss": 0.832,
+      "step": 10522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98165953272138e-06,
+      "loss": 0.7227,
+      "step": 10523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979870629782133e-06,
+      "loss": 0.7656,
+      "step": 10524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9780818331863055e-06,
+      "loss": 0.8086,
+      "step": 10525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976293142996883e-06,
+      "loss": 0.8672,
+      "step": 10526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974504559276844e-06,
+      "loss": 0.7656,
+      "step": 10527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.972716082089173e-06,
+      "loss": 0.8164,
+      "step": 10528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.970927711496844e-06,
+      "loss": 0.6797,
+      "step": 10529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9691394475628275e-06,
+      "loss": 0.8828,
+      "step": 10530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9673512903500886e-06,
+      "loss": 1.0234,
+      "step": 10531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965563239921593e-06,
+      "loss": 0.793,
+      "step": 10532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9637752963403026e-06,
+      "loss": 0.7188,
+      "step": 10533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961987459669173e-06,
+      "loss": 0.8594,
+      "step": 10534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960199729971152e-06,
+      "loss": 0.9141,
+      "step": 10535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9584121073091926e-06,
+      "loss": 0.7695,
+      "step": 10536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9566245917462415e-06,
+      "loss": 0.9609,
+      "step": 10537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954837183345238e-06,
+      "loss": 0.8516,
+      "step": 10538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9530498821691165e-06,
+      "loss": 0.8281,
+      "step": 10539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9512626882808135e-06,
+      "loss": 0.8125,
+      "step": 10540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.949475601743259e-06,
+      "loss": 0.8711,
+      "step": 10541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9476886226193785e-06,
+      "loss": 0.8672,
+      "step": 10542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.945901750972093e-06,
+      "loss": 0.793,
+      "step": 10543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944114986864325e-06,
+      "loss": 0.9219,
+      "step": 10544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942328330358984e-06,
+      "loss": 0.8164,
+      "step": 10545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940541781518986e-06,
+      "loss": 0.8203,
+      "step": 10546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938755340407233e-06,
+      "loss": 0.9141,
+      "step": 10547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936969007086633e-06,
+      "loss": 0.8164,
+      "step": 10548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9351827816200825e-06,
+      "loss": 0.875,
+      "step": 10549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93339666407048e-06,
+      "loss": 0.8984,
+      "step": 10550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931610654500713e-06,
+      "loss": 0.9141,
+      "step": 10551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929824752973675e-06,
+      "loss": 0.8086,
+      "step": 10552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928038959552246e-06,
+      "loss": 0.8516,
+      "step": 10553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926253274299311e-06,
+      "loss": 0.8672,
+      "step": 10554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924467697277743e-06,
+      "loss": 0.8633,
+      "step": 10555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922682228550415e-06,
+      "loss": 0.8789,
+      "step": 10556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920896868180195e-06,
+      "loss": 0.9688,
+      "step": 10557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919111616229954e-06,
+      "loss": 0.8438,
+      "step": 10558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.917326472762549e-06,
+      "loss": 0.7422,
+      "step": 10559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915541437840838e-06,
+      "loss": 0.9336,
+      "step": 10560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.913756511527674e-06,
+      "loss": 0.9219,
+      "step": 10561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.911971693885911e-06,
+      "loss": 0.7734,
+      "step": 10562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.910186984978393e-06,
+      "loss": 0.8125,
+      "step": 10563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908402384867958e-06,
+      "loss": 1.0391,
+      "step": 10564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906617893617451e-06,
+      "loss": 0.9102,
+      "step": 10565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9048335112897035e-06,
+      "loss": 0.8555,
+      "step": 10566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90304923794755e-06,
+      "loss": 1.0078,
+      "step": 10567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90126507365381e-06,
+      "loss": 0.7891,
+      "step": 10568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899481018471312e-06,
+      "loss": 0.9219,
+      "step": 10569
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8976970724628785e-06,
+      "loss": 0.7695,
+      "step": 10570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895913235691318e-06,
+      "loss": 0.8789,
+      "step": 10571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894129508219444e-06,
+      "loss": 0.8281,
+      "step": 10572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.892345890110066e-06,
+      "loss": 0.7969,
+      "step": 10573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890562381425991e-06,
+      "loss": 0.8398,
+      "step": 10574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888778982230011e-06,
+      "loss": 0.832,
+      "step": 10575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886995692584928e-06,
+      "loss": 0.7891,
+      "step": 10576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885212512553534e-06,
+      "loss": 0.9336,
+      "step": 10577
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.883429442198618e-06,
+      "loss": 0.7695,
+      "step": 10578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88164648158296e-06,
+      "loss": 0.8633,
+      "step": 10579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879863630769346e-06,
+      "loss": 0.7969,
+      "step": 10580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878080889820553e-06,
+      "loss": 1.0156,
+      "step": 10581
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876298258799349e-06,
+      "loss": 0.9258,
+      "step": 10582
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874515737768506e-06,
+      "loss": 0.875,
+      "step": 10583
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872733326790789e-06,
+      "loss": 0.8359,
+      "step": 10584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870951025928964e-06,
+      "loss": 0.832,
+      "step": 10585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869168835245783e-06,
+      "loss": 0.832,
+      "step": 10586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867386754804e-06,
+      "loss": 0.9062,
+      "step": 10587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865604784666365e-06,
+      "loss": 0.8711,
+      "step": 10588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.863822924895629e-06,
+      "loss": 0.8516,
+      "step": 10589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8620411755545276e-06,
+      "loss": 0.7695,
+      "step": 10590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8602595367058e-06,
+      "loss": 0.8164,
+      "step": 10591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858478008412183e-06,
+      "loss": 0.8359,
+      "step": 10592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.856696590736409e-06,
+      "loss": 0.8711,
+      "step": 10593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.854915283741198e-06,
+      "loss": 0.8398,
+      "step": 10594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853134087489275e-06,
+      "loss": 0.7578,
+      "step": 10595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85135300204336e-06,
+      "loss": 0.9023,
+      "step": 10596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849572027466171e-06,
+      "loss": 0.7383,
+      "step": 10597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847791163820412e-06,
+      "loss": 0.9258,
+      "step": 10598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846010411168793e-06,
+      "loss": 0.8789,
+      "step": 10599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844229769574016e-06,
+      "loss": 0.8477,
+      "step": 10600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842449239098787e-06,
+      "loss": 1.0547,
+      "step": 10601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84066881980579e-06,
+      "loss": 0.7969,
+      "step": 10602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838888511757722e-06,
+      "loss": 0.8594,
+      "step": 10603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837108315017271e-06,
+      "loss": 0.7383,
+      "step": 10604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835328229647125e-06,
+      "loss": 0.7852,
+      "step": 10605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833548255709952e-06,
+      "loss": 0.9297,
+      "step": 10606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8317683932684345e-06,
+      "loss": 1.125,
+      "step": 10607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829988642385243e-06,
+      "loss": 0.9023,
+      "step": 10608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828209003123051e-06,
+      "loss": 0.75,
+      "step": 10609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8264294755445124e-06,
+      "loss": 0.8086,
+      "step": 10610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824650059712292e-06,
+      "loss": 0.7383,
+      "step": 10611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822870755689046e-06,
+      "loss": 0.9375,
+      "step": 10612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821091563537431e-06,
+      "loss": 0.8828,
+      "step": 10613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819312483320087e-06,
+      "loss": 0.7969,
+      "step": 10614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8175335150996594e-06,
+      "loss": 0.8906,
+      "step": 10615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.81575465893879e-06,
+      "loss": 0.8281,
+      "step": 10616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813975914900121e-06,
+      "loss": 0.8516,
+      "step": 10617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8121972830462744e-06,
+      "loss": 0.8672,
+      "step": 10618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.810418763439883e-06,
+      "loss": 0.8945,
+      "step": 10619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8086403561435735e-06,
+      "loss": 0.7695,
+      "step": 10620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.806862061219968e-06,
+      "loss": 0.8906,
+      "step": 10621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805083878731674e-06,
+      "loss": 0.8164,
+      "step": 10622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803305808741311e-06,
+      "loss": 0.7891,
+      "step": 10623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801527851311485e-06,
+      "loss": 0.8516,
+      "step": 10624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799750006504806e-06,
+      "loss": 0.9062,
+      "step": 10625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7979722743838685e-06,
+      "loss": 0.8789,
+      "step": 10626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.796194655011269e-06,
+      "loss": 0.7617,
+      "step": 10627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.794417148449602e-06,
+      "loss": 0.8242,
+      "step": 10628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792639754761463e-06,
+      "loss": 0.9258,
+      "step": 10629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790862474009425e-06,
+      "loss": 0.793,
+      "step": 10630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789085306256074e-06,
+      "loss": 0.8984,
+      "step": 10631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.787308251563991e-06,
+      "loss": 0.875,
+      "step": 10632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785531309995741e-06,
+      "loss": 0.8672,
+      "step": 10633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.783754481613895e-06,
+      "loss": 0.8242,
+      "step": 10634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781977766481023e-06,
+      "loss": 0.8086,
+      "step": 10635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780201164659683e-06,
+      "loss": 0.8125,
+      "step": 10636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7784246762124294e-06,
+      "loss": 0.8164,
+      "step": 10637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776648301201817e-06,
+      "loss": 0.8125,
+      "step": 10638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7748720396903946e-06,
+      "loss": 0.8633,
+      "step": 10639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.773095891740711e-06,
+      "loss": 0.7891,
+      "step": 10640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7713198574153e-06,
+      "loss": 0.8438,
+      "step": 10641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7695439367767015e-06,
+      "loss": 0.8047,
+      "step": 10642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767768129887449e-06,
+      "loss": 0.8594,
+      "step": 10643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.765992436810074e-06,
+      "loss": 0.8711,
+      "step": 10644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764216857607096e-06,
+      "loss": 1.0234,
+      "step": 10645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.762441392341037e-06,
+      "loss": 0.8281,
+      "step": 10646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760666041074417e-06,
+      "loss": 0.8906,
+      "step": 10647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758890803869751e-06,
+      "loss": 0.8984,
+      "step": 10648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757115680789539e-06,
+      "loss": 0.7656,
+      "step": 10649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755340671896291e-06,
+      "loss": 0.8281,
+      "step": 10650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753565777252509e-06,
+      "loss": 0.8203,
+      "step": 10651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75179099692069e-06,
+      "loss": 0.832,
+      "step": 10652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750016330963324e-06,
+      "loss": 0.9219,
+      "step": 10653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748241779442899e-06,
+      "loss": 0.875,
+      "step": 10654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746467342421903e-06,
+      "loss": 0.8828,
+      "step": 10655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744693019962818e-06,
+      "loss": 0.9648,
+      "step": 10656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742918812128115e-06,
+      "loss": 0.7305,
+      "step": 10657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.741144718980269e-06,
+      "loss": 0.7969,
+      "step": 10658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73937074058175e-06,
+      "loss": 0.8867,
+      "step": 10659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7375968769950255e-06,
+      "loss": 0.9375,
+      "step": 10660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.735823128282547e-06,
+      "loss": 0.8164,
+      "step": 10661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734049494506776e-06,
+      "loss": 0.9062,
+      "step": 10662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732275975730166e-06,
+      "loss": 0.875,
+      "step": 10663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730502572015168e-06,
+      "loss": 0.7617,
+      "step": 10664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728729283424218e-06,
+      "loss": 0.9062,
+      "step": 10665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72695611001976e-06,
+      "loss": 0.8047,
+      "step": 10666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725183051864233e-06,
+      "loss": 0.832,
+      "step": 10667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.72341010902007e-06,
+      "loss": 0.832,
+      "step": 10668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.721637281549691e-06,
+      "loss": 0.8867,
+      "step": 10669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719864569515527e-06,
+      "loss": 0.8906,
+      "step": 10670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718091972979995e-06,
+      "loss": 0.8672,
+      "step": 10671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716319492005516e-06,
+      "loss": 0.875,
+      "step": 10672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714547126654494e-06,
+      "loss": 0.832,
+      "step": 10673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.712774876989341e-06,
+      "loss": 0.9023,
+      "step": 10674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711002743072459e-06,
+      "loss": 0.8008,
+      "step": 10675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709230724966252e-06,
+      "loss": 0.832,
+      "step": 10676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707458822733109e-06,
+      "loss": 0.9258,
+      "step": 10677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705687036435424e-06,
+      "loss": 0.7578,
+      "step": 10678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703915366135586e-06,
+      "loss": 0.8594,
+      "step": 10679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70214381189598e-06,
+      "loss": 0.8984,
+      "step": 10680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70037237377898e-06,
+      "loss": 0.8438,
+      "step": 10681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.698601051846961e-06,
+      "loss": 0.8516,
+      "step": 10682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.696829846162298e-06,
+      "loss": 0.9961,
+      "step": 10683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695058756787359e-06,
+      "loss": 0.8906,
+      "step": 10684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693287783784501e-06,
+      "loss": 0.8125,
+      "step": 10685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691516927216084e-06,
+      "loss": 0.7578,
+      "step": 10686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68974618714447e-06,
+      "loss": 0.8359,
+      "step": 10687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687975563632e-06,
+      "loss": 0.7461,
+      "step": 10688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686205056741023e-06,
+      "loss": 0.9141,
+      "step": 10689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684434666533883e-06,
+      "loss": 0.9258,
+      "step": 10690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682664393072922e-06,
+      "loss": 0.9492,
+      "step": 10691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680894236420464e-06,
+      "loss": 0.7852,
+      "step": 10692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679124196638847e-06,
+      "loss": 0.8477,
+      "step": 10693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.677354273790393e-06,
+      "loss": 0.8633,
+      "step": 10694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675584467937428e-06,
+      "loss": 0.918,
+      "step": 10695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673814779142265e-06,
+      "loss": 0.9375,
+      "step": 10696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672045207467218e-06,
+      "loss": 0.8633,
+      "step": 10697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6702757529746e-06,
+      "loss": 0.9844,
+      "step": 10698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.668506415726714e-06,
+      "loss": 0.8906,
+      "step": 10699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666737195785858e-06,
+      "loss": 0.9336,
+      "step": 10700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664968093214333e-06,
+      "loss": 0.8906,
+      "step": 10701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663199108074432e-06,
+      "loss": 0.918,
+      "step": 10702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6614302404284435e-06,
+      "loss": 0.8633,
+      "step": 10703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.659661490338648e-06,
+      "loss": 0.8594,
+      "step": 10704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65789285786733e-06,
+      "loss": 0.9453,
+      "step": 10705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6561243430767645e-06,
+      "loss": 0.9219,
+      "step": 10706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6543559460292254e-06,
+      "loss": 0.9336,
+      "step": 10707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652587666786976e-06,
+      "loss": 0.9062,
+      "step": 10708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650819505412285e-06,
+      "loss": 0.875,
+      "step": 10709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64905146196741e-06,
+      "loss": 0.8359,
+      "step": 10710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647283536514609e-06,
+      "loss": 0.8945,
+      "step": 10711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645515729116129e-06,
+      "loss": 0.8086,
+      "step": 10712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6437480398342215e-06,
+      "loss": 0.918,
+      "step": 10713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641980468731124e-06,
+      "loss": 0.9336,
+      "step": 10714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640213015869083e-06,
+      "loss": 0.8945,
+      "step": 10715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638445681310326e-06,
+      "loss": 0.832,
+      "step": 10716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6366784651170875e-06,
+      "loss": 0.8203,
+      "step": 10717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634911367351593e-06,
+      "loss": 0.8555,
+      "step": 10718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.633144388076067e-06,
+      "loss": 0.8789,
+      "step": 10719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631377527352724e-06,
+      "loss": 0.8633,
+      "step": 10720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629610785243781e-06,
+      "loss": 0.7891,
+      "step": 10721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.627844161811443e-06,
+      "loss": 0.8711,
+      "step": 10722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626077657117921e-06,
+      "loss": 0.8828,
+      "step": 10723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624311271225412e-06,
+      "loss": 0.8281,
+      "step": 10724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622545004196117e-06,
+      "loss": 0.8789,
+      "step": 10725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620778856092227e-06,
+      "loss": 0.9141,
+      "step": 10726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619012826975931e-06,
+      "loss": 0.9766,
+      "step": 10727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617246916909415e-06,
+      "loss": 0.8828,
+      "step": 10728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615481125954855e-06,
+      "loss": 0.8906,
+      "step": 10729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.61371545417443e-06,
+      "loss": 0.9688,
+      "step": 10730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611949901630314e-06,
+      "loss": 0.9375,
+      "step": 10731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610184468384675e-06,
+      "loss": 0.9062,
+      "step": 10732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.608419154499672e-06,
+      "loss": 0.8281,
+      "step": 10733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606653960037466e-06,
+      "loss": 0.8594,
+      "step": 10734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604888885060217e-06,
+      "loss": 0.8984,
+      "step": 10735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603123929630071e-06,
+      "loss": 0.8047,
+      "step": 10736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6013590938091746e-06,
+      "loss": 1.0938,
+      "step": 10737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599594377659672e-06,
+      "loss": 0.8594,
+      "step": 10738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.597829781243704e-06,
+      "loss": 0.7969,
+      "step": 10739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5960653046234025e-06,
+      "loss": 0.8711,
+      "step": 10740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5943009478608945e-06,
+      "loss": 0.7891,
+      "step": 10741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.592536711018309e-06,
+      "loss": 0.7812,
+      "step": 10742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590772594157771e-06,
+      "loss": 0.9297,
+      "step": 10743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5890085973413895e-06,
+      "loss": 0.8711,
+      "step": 10744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.587244720631282e-06,
+      "loss": 0.9102,
+      "step": 10745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585480964089559e-06,
+      "loss": 0.8008,
+      "step": 10746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5837173277783225e-06,
+      "loss": 0.8555,
+      "step": 10747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5819538117596735e-06,
+      "loss": 0.8281,
+      "step": 10748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580190416095707e-06,
+      "loss": 0.9297,
+      "step": 10749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578427140848519e-06,
+      "loss": 0.8438,
+      "step": 10750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.576663986080194e-06,
+      "loss": 0.9688,
+      "step": 10751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574900951852813e-06,
+      "loss": 0.8477,
+      "step": 10752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573138038228458e-06,
+      "loss": 0.7617,
+      "step": 10753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5713752452692084e-06,
+      "loss": 0.918,
+      "step": 10754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569612573037125e-06,
+      "loss": 0.793,
+      "step": 10755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56785002159428e-06,
+      "loss": 0.8945,
+      "step": 10756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566087591002735e-06,
+      "loss": 0.8594,
+      "step": 10757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564325281324553e-06,
+      "loss": 0.8086,
+      "step": 10758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562563092621776e-06,
+      "loss": 0.8203,
+      "step": 10759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560801024956462e-06,
+      "loss": 0.9062,
+      "step": 10760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559039078390654e-06,
+      "loss": 0.8555,
+      "step": 10761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557277252986396e-06,
+      "loss": 0.8711,
+      "step": 10762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5555155488057175e-06,
+      "loss": 0.8555,
+      "step": 10763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553753965910655e-06,
+      "loss": 0.8125,
+      "step": 10764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551992504363236e-06,
+      "loss": 0.9727,
+      "step": 10765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550231164225489e-06,
+      "loss": 0.9883,
+      "step": 10766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.548469945559426e-06,
+      "loss": 0.8555,
+      "step": 10767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.546708848427063e-06,
+      "loss": 0.8281,
+      "step": 10768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544947872890416e-06,
+      "loss": 0.8516,
+      "step": 10769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5431870190114925e-06,
+      "loss": 0.8594,
+      "step": 10770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541426286852286e-06,
+      "loss": 0.8516,
+      "step": 10771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5396656764748e-06,
+      "loss": 0.9141,
+      "step": 10772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.537905187941029e-06,
+      "loss": 0.9844,
+      "step": 10773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.536144821312967e-06,
+      "loss": 0.9648,
+      "step": 10774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5343845766525885e-06,
+      "loss": 0.7656,
+      "step": 10775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53262445402188e-06,
+      "loss": 0.8516,
+      "step": 10776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.530864453482818e-06,
+      "loss": 0.8984,
+      "step": 10777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529104575097379e-06,
+      "loss": 0.8555,
+      "step": 10778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527344818927524e-06,
+      "loss": 0.9648,
+      "step": 10779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.525585185035218e-06,
+      "loss": 0.8203,
+      "step": 10780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523825673482422e-06,
+      "loss": 0.8516,
+      "step": 10781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522066284331097e-06,
+      "loss": 0.8789,
+      "step": 10782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.520307017643184e-06,
+      "loss": 0.9727,
+      "step": 10783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518547873480632e-06,
+      "loss": 0.8867,
+      "step": 10784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.516788851905385e-06,
+      "loss": 0.9258,
+      "step": 10785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515029952979386e-06,
+      "loss": 0.8008,
+      "step": 10786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.513271176764557e-06,
+      "loss": 0.8516,
+      "step": 10787
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511512523322836e-06,
+      "loss": 0.8906,
+      "step": 10788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.509753992716143e-06,
+      "loss": 0.7891,
+      "step": 10789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507995585006406e-06,
+      "loss": 0.8984,
+      "step": 10790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506237300255531e-06,
+      "loss": 0.9727,
+      "step": 10791
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504479138525437e-06,
+      "loss": 0.8594,
+      "step": 10792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5027210998780284e-06,
+      "loss": 0.6875,
+      "step": 10793
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500963184375213e-06,
+      "loss": 0.9258,
+      "step": 10794
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499205392078883e-06,
+      "loss": 0.8398,
+      "step": 10795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497447723050937e-06,
+      "loss": 1.0234,
+      "step": 10796
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495690177353264e-06,
+      "loss": 0.8438,
+      "step": 10797
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493932755047754e-06,
+      "loss": 0.875,
+      "step": 10798
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4921754561962814e-06,
+      "loss": 0.9023,
+      "step": 10799
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490418280860728e-06,
+      "loss": 0.8359,
+      "step": 10800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488661229102968e-06,
+      "loss": 0.8906,
+      "step": 10801
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486904300984865e-06,
+      "loss": 0.9219,
+      "step": 10802
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.485147496568285e-06,
+      "loss": 0.8984,
+      "step": 10803
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483390815915089e-06,
+      "loss": 0.8281,
+      "step": 10804
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.481634259087136e-06,
+      "loss": 0.8828,
+      "step": 10805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479877826146269e-06,
+      "loss": 0.8164,
+      "step": 10806
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.478121517154339e-06,
+      "loss": 0.9062,
+      "step": 10807
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.476365332173188e-06,
+      "loss": 0.9297,
+      "step": 10808
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474609271264658e-06,
+      "loss": 0.8555,
+      "step": 10809
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472853334490574e-06,
+      "loss": 0.9102,
+      "step": 10810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4710975219127716e-06,
+      "loss": 0.8203,
+      "step": 10811
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469341833593073e-06,
+      "loss": 0.8203,
+      "step": 10812
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467586269593305e-06,
+      "loss": 0.7031,
+      "step": 10813
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465830829975273e-06,
+      "loss": 0.8828,
+      "step": 10814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464075514800794e-06,
+      "loss": 0.7422,
+      "step": 10815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462320324131677e-06,
+      "loss": 0.9141,
+      "step": 10816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460565258029726e-06,
+      "loss": 0.8594,
+      "step": 10817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458810316556734e-06,
+      "loss": 0.8906,
+      "step": 10818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457055499774498e-06,
+      "loss": 0.8945,
+      "step": 10819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.455300807744809e-06,
+      "loss": 0.7734,
+      "step": 10820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453546240529455e-06,
+      "loss": 0.8906,
+      "step": 10821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.451791798190209e-06,
+      "loss": 1.0625,
+      "step": 10822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450037480788853e-06,
+      "loss": 0.7812,
+      "step": 10823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4482832883871585e-06,
+      "loss": 0.7539,
+      "step": 10824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.446529221046897e-06,
+      "loss": 0.8398,
+      "step": 10825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444775278829825e-06,
+      "loss": 0.7578,
+      "step": 10826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.443021461797702e-06,
+      "loss": 0.8203,
+      "step": 10827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441267770012287e-06,
+      "loss": 0.8867,
+      "step": 10828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439514203535333e-06,
+      "loss": 0.9727,
+      "step": 10829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437760762428578e-06,
+      "loss": 0.8555,
+      "step": 10830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4360074467537644e-06,
+      "loss": 0.8359,
+      "step": 10831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434254256572633e-06,
+      "loss": 0.875,
+      "step": 10832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432501191946918e-06,
+      "loss": 0.8828,
+      "step": 10833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43074825293834e-06,
+      "loss": 0.8242,
+      "step": 10834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428995439608628e-06,
+      "loss": 0.7578,
+      "step": 10835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427242752019499e-06,
+      "loss": 0.8438,
+      "step": 10836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.425490190232674e-06,
+      "loss": 0.9297,
+      "step": 10837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4237377543098535e-06,
+      "loss": 0.8008,
+      "step": 10838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42198544431275e-06,
+      "loss": 0.8008,
+      "step": 10839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4202332603030615e-06,
+      "loss": 1.0625,
+      "step": 10840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4184812023424925e-06,
+      "loss": 0.875,
+      "step": 10841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.416729270492725e-06,
+      "loss": 0.8672,
+      "step": 10842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414977464815453e-06,
+      "loss": 0.8594,
+      "step": 10843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4132257853723605e-06,
+      "loss": 0.8867,
+      "step": 10844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411474232225129e-06,
+      "loss": 0.8008,
+      "step": 10845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409722805435427e-06,
+      "loss": 1.0156,
+      "step": 10846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.407971505064927e-06,
+      "loss": 0.9336,
+      "step": 10847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406220331175299e-06,
+      "loss": 0.8633,
+      "step": 10848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.404469283828204e-06,
+      "loss": 0.8242,
+      "step": 10849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402718363085294e-06,
+      "loss": 0.7148,
+      "step": 10850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400967569008223e-06,
+      "loss": 0.9219,
+      "step": 10851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3992169016586425e-06,
+      "loss": 0.7461,
+      "step": 10852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397466361098198e-06,
+      "loss": 0.8867,
+      "step": 10853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395715947388521e-06,
+      "loss": 0.9258,
+      "step": 10854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393965660591251e-06,
+      "loss": 0.8672,
+      "step": 10855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392215500768022e-06,
+      "loss": 0.7891,
+      "step": 10856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39046546798045e-06,
+      "loss": 0.8984,
+      "step": 10857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388715562290163e-06,
+      "loss": 0.8164,
+      "step": 10858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386965783758777e-06,
+      "loss": 0.9766,
+      "step": 10859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385216132447909e-06,
+      "loss": 0.875,
+      "step": 10860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383466608419157e-06,
+      "loss": 0.7656,
+      "step": 10861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3817172117341295e-06,
+      "loss": 0.8242,
+      "step": 10862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.379967942454426e-06,
+      "loss": 0.8633,
+      "step": 10863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.378218800641644e-06,
+      "loss": 0.8086,
+      "step": 10864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.376469786357368e-06,
+      "loss": 0.8203,
+      "step": 10865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374720899663183e-06,
+      "loss": 0.832,
+      "step": 10866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372972140620674e-06,
+      "loss": 0.9492,
+      "step": 10867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37122350929142e-06,
+      "loss": 0.7031,
+      "step": 10868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.369475005736984e-06,
+      "loss": 0.8984,
+      "step": 10869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36772663001894e-06,
+      "loss": 0.9062,
+      "step": 10870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365978382198852e-06,
+      "loss": 0.7773,
+      "step": 10871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364230262338274e-06,
+      "loss": 0.793,
+      "step": 10872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.362482270498762e-06,
+      "loss": 0.7852,
+      "step": 10873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360734406741864e-06,
+      "loss": 0.9219,
+      "step": 10874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358986671129129e-06,
+      "loss": 0.8516,
+      "step": 10875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357239063722096e-06,
+      "loss": 0.9297,
+      "step": 10876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.355491584582297e-06,
+      "loss": 1.0469,
+      "step": 10877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.353744233771266e-06,
+      "loss": 0.9141,
+      "step": 10878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351997011350533e-06,
+      "loss": 0.8125,
+      "step": 10879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350249917381619e-06,
+      "loss": 0.9453,
+      "step": 10880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348502951926036e-06,
+      "loss": 0.8477,
+      "step": 10881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.346756115045304e-06,
+      "loss": 0.8711,
+      "step": 10882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3450094068009305e-06,
+      "loss": 0.7656,
+      "step": 10883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343262827254421e-06,
+      "loss": 0.8125,
+      "step": 10884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.341516376467269e-06,
+      "loss": 0.8438,
+      "step": 10885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3397700545009775e-06,
+      "loss": 0.7617,
+      "step": 10886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338023861417031e-06,
+      "loss": 0.8242,
+      "step": 10887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336277797276921e-06,
+      "loss": 0.8359,
+      "step": 10888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3345318621421235e-06,
+      "loss": 0.918,
+      "step": 10889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332786056074123e-06,
+      "loss": 0.8438,
+      "step": 10890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3310403791343835e-06,
+      "loss": 0.8438,
+      "step": 10891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329294831384379e-06,
+      "loss": 0.8203,
+      "step": 10892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32754941288557e-06,
+      "loss": 0.8086,
+      "step": 10893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3258041236994174e-06,
+      "loss": 0.8164,
+      "step": 10894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324058963887374e-06,
+      "loss": 0.8359,
+      "step": 10895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3223139335108905e-06,
+      "loss": 0.7031,
+      "step": 10896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.320569032631412e-06,
+      "loss": 0.9648,
+      "step": 10897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318824261310379e-06,
+      "loss": 0.9141,
+      "step": 10898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317079619609226e-06,
+      "loss": 0.9219,
+      "step": 10899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3153351075893874e-06,
+      "loss": 0.8867,
+      "step": 10900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3135907253122885e-06,
+      "loss": 0.7617,
+      "step": 10901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3118464728393515e-06,
+      "loss": 0.8516,
+      "step": 10902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310102350231995e-06,
+      "loss": 0.9844,
+      "step": 10903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308358357551632e-06,
+      "loss": 0.832,
+      "step": 10904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306614494859672e-06,
+      "loss": 0.8047,
+      "step": 10905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.304870762217516e-06,
+      "loss": 0.8945,
+      "step": 10906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303127159686567e-06,
+      "loss": 0.875,
+      "step": 10907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301383687328219e-06,
+      "loss": 0.7812,
+      "step": 10908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299640345203862e-06,
+      "loss": 0.8398,
+      "step": 10909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297897133374881e-06,
+      "loss": 0.8438,
+      "step": 10910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296154051902656e-06,
+      "loss": 0.8828,
+      "step": 10911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.29441110084857e-06,
+      "loss": 0.8594,
+      "step": 10912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292668280273991e-06,
+      "loss": 0.8984,
+      "step": 10913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290925590240281e-06,
+      "loss": 0.9219,
+      "step": 10914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.289183030808812e-06,
+      "loss": 0.8867,
+      "step": 10915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287440602040937e-06,
+      "loss": 0.9961,
+      "step": 10916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285698303998009e-06,
+      "loss": 0.8906,
+      "step": 10917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.283956136741378e-06,
+      "loss": 0.8438,
+      "step": 10918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2822141003323915e-06,
+      "loss": 0.8516,
+      "step": 10919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280472194832385e-06,
+      "loss": 0.8125,
+      "step": 10920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278730420302697e-06,
+      "loss": 0.7695,
+      "step": 10921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2769887768046535e-06,
+      "loss": 0.7383,
+      "step": 10922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275247264399587e-06,
+      "loss": 0.8633,
+      "step": 10923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273505883148815e-06,
+      "loss": 1.0078,
+      "step": 10924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271764633113653e-06,
+      "loss": 0.8867,
+      "step": 10925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270023514355413e-06,
+      "loss": 0.8711,
+      "step": 10926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26828252693541e-06,
+      "loss": 0.8398,
+      "step": 10927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266541670914936e-06,
+      "loss": 0.8867,
+      "step": 10928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.264800946355294e-06,
+      "loss": 0.8906,
+      "step": 10929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263060353317778e-06,
+      "loss": 0.9375,
+      "step": 10930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261319891863682e-06,
+      "loss": 1.0703,
+      "step": 10931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259579562054279e-06,
+      "loss": 0.9062,
+      "step": 10932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2578393639508565e-06,
+      "loss": 0.957,
+      "step": 10933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256099297614687e-06,
+      "loss": 0.9219,
+      "step": 10934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254359363107046e-06,
+      "loss": 0.9492,
+      "step": 10935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252619560489192e-06,
+      "loss": 0.875,
+      "step": 10936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250879889822388e-06,
+      "loss": 0.9219,
+      "step": 10937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249140351167892e-06,
+      "loss": 0.8672,
+      "step": 10938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24740094458696e-06,
+      "loss": 0.7617,
+      "step": 10939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2456616701408305e-06,
+      "loss": 0.8945,
+      "step": 10940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243922527890749e-06,
+      "loss": 0.9023,
+      "step": 10941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242183517897956e-06,
+      "loss": 0.7148,
+      "step": 10942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240444640223688e-06,
+      "loss": 0.8672,
+      "step": 10943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238705894929164e-06,
+      "loss": 0.7734,
+      "step": 10944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236967282075612e-06,
+      "loss": 0.7969,
+      "step": 10945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235228801724254e-06,
+      "loss": 0.8047,
+      "step": 10946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2334904539363065e-06,
+      "loss": 0.7852,
+      "step": 10947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23175223877297e-06,
+      "loss": 0.8086,
+      "step": 10948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230014156295457e-06,
+      "loss": 0.832,
+      "step": 10949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228276206564965e-06,
+      "loss": 0.8125,
+      "step": 10950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226538389642695e-06,
+      "loss": 0.8711,
+      "step": 10951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224800705589833e-06,
+      "loss": 0.9062,
+      "step": 10952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223063154467565e-06,
+      "loss": 0.7812,
+      "step": 10953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221325736337076e-06,
+      "loss": 0.8281,
+      "step": 10954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.219588451259545e-06,
+      "loss": 0.8242,
+      "step": 10955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217851299296138e-06,
+      "loss": 0.8945,
+      "step": 10956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2161142805080264e-06,
+      "loss": 0.793,
+      "step": 10957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214377394956373e-06,
+      "loss": 0.9102,
+      "step": 10958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212640642702341e-06,
+      "loss": 0.8242,
+      "step": 10959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.210904023807076e-06,
+      "loss": 0.7891,
+      "step": 10960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209167538331731e-06,
+      "loss": 0.9102,
+      "step": 10961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207431186337448e-06,
+      "loss": 0.7227,
+      "step": 10962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.205694967885374e-06,
+      "loss": 0.8672,
+      "step": 10963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.203958883036634e-06,
+      "loss": 0.8008,
+      "step": 10964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202222931852364e-06,
+      "loss": 0.9062,
+      "step": 10965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200487114393687e-06,
+      "loss": 0.8945,
+      "step": 10966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19875143072173e-06,
+      "loss": 0.8828,
+      "step": 10967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197015880897599e-06,
+      "loss": 0.75,
+      "step": 10968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.195280464982412e-06,
+      "loss": 0.8594,
+      "step": 10969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193545183037277e-06,
+      "loss": 0.8359,
+      "step": 10970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191810035123291e-06,
+      "loss": 0.7773,
+      "step": 10971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1900750213015515e-06,
+      "loss": 0.8438,
+      "step": 10972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188340141633155e-06,
+      "loss": 0.8516,
+      "step": 10973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1866053961791905e-06,
+      "loss": 0.832,
+      "step": 10974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.184870785000734e-06,
+      "loss": 0.7969,
+      "step": 10975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183136308158867e-06,
+      "loss": 0.832,
+      "step": 10976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181401965714664e-06,
+      "loss": 0.7695,
+      "step": 10977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179667757729197e-06,
+      "loss": 0.8359,
+      "step": 10978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177933684263524e-06,
+      "loss": 0.8867,
+      "step": 10979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.176199745378708e-06,
+      "loss": 0.9922,
+      "step": 10980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.174465941135801e-06,
+      "loss": 0.7617,
+      "step": 10981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.172732271595861e-06,
+      "loss": 0.8594,
+      "step": 10982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170998736819921e-06,
+      "loss": 0.8398,
+      "step": 10983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.169265336869028e-06,
+      "loss": 0.8945,
+      "step": 10984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167532071804218e-06,
+      "loss": 0.8398,
+      "step": 10985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165798941686524e-06,
+      "loss": 0.9141,
+      "step": 10986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164065946576965e-06,
+      "loss": 0.7578,
+      "step": 10987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.162333086536567e-06,
+      "loss": 0.9102,
+      "step": 10988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160600361626348e-06,
+      "loss": 0.8672,
+      "step": 10989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158867771907319e-06,
+      "loss": 0.7461,
+      "step": 10990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.157135317440486e-06,
+      "loss": 0.7188,
+      "step": 10991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.15540299828685e-06,
+      "loss": 0.9336,
+      "step": 10992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153670814507411e-06,
+      "loss": 0.8711,
+      "step": 10993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1519387661631645e-06,
+      "loss": 0.8359,
+      "step": 10994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150206853315091e-06,
+      "loss": 1.0312,
+      "step": 10995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1484750760241806e-06,
+      "loss": 0.7891,
+      "step": 10996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1467434343514075e-06,
+      "loss": 0.7812,
+      "step": 10997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1450119283577525e-06,
+      "loss": 0.875,
+      "step": 10998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143280558104174e-06,
+      "loss": 0.7812,
+      "step": 10999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1415493236516435e-06,
+      "loss": 0.8906,
+      "step": 11000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139818225061118e-06,
+      "loss": 0.9727,
+      "step": 11001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138087262393557e-06,
+      "loss": 0.7891,
+      "step": 11002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136356435709902e-06,
+      "loss": 0.8711,
+      "step": 11003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1346257450711035e-06,
+      "loss": 0.7773,
+      "step": 11004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1328951905381004e-06,
+      "loss": 0.8047,
+      "step": 11005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131164772171831e-06,
+      "loss": 0.9492,
+      "step": 11006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129434490033221e-06,
+      "loss": 0.8789,
+      "step": 11007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127704344183198e-06,
+      "loss": 0.875,
+      "step": 11008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1259743346826846e-06,
+      "loss": 0.9453,
+      "step": 11009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1242444615926e-06,
+      "loss": 0.9297,
+      "step": 11010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122514724973847e-06,
+      "loss": 0.8828,
+      "step": 11011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120785124887339e-06,
+      "loss": 0.6836,
+      "step": 11012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119055661393976e-06,
+      "loss": 0.8281,
+      "step": 11013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117326334554658e-06,
+      "loss": 0.9102,
+      "step": 11014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115597144430271e-06,
+      "loss": 0.8125,
+      "step": 11015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113868091081707e-06,
+      "loss": 0.8125,
+      "step": 11016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.112139174569847e-06,
+      "loss": 0.7461,
+      "step": 11017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.110410394955573e-06,
+      "loss": 0.9023,
+      "step": 11018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108681752299751e-06,
+      "loss": 0.8086,
+      "step": 11019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106953246663253e-06,
+      "loss": 0.7891,
+      "step": 11020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1052248781069416e-06,
+      "loss": 0.832,
+      "step": 11021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.10349664669168e-06,
+      "loss": 0.8555,
+      "step": 11022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101768552478314e-06,
+      "loss": 0.832,
+      "step": 11023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100040595527696e-06,
+      "loss": 0.8438,
+      "step": 11024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098312775900673e-06,
+      "loss": 0.9258,
+      "step": 11025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096585093658079e-06,
+      "loss": 0.9297,
+      "step": 11026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0948575488607486e-06,
+      "loss": 0.8398,
+      "step": 11027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0931301415695154e-06,
+      "loss": 0.8203,
+      "step": 11028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091402871845206e-06,
+      "loss": 0.8984,
+      "step": 11029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08967573974863e-06,
+      "loss": 0.9297,
+      "step": 11030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.087948745340611e-06,
+      "loss": 0.9102,
+      "step": 11031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.086221888681958e-06,
+      "loss": 0.918,
+      "step": 11032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0844951698334775e-06,
+      "loss": 0.9922,
+      "step": 11033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082768588855963e-06,
+      "loss": 0.8828,
+      "step": 11034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081042145810216e-06,
+      "loss": 0.9141,
+      "step": 11035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079315840757025e-06,
+      "loss": 0.9844,
+      "step": 11036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077589673757181e-06,
+      "loss": 0.8164,
+      "step": 11037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075863644871456e-06,
+      "loss": 0.8633,
+      "step": 11038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.074137754160632e-06,
+      "loss": 0.7188,
+      "step": 11039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0724120016854795e-06,
+      "loss": 0.8984,
+      "step": 11040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.070686387506768e-06,
+      "loss": 0.8672,
+      "step": 11041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068960911685251e-06,
+      "loss": 0.8867,
+      "step": 11042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06723557428169e-06,
+      "loss": 0.8477,
+      "step": 11043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06551037535684e-06,
+      "loss": 0.8477,
+      "step": 11044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063785314971443e-06,
+      "loss": 0.7578,
+      "step": 11045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062060393186241e-06,
+      "loss": 0.9141,
+      "step": 11046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060335610061973e-06,
+      "loss": 0.8008,
+      "step": 11047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0586109656593726e-06,
+      "loss": 0.8008,
+      "step": 11048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056886460039166e-06,
+      "loss": 0.918,
+      "step": 11049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0551620932620725e-06,
+      "loss": 0.8789,
+      "step": 11050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0534378653888135e-06,
+      "loss": 0.8438,
+      "step": 11051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0517137764801025e-06,
+      "loss": 0.875,
+      "step": 11052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049989826596645e-06,
+      "loss": 1.1016,
+      "step": 11053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048266015799144e-06,
+      "loss": 0.7812,
+      "step": 11054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046542344148296e-06,
+      "loss": 0.8398,
+      "step": 11055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0448188117048e-06,
+      "loss": 0.8281,
+      "step": 11056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043095418529339e-06,
+      "loss": 0.7539,
+      "step": 11057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0413721646825964e-06,
+      "loss": 0.7617,
+      "step": 11058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039649050225254e-06,
+      "loss": 0.7773,
+      "step": 11059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0379260752179815e-06,
+      "loss": 0.9688,
+      "step": 11060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03620323972145e-06,
+      "loss": 0.8398,
+      "step": 11061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034480543796322e-06,
+      "loss": 0.8281,
+      "step": 11062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032757987503259e-06,
+      "loss": 0.8008,
+      "step": 11063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0310355709029104e-06,
+      "loss": 0.8477,
+      "step": 11064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029313294055928e-06,
+      "loss": 0.7695,
+      "step": 11065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027591157022954e-06,
+      "loss": 0.9258,
+      "step": 11066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0258691598646316e-06,
+      "loss": 0.8086,
+      "step": 11067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02414730264159e-06,
+      "loss": 0.7617,
+      "step": 11068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022425585414462e-06,
+      "loss": 0.8906,
+      "step": 11069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.020704008243868e-06,
+      "loss": 0.9258,
+      "step": 11070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018982571190432e-06,
+      "loss": 0.8711,
+      "step": 11071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017261274314765e-06,
+      "loss": 0.8281,
+      "step": 11072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015540117677481e-06,
+      "loss": 0.8672,
+      "step": 11073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01381910133918e-06,
+      "loss": 0.7812,
+      "step": 11074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012098225360461e-06,
+      "loss": 0.8242,
+      "step": 11075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010377489801921e-06,
+      "loss": 0.9727,
+      "step": 11076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008656894724152e-06,
+      "loss": 0.9258,
+      "step": 11077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.006936440187735e-06,
+      "loss": 0.7812,
+      "step": 11078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00521612625325e-06,
+      "loss": 0.8867,
+      "step": 11079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003495952981273e-06,
+      "loss": 0.8672,
+      "step": 11080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001775920432378e-06,
+      "loss": 0.9297,
+      "step": 11081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000056028667123e-06,
+      "loss": 0.8281,
+      "step": 11082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.998336277746071e-06,
+      "loss": 0.7891,
+      "step": 11083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996616667729779e-06,
+      "loss": 0.8164,
+      "step": 11084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994897198678793e-06,
+      "loss": 0.8438,
+      "step": 11085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993177870653662e-06,
+      "loss": 0.8789,
+      "step": 11086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991458683714923e-06,
+      "loss": 0.8164,
+      "step": 11087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.989739637923115e-06,
+      "loss": 1.0391,
+      "step": 11088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988020733338767e-06,
+      "loss": 0.8984,
+      "step": 11089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986301970022401e-06,
+      "loss": 0.918,
+      "step": 11090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.984583348034541e-06,
+      "loss": 0.9414,
+      "step": 11091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982864867435701e-06,
+      "loss": 0.8047,
+      "step": 11092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9811465282863924e-06,
+      "loss": 0.7812,
+      "step": 11093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979428330647117e-06,
+      "loss": 0.7383,
+      "step": 11094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977710274578379e-06,
+      "loss": 0.8125,
+      "step": 11095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975992360140673e-06,
+      "loss": 0.7969,
+      "step": 11096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974274587394491e-06,
+      "loss": 0.8984,
+      "step": 11097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972556956400312e-06,
+      "loss": 0.8086,
+      "step": 11098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.970839467218623e-06,
+      "loss": 0.8438,
+      "step": 11099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9691221199099e-06,
+      "loss": 0.8164,
+      "step": 11100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.967404914534608e-06,
+      "loss": 0.875,
+      "step": 11101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965687851153214e-06,
+      "loss": 0.8281,
+      "step": 11102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963970929826181e-06,
+      "loss": 0.7734,
+      "step": 11103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962254150613966e-06,
+      "loss": 0.9141,
+      "step": 11104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960537513577014e-06,
+      "loss": 0.9219,
+      "step": 11105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958821018775773e-06,
+      "loss": 0.9297,
+      "step": 11106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957104666270684e-06,
+      "loss": 0.7891,
+      "step": 11107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955388456122184e-06,
+      "loss": 0.8789,
+      "step": 11108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.953672388390698e-06,
+      "loss": 0.8281,
+      "step": 11109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9519564631366544e-06,
+      "loss": 0.8789,
+      "step": 11110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9502406804204735e-06,
+      "loss": 0.8672,
+      "step": 11111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948525040302577e-06,
+      "loss": 0.8867,
+      "step": 11112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946809542843363e-06,
+      "loss": 0.8633,
+      "step": 11113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945094188103245e-06,
+      "loss": 0.7344,
+      "step": 11114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9433789761426205e-06,
+      "loss": 0.9766,
+      "step": 11115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9416639070218906e-06,
+      "loss": 0.7812,
+      "step": 11116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.939948980801435e-06,
+      "loss": 0.8008,
+      "step": 11117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938234197541646e-06,
+      "loss": 0.7578,
+      "step": 11118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936519557302901e-06,
+      "loss": 0.957,
+      "step": 11119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934805060145581e-06,
+      "loss": 0.7773,
+      "step": 11120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933090706130047e-06,
+      "loss": 0.8438,
+      "step": 11121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93137649531667e-06,
+      "loss": 0.7539,
+      "step": 11122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929662427765808e-06,
+      "loss": 0.9531,
+      "step": 11123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927948503537819e-06,
+      "loss": 0.7852,
+      "step": 11124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926234722693047e-06,
+      "loss": 0.8203,
+      "step": 11125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924521085291841e-06,
+      "loss": 0.8711,
+      "step": 11126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922807591394539e-06,
+      "loss": 0.875,
+      "step": 11127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9210942410614824e-06,
+      "loss": 0.8516,
+      "step": 11128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919381034352991e-06,
+      "loss": 0.8711,
+      "step": 11129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9176679713293926e-06,
+      "loss": 0.9258,
+      "step": 11130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.91595505205101e-06,
+      "loss": 0.8164,
+      "step": 11131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.914242276578158e-06,
+      "loss": 0.7891,
+      "step": 11132
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.912529644971141e-06,
+      "loss": 0.832,
+      "step": 11133
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.910817157290265e-06,
+      "loss": 0.7969,
+      "step": 11134
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909104813595832e-06,
+      "loss": 0.75,
+      "step": 11135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907392613948139e-06,
+      "loss": 0.7266,
+      "step": 11136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905680558407465e-06,
+      "loss": 0.9297,
+      "step": 11137
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9039686470341016e-06,
+      "loss": 0.8438,
+      "step": 11138
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90225687988833e-06,
+      "loss": 0.8516,
+      "step": 11139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900545257030417e-06,
+      "loss": 0.832,
+      "step": 11140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898833778520635e-06,
+      "loss": 0.9102,
+      "step": 11141
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.897122444419248e-06,
+      "loss": 0.9961,
+      "step": 11142
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.895411254786518e-06,
+      "loss": 0.7891,
+      "step": 11143
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893700209682691e-06,
+      "loss": 0.7852,
+      "step": 11144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891989309168021e-06,
+      "loss": 0.8242,
+      "step": 11145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.890278553302749e-06,
+      "loss": 0.8633,
+      "step": 11146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888567942147118e-06,
+      "loss": 0.9453,
+      "step": 11147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886857475761355e-06,
+      "loss": 0.7891,
+      "step": 11148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8851471542056905e-06,
+      "loss": 0.8828,
+      "step": 11149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883436977540349e-06,
+      "loss": 0.9141,
+      "step": 11150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88172694582555e-06,
+      "loss": 0.9336,
+      "step": 11151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8800170591215e-06,
+      "loss": 1.0391,
+      "step": 11152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8783073174884105e-06,
+      "loss": 0.9102,
+      "step": 11153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876597720986486e-06,
+      "loss": 0.8516,
+      "step": 11154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874888269675925e-06,
+      "loss": 0.8945,
+      "step": 11155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873178963616913e-06,
+      "loss": 0.9492,
+      "step": 11156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87146980286964e-06,
+      "loss": 0.7812,
+      "step": 11157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869760787494292e-06,
+      "loss": 0.875,
+      "step": 11158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868051917551047e-06,
+      "loss": 0.8203,
+      "step": 11159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8663431931000705e-06,
+      "loss": 1.0078,
+      "step": 11160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864634614201531e-06,
+      "loss": 0.8594,
+      "step": 11161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8629261809155925e-06,
+      "loss": 0.8438,
+      "step": 11162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.861217893302416e-06,
+      "loss": 0.8086,
+      "step": 11163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.859509751422141e-06,
+      "loss": 0.8594,
+      "step": 11164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857801755334922e-06,
+      "loss": 0.875,
+      "step": 11165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856093905100899e-06,
+      "loss": 0.9141,
+      "step": 11166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854386200780212e-06,
+      "loss": 0.8516,
+      "step": 11167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.852678642432983e-06,
+      "loss": 0.8828,
+      "step": 11168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850971230119342e-06,
+      "loss": 0.9062,
+      "step": 11169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849263963899412e-06,
+      "loss": 0.8945,
+      "step": 11170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8475568438333084e-06,
+      "loss": 0.8477,
+      "step": 11171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.9297,
+      "step": 11172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.844143042403005e-06,
+      "loss": 0.8398,
+      "step": 11173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.842436361159014e-06,
+      "loss": 0.8828,
+      "step": 11174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840729826309261e-06,
+      "loss": 0.957,
+      "step": 11175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839023437913829e-06,
+      "loss": 0.7969,
+      "step": 11176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837317196032808e-06,
+      "loss": 0.918,
+      "step": 11177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.835611100726275e-06,
+      "loss": 0.7227,
+      "step": 11178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833905152054304e-06,
+      "loss": 0.8984,
+      "step": 11179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832199350076968e-06,
+      "loss": 0.832,
+      "step": 11180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830493694854326e-06,
+      "loss": 0.9258,
+      "step": 11181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8287881864464365e-06,
+      "loss": 0.9609,
+      "step": 11182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827082824913362e-06,
+      "loss": 0.8359,
+      "step": 11183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.825377610315137e-06,
+      "loss": 0.8906,
+      "step": 11184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823672542711815e-06,
+      "loss": 0.8359,
+      "step": 11185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.821967622163429e-06,
+      "loss": 0.8594,
+      "step": 11186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820262848730019e-06,
+      "loss": 0.8789,
+      "step": 11187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818558222471603e-06,
+      "loss": 0.9414,
+      "step": 11188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816853743448207e-06,
+      "loss": 0.8867,
+      "step": 11189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815149411719851e-06,
+      "loss": 0.8398,
+      "step": 11190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81344522734655e-06,
+      "loss": 0.7969,
+      "step": 11191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811741190388302e-06,
+      "loss": 0.8789,
+      "step": 11192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.810037300905114e-06,
+      "loss": 1.0078,
+      "step": 11193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8083335589569845e-06,
+      "loss": 0.7578,
+      "step": 11194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806629964603898e-06,
+      "loss": 0.7461,
+      "step": 11195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804926517905846e-06,
+      "loss": 0.8594,
+      "step": 11196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8032232189228084e-06,
+      "loss": 0.8867,
+      "step": 11197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801520067714766e-06,
+      "loss": 0.8906,
+      "step": 11198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7998170643416795e-06,
+      "loss": 0.9141,
+      "step": 11199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798114208863519e-06,
+      "loss": 0.8477,
+      "step": 11200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796411501340244e-06,
+      "loss": 0.9609,
+      "step": 11201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794708941831817e-06,
+      "loss": 0.875,
+      "step": 11202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.793006530398175e-06,
+      "loss": 0.8438,
+      "step": 11203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791304267099269e-06,
+      "loss": 0.8516,
+      "step": 11204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789602151995039e-06,
+      "loss": 0.8242,
+      "step": 11205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7879001851454206e-06,
+      "loss": 0.8203,
+      "step": 11206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786198366610337e-06,
+      "loss": 0.8359,
+      "step": 11207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784496696449715e-06,
+      "loss": 0.8086,
+      "step": 11208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7827951747234724e-06,
+      "loss": 0.6992,
+      "step": 11209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7810938014915284e-06,
+      "loss": 0.9258,
+      "step": 11210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77939257681378e-06,
+      "loss": 0.8008,
+      "step": 11211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7776915007501365e-06,
+      "loss": 0.9766,
+      "step": 11212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7759905733604945e-06,
+      "loss": 0.9219,
+      "step": 11213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774289794704751e-06,
+      "loss": 1.0312,
+      "step": 11214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772589164842782e-06,
+      "loss": 0.8086,
+      "step": 11215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770888683834477e-06,
+      "loss": 0.8984,
+      "step": 11216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.769188351739712e-06,
+      "loss": 0.8633,
+      "step": 11217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76748816861836e-06,
+      "loss": 0.8555,
+      "step": 11218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765788134530281e-06,
+      "loss": 0.9297,
+      "step": 11219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76408824953534e-06,
+      "loss": 0.918,
+      "step": 11220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762388513693391e-06,
+      "loss": 0.8477,
+      "step": 11221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7606889270642906e-06,
+      "loss": 0.8008,
+      "step": 11222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.758989489707874e-06,
+      "loss": 0.9102,
+      "step": 11223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757290201683985e-06,
+      "loss": 0.8359,
+      "step": 11224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755591063052459e-06,
+      "loss": 0.9727,
+      "step": 11225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753892073873129e-06,
+      "loss": 0.8633,
+      "step": 11226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752193234205812e-06,
+      "loss": 0.8711,
+      "step": 11227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750494544110328e-06,
+      "loss": 0.7852,
+      "step": 11228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7487960036464944e-06,
+      "loss": 0.8125,
+      "step": 11229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74709761287412e-06,
+      "loss": 0.8672,
+      "step": 11230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745399371853001e-06,
+      "loss": 0.9609,
+      "step": 11231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74370128064294e-06,
+      "loss": 0.8672,
+      "step": 11232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742003339303728e-06,
+      "loss": 0.75,
+      "step": 11233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740305547895158e-06,
+      "loss": 0.8203,
+      "step": 11234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.738607906477001e-06,
+      "loss": 0.8984,
+      "step": 11235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73691041510904e-06,
+      "loss": 0.8711,
+      "step": 11236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7352130738510445e-06,
+      "loss": 0.8125,
+      "step": 11237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733515882762785e-06,
+      "loss": 0.8672,
+      "step": 11238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731818841904018e-06,
+      "loss": 0.9062,
+      "step": 11239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730121951334498e-06,
+      "loss": 0.8203,
+      "step": 11240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.728425211113975e-06,
+      "loss": 0.7891,
+      "step": 11241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726728621302201e-06,
+      "loss": 0.8789,
+      "step": 11242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7250321819589065e-06,
+      "loss": 0.8047,
+      "step": 11243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.723335893143829e-06,
+      "loss": 0.7812,
+      "step": 11244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721639754916697e-06,
+      "loss": 0.8672,
+      "step": 11245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71994376733724e-06,
+      "loss": 0.875,
+      "step": 11246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718247930465168e-06,
+      "loss": 0.9102,
+      "step": 11247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716552244360196e-06,
+      "loss": 0.8477,
+      "step": 11248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714856709082037e-06,
+      "loss": 0.8828,
+      "step": 11249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7131613246903874e-06,
+      "loss": 0.8477,
+      "step": 11250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711466091244945e-06,
+      "loss": 0.8477,
+      "step": 11251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709771008805403e-06,
+      "loss": 0.7305,
+      "step": 11252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708076077431452e-06,
+      "loss": 0.9141,
+      "step": 11253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7063812971827655e-06,
+      "loss": 0.9219,
+      "step": 11254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704686668119022e-06,
+      "loss": 0.7227,
+      "step": 11255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.702992190299894e-06,
+      "loss": 0.8047,
+      "step": 11256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701297863785049e-06,
+      "loss": 0.8984,
+      "step": 11257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699603688634139e-06,
+      "loss": 0.7812,
+      "step": 11258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697909664906823e-06,
+      "loss": 0.8789,
+      "step": 11259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.69621579266275e-06,
+      "loss": 0.9219,
+      "step": 11260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694522071961568e-06,
+      "loss": 0.7578,
+      "step": 11261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.692828502862908e-06,
+      "loss": 0.8867,
+      "step": 11262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691135085426407e-06,
+      "loss": 0.7695,
+      "step": 11263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689441819711692e-06,
+      "loss": 1.0703,
+      "step": 11264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68774870577839e-06,
+      "loss": 0.9453,
+      "step": 11265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68605574368611e-06,
+      "loss": 0.7695,
+      "step": 11266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684362933494468e-06,
+      "loss": 0.9531,
+      "step": 11267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68267027526307e-06,
+      "loss": 0.8906,
+      "step": 11268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680977769051523e-06,
+      "loss": 0.9648,
+      "step": 11269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679285414919412e-06,
+      "loss": 0.8438,
+      "step": 11270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.677593212926333e-06,
+      "loss": 0.8125,
+      "step": 11271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675901163131871e-06,
+      "loss": 0.793,
+      "step": 11272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67420926559561e-06,
+      "loss": 0.8828,
+      "step": 11273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672517520377115e-06,
+      "loss": 0.9141,
+      "step": 11274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.670825927535959e-06,
+      "loss": 0.9648,
+      "step": 11275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.669134487131707e-06,
+      "loss": 0.9883,
+      "step": 11276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667443199223921e-06,
+      "loss": 0.8047,
+      "step": 11277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665752063872145e-06,
+      "loss": 0.8555,
+      "step": 11278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664061081135931e-06,
+      "loss": 0.8672,
+      "step": 11279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66237025107482e-06,
+      "loss": 0.8867,
+      "step": 11280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6606795737483555e-06,
+      "loss": 0.8516,
+      "step": 11281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6589890492160585e-06,
+      "loss": 0.8477,
+      "step": 11282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6572986775374596e-06,
+      "loss": 0.9219,
+      "step": 11283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6556084587720795e-06,
+      "loss": 0.7891,
+      "step": 11284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653918392979436e-06,
+      "loss": 0.9727,
+      "step": 11285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.652228480219033e-06,
+      "loss": 0.9336,
+      "step": 11286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6505387205503785e-06,
+      "loss": 0.8477,
+      "step": 11287
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64884911403297e-06,
+      "loss": 0.8242,
+      "step": 11288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647159660726309e-06,
+      "loss": 0.9453,
+      "step": 11289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645470360689871e-06,
+      "loss": 0.9023,
+      "step": 11290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.643781213983146e-06,
+      "loss": 0.8828,
+      "step": 11291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642092220665609e-06,
+      "loss": 0.832,
+      "step": 11292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640403380796738e-06,
+      "loss": 0.8125,
+      "step": 11293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638714694435992e-06,
+      "loss": 0.832,
+      "step": 11294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637026161642836e-06,
+      "loss": 0.875,
+      "step": 11295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635337782476724e-06,
+      "loss": 0.75,
+      "step": 11296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6336495569971114e-06,
+      "loss": 0.8594,
+      "step": 11297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631961485263436e-06,
+      "loss": 0.8633,
+      "step": 11298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630273567335141e-06,
+      "loss": 0.8633,
+      "step": 11299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6285858032716605e-06,
+      "loss": 0.875,
+      "step": 11300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626898193132428e-06,
+      "loss": 0.8945,
+      "step": 11301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625210736976857e-06,
+      "loss": 0.8516,
+      "step": 11302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6235234348643706e-06,
+      "loss": 0.8828,
+      "step": 11303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621836286854383e-06,
+      "loss": 0.8008,
+      "step": 11304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620149293006302e-06,
+      "loss": 1.0625,
+      "step": 11305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.618462453379523e-06,
+      "loss": 0.9844,
+      "step": 11306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616775768033446e-06,
+      "loss": 0.7305,
+      "step": 11307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.615089237027466e-06,
+      "loss": 0.7891,
+      "step": 11308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613402860420962e-06,
+      "loss": 0.8984,
+      "step": 11309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611716638273314e-06,
+      "loss": 0.7812,
+      "step": 11310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610030570643899e-06,
+      "loss": 0.875,
+      "step": 11311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608344657592091e-06,
+      "loss": 0.957,
+      "step": 11312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606658899177243e-06,
+      "loss": 1.0,
+      "step": 11313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604973295458719e-06,
+      "loss": 0.8906,
+      "step": 11314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60328784649587e-06,
+      "loss": 0.8672,
+      "step": 11315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601602552348051e-06,
+      "loss": 0.918,
+      "step": 11316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599917413074591e-06,
+      "loss": 0.7773,
+      "step": 11317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598232428734834e-06,
+      "loss": 0.7695,
+      "step": 11318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596547599388108e-06,
+      "loss": 0.9141,
+      "step": 11319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594862925093745e-06,
+      "loss": 0.8594,
+      "step": 11320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593178405911055e-06,
+      "loss": 0.9336,
+      "step": 11321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.591494041899359e-06,
+      "loss": 0.8789,
+      "step": 11322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.589809833117963e-06,
+      "loss": 0.8438,
+      "step": 11323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588125779626177e-06,
+      "loss": 0.8828,
+      "step": 11324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58644188148329e-06,
+      "loss": 0.8359,
+      "step": 11325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584758138748598e-06,
+      "loss": 1.0703,
+      "step": 11326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.583074551481391e-06,
+      "loss": 0.9375,
+      "step": 11327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581391119740951e-06,
+      "loss": 0.8438,
+      "step": 11328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579707843586548e-06,
+      "loss": 0.8867,
+      "step": 11329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5780247230774576e-06,
+      "loss": 0.8672,
+      "step": 11330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576341758272944e-06,
+      "loss": 0.9414,
+      "step": 11331
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574658949232271e-06,
+      "loss": 1.0547,
+      "step": 11332
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572976296014685e-06,
+      "loss": 0.793,
+      "step": 11333
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57129379867944e-06,
+      "loss": 0.7969,
+      "step": 11334
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5696114572857774e-06,
+      "loss": 0.7656,
+      "step": 11335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56792927189294e-06,
+      "loss": 0.8555,
+      "step": 11336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5662472425601524e-06,
+      "loss": 0.8945,
+      "step": 11337
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564565369346644e-06,
+      "loss": 0.8281,
+      "step": 11338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562883652311637e-06,
+      "loss": 0.8203,
+      "step": 11339
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5612020915143526e-06,
+      "loss": 0.8516,
+      "step": 11340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559520687013993e-06,
+      "loss": 0.8008,
+      "step": 11341
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557839438869764e-06,
+      "loss": 0.918,
+      "step": 11342
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5561583471408676e-06,
+      "loss": 0.9648,
+      "step": 11343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554477411886501e-06,
+      "loss": 0.9023,
+      "step": 11344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552796633165845e-06,
+      "loss": 0.8008,
+      "step": 11345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551116011038087e-06,
+      "loss": 0.875,
+      "step": 11346
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549435545562403e-06,
+      "loss": 0.9492,
+      "step": 11347
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547755236797967e-06,
+      "loss": 0.9648,
+      "step": 11348
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546075084803941e-06,
+      "loss": 0.9531,
+      "step": 11349
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544395089639487e-06,
+      "loss": 0.7305,
+      "step": 11350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.542715251363764e-06,
+      "loss": 0.8711,
+      "step": 11351
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541035570035919e-06,
+      "loss": 0.918,
+      "step": 11352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539356045715096e-06,
+      "loss": 0.832,
+      "step": 11353
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537676678460432e-06,
+      "loss": 0.7969,
+      "step": 11354
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5359974683310626e-06,
+      "loss": 0.8242,
+      "step": 11355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5343184153861175e-06,
+      "loss": 0.9609,
+      "step": 11356
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532639519684713e-06,
+      "loss": 0.8594,
+      "step": 11357
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530960781285969e-06,
+      "loss": 0.8477,
+      "step": 11358
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529282200248995e-06,
+      "loss": 0.7695,
+      "step": 11359
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5276037766329026e-06,
+      "loss": 0.8242,
+      "step": 11360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525925510496781e-06,
+      "loss": 0.9492,
+      "step": 11361
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524247401899732e-06,
+      "loss": 0.7969,
+      "step": 11362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522569450900845e-06,
+      "loss": 0.9023,
+      "step": 11363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.520891657559197e-06,
+      "loss": 0.7969,
+      "step": 11364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519214021933872e-06,
+      "loss": 0.8945,
+      "step": 11365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.517536544083936e-06,
+      "loss": 0.8164,
+      "step": 11366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515859224068465e-06,
+      "loss": 0.8008,
+      "step": 11367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51418206194651e-06,
+      "loss": 0.8672,
+      "step": 11368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512505057777131e-06,
+      "loss": 1.0,
+      "step": 11369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5108282116193765e-06,
+      "loss": 0.793,
+      "step": 11370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509151523532297e-06,
+      "loss": 0.7539,
+      "step": 11371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5074749935749215e-06,
+      "loss": 0.9297,
+      "step": 11372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.505798621806289e-06,
+      "loss": 0.918,
+      "step": 11373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504122408285425e-06,
+      "loss": 0.8789,
+      "step": 11374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5024463530713565e-06,
+      "loss": 0.9531,
+      "step": 11375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500770456223093e-06,
+      "loss": 0.6367,
+      "step": 11376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.499094717799648e-06,
+      "loss": 0.793,
+      "step": 11377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497419137860028e-06,
+      "loss": 0.8125,
+      "step": 11378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.495743716463235e-06,
+      "loss": 0.8047,
+      "step": 11379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494068453668258e-06,
+      "loss": 0.7734,
+      "step": 11380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492393349534086e-06,
+      "loss": 0.9414,
+      "step": 11381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.490718404119704e-06,
+      "loss": 0.9062,
+      "step": 11382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489043617484094e-06,
+      "loss": 0.793,
+      "step": 11383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.487368989686219e-06,
+      "loss": 0.918,
+      "step": 11384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48569452078505e-06,
+      "loss": 0.7656,
+      "step": 11385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484020210839547e-06,
+      "loss": 0.7969,
+      "step": 11386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482346059908668e-06,
+      "loss": 0.8047,
+      "step": 11387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480672068051356e-06,
+      "loss": 0.7266,
+      "step": 11388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.478998235326558e-06,
+      "loss": 0.8711,
+      "step": 11389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477324561793212e-06,
+      "loss": 0.9062,
+      "step": 11390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475651047510257e-06,
+      "loss": 0.8906,
+      "step": 11391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4739776925366085e-06,
+      "loss": 0.8086,
+      "step": 11392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4723044969311945e-06,
+      "loss": 0.9023,
+      "step": 11393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470631460752928e-06,
+      "loss": 0.8242,
+      "step": 11394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468958584060728e-06,
+      "loss": 0.8164,
+      "step": 11395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.467285866913487e-06,
+      "loss": 0.8594,
+      "step": 11396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465613309370109e-06,
+      "loss": 0.9141,
+      "step": 11397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463940911489487e-06,
+      "loss": 0.8516,
+      "step": 11398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.462268673330514e-06,
+      "loss": 0.875,
+      "step": 11399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460596594952063e-06,
+      "loss": 0.9062,
+      "step": 11400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.458924676413015e-06,
+      "loss": 0.8359,
+      "step": 11401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45725291777224e-06,
+      "loss": 0.8945,
+      "step": 11402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.455581319088608e-06,
+      "loss": 0.8281,
+      "step": 11403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45390988042097e-06,
+      "loss": 0.875,
+      "step": 11404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452238601828184e-06,
+      "loss": 0.8086,
+      "step": 11405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4505674833690995e-06,
+      "loss": 0.8125,
+      "step": 11406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448896525102562e-06,
+      "loss": 0.8398,
+      "step": 11407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447225727087399e-06,
+      "loss": 0.8164,
+      "step": 11408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445555089382451e-06,
+      "loss": 0.7656,
+      "step": 11409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443884612046537e-06,
+      "loss": 0.8242,
+      "step": 11410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442214295138486e-06,
+      "loss": 0.8633,
+      "step": 11411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440544138717103e-06,
+      "loss": 0.9453,
+      "step": 11412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4388741428412004e-06,
+      "loss": 0.8945,
+      "step": 11413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437204307569581e-06,
+      "loss": 0.8438,
+      "step": 11414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.435534632961048e-06,
+      "loss": 0.8633,
+      "step": 11415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433865119074384e-06,
+      "loss": 0.8516,
+      "step": 11416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432195765968379e-06,
+      "loss": 0.9219,
+      "step": 11417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430526573701817e-06,
+      "loss": 0.875,
+      "step": 11418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428857542333465e-06,
+      "loss": 0.918,
+      "step": 11419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427188671922099e-06,
+      "loss": 0.7969,
+      "step": 11420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4255199625264794e-06,
+      "loss": 0.832,
+      "step": 11421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423851414205367e-06,
+      "loss": 0.8672,
+      "step": 11422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4221830270175114e-06,
+      "loss": 0.9531,
+      "step": 11423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420514801021657e-06,
+      "loss": 0.8359,
+      "step": 11424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4188467362765484e-06,
+      "loss": 0.7383,
+      "step": 11425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417178832840922e-06,
+      "loss": 0.8906,
+      "step": 11426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415511090773502e-06,
+      "loss": 0.9141,
+      "step": 11427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413843510133014e-06,
+      "loss": 0.8047,
+      "step": 11428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412176090978178e-06,
+      "loss": 1.0234,
+      "step": 11429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410508833367709e-06,
+      "loss": 0.8203,
+      "step": 11430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4088417373603065e-06,
+      "loss": 0.7539,
+      "step": 11431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4071748030146755e-06,
+      "loss": 0.8828,
+      "step": 11432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4055080303895105e-06,
+      "loss": 0.8203,
+      "step": 11433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403841419543505e-06,
+      "loss": 0.7891,
+      "step": 11434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402174970535337e-06,
+      "loss": 0.8242,
+      "step": 11435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4005086834236855e-06,
+      "loss": 0.8359,
+      "step": 11436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398842558267226e-06,
+      "loss": 0.7734,
+      "step": 11437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397176595124627e-06,
+      "loss": 0.8594,
+      "step": 11438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.395510794054544e-06,
+      "loss": 0.8242,
+      "step": 11439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393845155115636e-06,
+      "loss": 0.9023,
+      "step": 11440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39217967836655e-06,
+      "loss": 0.8555,
+      "step": 11441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390514363865937e-06,
+      "loss": 0.8555,
+      "step": 11442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388849211672426e-06,
+      "loss": 0.7969,
+      "step": 11443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387184221844655e-06,
+      "loss": 0.8047,
+      "step": 11444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.385519394441248e-06,
+      "loss": 0.9922,
+      "step": 11445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383854729520832e-06,
+      "loss": 0.9648,
+      "step": 11446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382190227142015e-06,
+      "loss": 0.7148,
+      "step": 11447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38052588736341e-06,
+      "loss": 0.8477,
+      "step": 11448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378861710243622e-06,
+      "loss": 0.9492,
+      "step": 11449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37719769584125e-06,
+      "loss": 0.957,
+      "step": 11450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375533844214882e-06,
+      "loss": 0.8906,
+      "step": 11451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373870155423106e-06,
+      "loss": 0.8203,
+      "step": 11452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372206629524508e-06,
+      "loss": 0.8438,
+      "step": 11453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370543266577662e-06,
+      "loss": 0.8516,
+      "step": 11454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36888006664113e-06,
+      "loss": 0.8789,
+      "step": 11455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3672170297734816e-06,
+      "loss": 0.8555,
+      "step": 11456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365554156033277e-06,
+      "loss": 1.0391,
+      "step": 11457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363891445479068e-06,
+      "loss": 0.7734,
+      "step": 11458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362228898169395e-06,
+      "loss": 0.8633,
+      "step": 11459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360566514162805e-06,
+      "loss": 0.7969,
+      "step": 11460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35890429351783e-06,
+      "loss": 0.8672,
+      "step": 11461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357242236293005e-06,
+      "loss": 0.8633,
+      "step": 11462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.355580342546845e-06,
+      "loss": 0.8203,
+      "step": 11463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353918612337873e-06,
+      "loss": 0.8281,
+      "step": 11464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352257045724599e-06,
+      "loss": 0.8086,
+      "step": 11465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.350595642765535e-06,
+      "loss": 0.8281,
+      "step": 11466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348934403519174e-06,
+      "loss": 0.9375,
+      "step": 11467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347273328044013e-06,
+      "loss": 0.8828,
+      "step": 11468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345612416398545e-06,
+      "loss": 0.9141,
+      "step": 11469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3439516686412476e-06,
+      "loss": 0.7656,
+      "step": 11470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3422910848306e-06,
+      "loss": 0.7578,
+      "step": 11471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340630665025075e-06,
+      "loss": 0.8906,
+      "step": 11472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338970409283144e-06,
+      "loss": 1.0156,
+      "step": 11473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337310317663255e-06,
+      "loss": 0.9297,
+      "step": 11474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.33565039022387e-06,
+      "loss": 0.8438,
+      "step": 11475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333990627023437e-06,
+      "loss": 0.7422,
+      "step": 11476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.332331028120401e-06,
+      "loss": 0.8789,
+      "step": 11477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330671593573193e-06,
+      "loss": 0.7891,
+      "step": 11478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329012323440248e-06,
+      "loss": 0.7617,
+      "step": 11479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327353217779991e-06,
+      "loss": 0.957,
+      "step": 11480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325694276650845e-06,
+      "loss": 0.8633,
+      "step": 11481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324035500111218e-06,
+      "loss": 0.75,
+      "step": 11482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3223768882195205e-06,
+      "loss": 0.8359,
+      "step": 11483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3207184410341575e-06,
+      "loss": 0.8477,
+      "step": 11484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3190601586135205e-06,
+      "loss": 0.8945,
+      "step": 11485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317402041016002e-06,
+      "loss": 0.9531,
+      "step": 11486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315744088299988e-06,
+      "loss": 0.9688,
+      "step": 11487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31408630052386e-06,
+      "loss": 0.8672,
+      "step": 11488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.312428677745984e-06,
+      "loss": 0.8359,
+      "step": 11489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310771220024733e-06,
+      "loss": 0.8555,
+      "step": 11490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309113927418468e-06,
+      "loss": 0.7461,
+      "step": 11491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307456799985548e-06,
+      "loss": 0.8945,
+      "step": 11492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3057998377843155e-06,
+      "loss": 1.0,
+      "step": 11493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304143040873117e-06,
+      "loss": 0.875,
+      "step": 11494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302486409310294e-06,
+      "loss": 0.8008,
+      "step": 11495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300829943154181e-06,
+      "loss": 0.9219,
+      "step": 11496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299173642463098e-06,
+      "loss": 0.8984,
+      "step": 11497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297517507295368e-06,
+      "loss": 0.7852,
+      "step": 11498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2958615377093114e-06,
+      "loss": 0.7734,
+      "step": 11499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294205733763229e-06,
+      "loss": 0.7578,
+      "step": 11500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2925500955154296e-06,
+      "loss": 0.8906,
+      "step": 11501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2908946230242075e-06,
+      "loss": 0.9062,
+      "step": 11502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289239316347863e-06,
+      "loss": 0.8281,
+      "step": 11503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28758417554467e-06,
+      "loss": 0.957,
+      "step": 11504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285929200672915e-06,
+      "loss": 0.8281,
+      "step": 11505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284274391790872e-06,
+      "loss": 0.9141,
+      "step": 11506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28261974895681e-06,
+      "loss": 0.9609,
+      "step": 11507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280965272228989e-06,
+      "loss": 0.8125,
+      "step": 11508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279310961665664e-06,
+      "loss": 0.9453,
+      "step": 11509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277656817325092e-06,
+      "loss": 0.8281,
+      "step": 11510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2760028392655175e-06,
+      "loss": 0.8438,
+      "step": 11511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274349027545171e-06,
+      "loss": 0.9102,
+      "step": 11512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272695382222295e-06,
+      "loss": 0.7695,
+      "step": 11513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.271041903355115e-06,
+      "loss": 0.918,
+      "step": 11514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269388591001848e-06,
+      "loss": 0.9219,
+      "step": 11515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.267735445220713e-06,
+      "loss": 0.7891,
+      "step": 11516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.266082466069919e-06,
+      "loss": 0.8438,
+      "step": 11517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264429653607675e-06,
+      "loss": 0.8516,
+      "step": 11518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2627770078921704e-06,
+      "loss": 0.9922,
+      "step": 11519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2611245289816014e-06,
+      "loss": 0.9102,
+      "step": 11520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259472216934155e-06,
+      "loss": 0.832,
+      "step": 11521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.257820071808016e-06,
+      "loss": 0.8672,
+      "step": 11522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256168093661348e-06,
+      "loss": 0.9219,
+      "step": 11523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254516282552326e-06,
+      "loss": 1.0156,
+      "step": 11524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2528646385391144e-06,
+      "loss": 0.7578,
+      "step": 11525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251213161679871e-06,
+      "loss": 0.918,
+      "step": 11526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249561852032742e-06,
+      "loss": 0.8633,
+      "step": 11527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247910709655874e-06,
+      "loss": 0.9922,
+      "step": 11528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246259734607411e-06,
+      "loss": 0.7852,
+      "step": 11529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244608926945478e-06,
+      "loss": 0.9922,
+      "step": 11530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242958286728208e-06,
+      "loss": 0.8008,
+      "step": 11531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241307814013721e-06,
+      "loss": 0.8789,
+      "step": 11532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239657508860136e-06,
+      "loss": 0.9062,
+      "step": 11533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238007371325557e-06,
+      "loss": 0.8516,
+      "step": 11534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.236357401468093e-06,
+      "loss": 0.8945,
+      "step": 11535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234707599345839e-06,
+      "loss": 0.9141,
+      "step": 11536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23305796501689e-06,
+      "loss": 0.8789,
+      "step": 11537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231408498539328e-06,
+      "loss": 0.7305,
+      "step": 11538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.229759199971234e-06,
+      "loss": 0.9062,
+      "step": 11539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228110069370689e-06,
+      "loss": 0.8906,
+      "step": 11540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226461106795753e-06,
+      "loss": 0.9023,
+      "step": 11541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224812312304491e-06,
+      "loss": 0.8359,
+      "step": 11542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22316368595496e-06,
+      "loss": 0.7969,
+      "step": 11543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221515227805214e-06,
+      "loss": 0.8086,
+      "step": 11544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219866937913292e-06,
+      "loss": 0.8086,
+      "step": 11545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218218816337235e-06,
+      "loss": 0.9727,
+      "step": 11546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216570863135077e-06,
+      "loss": 0.9219,
+      "step": 11547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214923078364848e-06,
+      "loss": 0.7656,
+      "step": 11548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213275462084564e-06,
+      "loss": 0.75,
+      "step": 11549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211628014352239e-06,
+      "loss": 0.8906,
+      "step": 11550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209980735225886e-06,
+      "loss": 0.9336,
+      "step": 11551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208333624763511e-06,
+      "loss": 0.8398,
+      "step": 11552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2066866830231025e-06,
+      "loss": 0.9102,
+      "step": 11553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.205039910062657e-06,
+      "loss": 0.8945,
+      "step": 11554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203393305940161e-06,
+      "loss": 0.9531,
+      "step": 11555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201746870713595e-06,
+      "loss": 0.7891,
+      "step": 11556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200100604440926e-06,
+      "loss": 0.7773,
+      "step": 11557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198454507180127e-06,
+      "loss": 0.9609,
+      "step": 11558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196808578989159e-06,
+      "loss": 0.9648,
+      "step": 11559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195162819925979e-06,
+      "loss": 0.8438,
+      "step": 11560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.193517230048531e-06,
+      "loss": 0.8633,
+      "step": 11561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191871809414763e-06,
+      "loss": 0.8672,
+      "step": 11562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1902265580826135e-06,
+      "loss": 0.8711,
+      "step": 11563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188581476110016e-06,
+      "loss": 0.9141,
+      "step": 11564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18693656355489e-06,
+      "loss": 0.957,
+      "step": 11565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.185291820475158e-06,
+      "loss": 0.9922,
+      "step": 11566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1836472469287355e-06,
+      "loss": 0.9219,
+      "step": 11567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182002842973534e-06,
+      "loss": 0.875,
+      "step": 11568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180358608667449e-06,
+      "loss": 0.8125,
+      "step": 11569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178714544068376e-06,
+      "loss": 0.9375,
+      "step": 11570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.177070649234209e-06,
+      "loss": 0.875,
+      "step": 11571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175426924222835e-06,
+      "loss": 0.8711,
+      "step": 11572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173783369092125e-06,
+      "loss": 0.8438,
+      "step": 11573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.172139983899953e-06,
+      "loss": 0.7422,
+      "step": 11574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.170496768704188e-06,
+      "loss": 0.8945,
+      "step": 11575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1688537235626905e-06,
+      "loss": 1.1016,
+      "step": 11576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16721084853331e-06,
+      "loss": 0.8047,
+      "step": 11577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165568143673897e-06,
+      "loss": 0.8867,
+      "step": 11578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163925609042294e-06,
+      "loss": 0.8242,
+      "step": 11579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16228324469634e-06,
+      "loss": 0.8125,
+      "step": 11580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1606410506938596e-06,
+      "loss": 0.8359,
+      "step": 11581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.158999027092681e-06,
+      "loss": 0.8203,
+      "step": 11582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157357173950619e-06,
+      "loss": 0.8203,
+      "step": 11583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.155715491325493e-06,
+      "loss": 0.957,
+      "step": 11584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1540739792750994e-06,
+      "loss": 1.0625,
+      "step": 11585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152432637857243e-06,
+      "loss": 0.7461,
+      "step": 11586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1507914671297235e-06,
+      "loss": 0.793,
+      "step": 11587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149150467150319e-06,
+      "loss": 0.8164,
+      "step": 11588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.147509637976815e-06,
+      "loss": 0.9062,
+      "step": 11589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1458689796669905e-06,
+      "loss": 0.875,
+      "step": 11590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1442284922786165e-06,
+      "loss": 0.8906,
+      "step": 11591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142588175869452e-06,
+      "loss": 0.9531,
+      "step": 11592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140948030497257e-06,
+      "loss": 0.9531,
+      "step": 11593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.139308056219783e-06,
+      "loss": 0.8945,
+      "step": 11594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137668253094783e-06,
+      "loss": 0.8789,
+      "step": 11595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136028621179987e-06,
+      "loss": 0.8516,
+      "step": 11596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134389160533133e-06,
+      "loss": 0.8945,
+      "step": 11597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132749871211948e-06,
+      "loss": 0.8906,
+      "step": 11598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13111075327416e-06,
+      "loss": 0.7734,
+      "step": 11599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1294718067774765e-06,
+      "loss": 0.8594,
+      "step": 11600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12783303177961e-06,
+      "loss": 0.7539,
+      "step": 11601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126194428338265e-06,
+      "loss": 0.875,
+      "step": 11602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124555996511144e-06,
+      "loss": 0.8828,
+      "step": 11603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1229177363559305e-06,
+      "loss": 0.7695,
+      "step": 11604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1212796479303115e-06,
+      "loss": 0.9609,
+      "step": 11605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119641731291971e-06,
+      "loss": 0.8672,
+      "step": 11606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.118003986498583e-06,
+      "loss": 0.9727,
+      "step": 11607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1163664136078095e-06,
+      "loss": 1.0,
+      "step": 11608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114729012677313e-06,
+      "loss": 0.7812,
+      "step": 11609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113091783764752e-06,
+      "loss": 0.8359,
+      "step": 11610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111454726927779e-06,
+      "loss": 0.9219,
+      "step": 11611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10981784222403e-06,
+      "loss": 0.9414,
+      "step": 11612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108181129711143e-06,
+      "loss": 0.9258,
+      "step": 11613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106544589446752e-06,
+      "loss": 0.957,
+      "step": 11614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104908221488486e-06,
+      "loss": 0.8867,
+      "step": 11615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103272025893955e-06,
+      "loss": 0.8281,
+      "step": 11616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101636002720778e-06,
+      "loss": 0.75,
+      "step": 11617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100000152026559e-06,
+      "loss": 0.8125,
+      "step": 11618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0983644738689046e-06,
+      "loss": 0.9375,
+      "step": 11619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0967289683054e-06,
+      "loss": 0.6914,
+      "step": 11620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0950936353936395e-06,
+      "loss": 0.9375,
+      "step": 11621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0934584751912065e-06,
+      "loss": 0.8438,
+      "step": 11622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091823487755681e-06,
+      "loss": 0.8477,
+      "step": 11623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090188673144622e-06,
+      "loss": 0.875,
+      "step": 11624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088554031415602e-06,
+      "loss": 0.7656,
+      "step": 11625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086919562626178e-06,
+      "loss": 1.0,
+      "step": 11626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.085285266833906e-06,
+      "loss": 0.8945,
+      "step": 11627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083651144096323e-06,
+      "loss": 0.9141,
+      "step": 11628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082017194470974e-06,
+      "loss": 0.7656,
+      "step": 11629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.080383418015394e-06,
+      "loss": 0.8359,
+      "step": 11630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078749814787114e-06,
+      "loss": 0.918,
+      "step": 11631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077116384843647e-06,
+      "loss": 0.7734,
+      "step": 11632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.075483128242512e-06,
+      "loss": 0.9609,
+      "step": 11633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07385004504122e-06,
+      "loss": 0.8828,
+      "step": 11634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072217135297279e-06,
+      "loss": 0.8164,
+      "step": 11635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.070584399068176e-06,
+      "loss": 0.9258,
+      "step": 11636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0689518364114076e-06,
+      "loss": 0.8008,
+      "step": 11637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.067319447384458e-06,
+      "loss": 0.9453,
+      "step": 11638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065687232044811e-06,
+      "loss": 0.8594,
+      "step": 11639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064055190449933e-06,
+      "loss": 0.8672,
+      "step": 11640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0624233226572905e-06,
+      "loss": 0.8516,
+      "step": 11641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060791628724351e-06,
+      "loss": 0.8047,
+      "step": 11642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05916010870856e-06,
+      "loss": 0.8125,
+      "step": 11643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057528762667371e-06,
+      "loss": 0.8359,
+      "step": 11644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055897590658225e-06,
+      "loss": 0.7383,
+      "step": 11645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054266592738563e-06,
+      "loss": 0.8242,
+      "step": 11646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0526357689658055e-06,
+      "loss": 0.7891,
+      "step": 11647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051005119397382e-06,
+      "loss": 0.8281,
+      "step": 11648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.04937464409071e-06,
+      "loss": 0.8281,
+      "step": 11649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.047744343103203e-06,
+      "loss": 0.8398,
+      "step": 11650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.046114216492261e-06,
+      "loss": 0.8438,
+      "step": 11651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044484264315286e-06,
+      "loss": 0.8477,
+      "step": 11652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0428544866296715e-06,
+      "loss": 0.8125,
+      "step": 11653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041224883492808e-06,
+      "loss": 0.8438,
+      "step": 11654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03959545496207e-06,
+      "loss": 0.8242,
+      "step": 11655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0379662010948325e-06,
+      "loss": 0.7109,
+      "step": 11656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.036337121948468e-06,
+      "loss": 0.8711,
+      "step": 11657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.034708217580341e-06,
+      "loss": 0.7891,
+      "step": 11658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033079488047801e-06,
+      "loss": 0.7148,
+      "step": 11659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.031450933408199e-06,
+      "loss": 0.7461,
+      "step": 11660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0298225537188824e-06,
+      "loss": 0.8867,
+      "step": 11661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028194349037192e-06,
+      "loss": 0.9023,
+      "step": 11662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026566319420449e-06,
+      "loss": 0.7812,
+      "step": 11663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.024938464925988e-06,
+      "loss": 0.8438,
+      "step": 11664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023310785611123e-06,
+      "loss": 0.9141,
+      "step": 11665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021683281533174e-06,
+      "loss": 0.7891,
+      "step": 11666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0200559527494386e-06,
+      "loss": 0.8906,
+      "step": 11667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018428799317222e-06,
+      "loss": 0.8906,
+      "step": 11668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016801821293819e-06,
+      "loss": 0.7383,
+      "step": 11669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015175018736522e-06,
+      "loss": 1.0312,
+      "step": 11670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.013548391702605e-06,
+      "loss": 0.7812,
+      "step": 11671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011921940249349e-06,
+      "loss": 0.8594,
+      "step": 11672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010295664434023e-06,
+      "loss": 0.8711,
+      "step": 11673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008669564313894e-06,
+      "loss": 0.8359,
+      "step": 11674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007043639946212e-06,
+      "loss": 0.7539,
+      "step": 11675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0054178913882335e-06,
+      "loss": 0.7773,
+      "step": 11676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003792318697203e-06,
+      "loss": 0.8281,
+      "step": 11677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002166921930361e-06,
+      "loss": 0.8203,
+      "step": 11678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000541701144936e-06,
+      "loss": 0.8086,
+      "step": 11679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998916656398157e-06,
+      "loss": 0.7969,
+      "step": 11680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9972917877472435e-06,
+      "loss": 0.8672,
+      "step": 11681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.995667095249415e-06,
+      "loss": 0.8789,
+      "step": 11682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9940425789618695e-06,
+      "loss": 0.7969,
+      "step": 11683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992418238941815e-06,
+      "loss": 0.9141,
+      "step": 11684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9907940752464454e-06,
+      "loss": 0.7969,
+      "step": 11685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989170087932955e-06,
+      "loss": 0.8398,
+      "step": 11686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987546277058518e-06,
+      "loss": 0.7422,
+      "step": 11687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985922642680316e-06,
+      "loss": 0.9766,
+      "step": 11688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98429918485552e-06,
+      "loss": 0.8633,
+      "step": 11689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982675903641297e-06,
+      "loss": 1.0234,
+      "step": 11690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9810527990948e-06,
+      "loss": 1.0078,
+      "step": 11691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9794298712731814e-06,
+      "loss": 0.8867,
+      "step": 11692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97780712023359e-06,
+      "loss": 0.8125,
+      "step": 11693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976184546033167e-06,
+      "loss": 0.957,
+      "step": 11694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974562148729039e-06,
+      "loss": 0.8281,
+      "step": 11695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972939928378338e-06,
+      "loss": 0.8555,
+      "step": 11696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971317885038183e-06,
+      "loss": 0.9102,
+      "step": 11697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969696018765693e-06,
+      "loss": 0.793,
+      "step": 11698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96807432961797e-06,
+      "loss": 0.8008,
+      "step": 11699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966452817652119e-06,
+      "loss": 0.8516,
+      "step": 11700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9648314829252394e-06,
+      "loss": 0.832,
+      "step": 11701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963210325494414e-06,
+      "loss": 0.9414,
+      "step": 11702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96158934541673e-06,
+      "loss": 0.8438,
+      "step": 11703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959968542749264e-06,
+      "loss": 0.8789,
+      "step": 11704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958347917549092e-06,
+      "loss": 0.7578,
+      "step": 11705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956727469873269e-06,
+      "loss": 0.9492,
+      "step": 11706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.95510719977886e-06,
+      "loss": 0.9297,
+      "step": 11707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.953487107322915e-06,
+      "loss": 0.9141,
+      "step": 11708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951867192562485e-06,
+      "loss": 0.7656,
+      "step": 11709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.950247455554602e-06,
+      "loss": 0.8789,
+      "step": 11710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948627896356302e-06,
+      "loss": 0.793,
+      "step": 11711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947008515024613e-06,
+      "loss": 0.8945,
+      "step": 11712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94538931161656e-06,
+      "loss": 0.8984,
+      "step": 11713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94377028618915e-06,
+      "loss": 0.9492,
+      "step": 11714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942151438799395e-06,
+      "loss": 0.8984,
+      "step": 11715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940532769504297e-06,
+      "loss": 1.0781,
+      "step": 11716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938914278360857e-06,
+      "loss": 0.7852,
+      "step": 11717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937295965426054e-06,
+      "loss": 0.9492,
+      "step": 11718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935677830756878e-06,
+      "loss": 0.8438,
+      "step": 11719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.934059874410304e-06,
+      "loss": 0.8438,
+      "step": 11720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932442096443308e-06,
+      "loss": 1.0312,
+      "step": 11721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.930824496912846e-06,
+      "loss": 0.8672,
+      "step": 11722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92920707587588e-06,
+      "loss": 0.9062,
+      "step": 11723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927589833389362e-06,
+      "loss": 0.7539,
+      "step": 11724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925972769510243e-06,
+      "loss": 0.8438,
+      "step": 11725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924355884295452e-06,
+      "loss": 0.7383,
+      "step": 11726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922739177801927e-06,
+      "loss": 0.9141,
+      "step": 11727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921122650086596e-06,
+      "loss": 0.8242,
+      "step": 11728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919506301206381e-06,
+      "loss": 0.8672,
+      "step": 11729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9178901312181904e-06,
+      "loss": 0.832,
+      "step": 11730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916274140178934e-06,
+      "loss": 0.9688,
+      "step": 11731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914658328145515e-06,
+      "loss": 0.9141,
+      "step": 11732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913042695174831e-06,
+      "loss": 0.8555,
+      "step": 11733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911427241323765e-06,
+      "loss": 0.7773,
+      "step": 11734
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909811966649203e-06,
+      "loss": 0.8516,
+      "step": 11735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9081968712080195e-06,
+      "loss": 0.8438,
+      "step": 11736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906581955057091e-06,
+      "loss": 0.8438,
+      "step": 11737
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904967218253271e-06,
+      "loss": 0.9062,
+      "step": 11738
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903352660853423e-06,
+      "loss": 0.7578,
+      "step": 11739
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901738282914398e-06,
+      "loss": 0.8125,
+      "step": 11740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900124084493042e-06,
+      "loss": 0.832,
+      "step": 11741
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898510065646187e-06,
+      "loss": 0.7852,
+      "step": 11742
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89689622643067e-06,
+      "loss": 0.8945,
+      "step": 11743
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895282566903317e-06,
+      "loss": 0.9297,
+      "step": 11744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893669087120948e-06,
+      "loss": 0.8398,
+      "step": 11745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892055787140373e-06,
+      "loss": 1.0,
+      "step": 11746
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890442667018399e-06,
+      "loss": 0.8281,
+      "step": 11747
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888829726811828e-06,
+      "loss": 0.8125,
+      "step": 11748
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887216966577458e-06,
+      "loss": 0.8906,
+      "step": 11749
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885604386372069e-06,
+      "loss": 0.875,
+      "step": 11750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.883991986252447e-06,
+      "loss": 0.8477,
+      "step": 11751
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882379766275365e-06,
+      "loss": 0.8164,
+      "step": 11752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880767726497596e-06,
+      "loss": 0.8086,
+      "step": 11753
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879155866975898e-06,
+      "loss": 0.7852,
+      "step": 11754
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877544187767027e-06,
+      "loss": 0.8125,
+      "step": 11755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8759326889277384e-06,
+      "loss": 0.8242,
+      "step": 11756
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874321370514768e-06,
+      "loss": 0.8164,
+      "step": 11757
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872710232584854e-06,
+      "loss": 0.8594,
+      "step": 11758
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871099275194732e-06,
+      "loss": 0.793,
+      "step": 11759
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869488498401126e-06,
+      "loss": 0.8164,
+      "step": 11760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867877902260747e-06,
+      "loss": 0.8008,
+      "step": 11761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866267486830312e-06,
+      "loss": 0.793,
+      "step": 11762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864657252166525e-06,
+      "loss": 1.0156,
+      "step": 11763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863047198326089e-06,
+      "loss": 0.7852,
+      "step": 11764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861437325365689e-06,
+      "loss": 0.8984,
+      "step": 11765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859827633342014e-06,
+      "loss": 1.0078,
+      "step": 11766
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.858218122311745e-06,
+      "loss": 0.875,
+      "step": 11767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856608792331558e-06,
+      "loss": 0.9258,
+      "step": 11768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854999643458115e-06,
+      "loss": 0.8281,
+      "step": 11769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853390675748077e-06,
+      "loss": 0.9062,
+      "step": 11770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851781889258099e-06,
+      "loss": 0.8203,
+      "step": 11771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850173284044833e-06,
+      "loss": 1.0391,
+      "step": 11772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.848564860164915e-06,
+      "loss": 0.8281,
+      "step": 11773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846956617674982e-06,
+      "loss": 0.957,
+      "step": 11774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845348556631661e-06,
+      "loss": 0.8984,
+      "step": 11775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84374067709158e-06,
+      "loss": 0.9883,
+      "step": 11776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.842132979111349e-06,
+      "loss": 0.957,
+      "step": 11777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8405254627475785e-06,
+      "loss": 0.9141,
+      "step": 11778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838918128056872e-06,
+      "loss": 0.9023,
+      "step": 11779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8373109750958316e-06,
+      "loss": 0.8477,
+      "step": 11780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8357040039210394e-06,
+      "loss": 0.7539,
+      "step": 11781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834097214589082e-06,
+      "loss": 0.7773,
+      "step": 11782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832490607156538e-06,
+      "loss": 0.8281,
+      "step": 11783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830884181679983e-06,
+      "loss": 0.7734,
+      "step": 11784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829277938215973e-06,
+      "loss": 0.8438,
+      "step": 11785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.827671876821069e-06,
+      "loss": 0.8398,
+      "step": 11786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8260659975518255e-06,
+      "loss": 0.8516,
+      "step": 11787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.82446030046479e-06,
+      "loss": 0.8047,
+      "step": 11788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822854785616495e-06,
+      "loss": 0.8945,
+      "step": 11789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821249453063477e-06,
+      "loss": 0.8398,
+      "step": 11790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8196443028622604e-06,
+      "loss": 0.7734,
+      "step": 11791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818039335069371e-06,
+      "loss": 0.957,
+      "step": 11792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.816434549741313e-06,
+      "loss": 0.9297,
+      "step": 11793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814829946934599e-06,
+      "loss": 0.7148,
+      "step": 11794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813225526705728e-06,
+      "loss": 0.8242,
+      "step": 11795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811621289111198e-06,
+      "loss": 0.8828,
+      "step": 11796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81001723420749e-06,
+      "loss": 0.9922,
+      "step": 11797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8084133620510875e-06,
+      "loss": 0.7617,
+      "step": 11798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806809672698467e-06,
+      "loss": 0.9414,
+      "step": 11799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8052061662061e-06,
+      "loss": 0.875,
+      "step": 11800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8036028426304415e-06,
+      "loss": 0.8203,
+      "step": 11801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8019997020279495e-06,
+      "loss": 0.8594,
+      "step": 11802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800396744455073e-06,
+      "loss": 0.8242,
+      "step": 11803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7987939699682605e-06,
+      "loss": 0.8398,
+      "step": 11804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7971913786239374e-06,
+      "loss": 0.875,
+      "step": 11805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7955889704785405e-06,
+      "loss": 0.8945,
+      "step": 11806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793986745588491e-06,
+      "loss": 0.7578,
+      "step": 11807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79238470401021e-06,
+      "loss": 0.7969,
+      "step": 11808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790782845800101e-06,
+      "loss": 0.8594,
+      "step": 11809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78918117101457e-06,
+      "loss": 0.8359,
+      "step": 11810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78757967971002e-06,
+      "loss": 0.8438,
+      "step": 11811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785978371942833e-06,
+      "loss": 0.8555,
+      "step": 11812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784377247769399e-06,
+      "loss": 0.7852,
+      "step": 11813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7827763072460935e-06,
+      "loss": 0.8359,
+      "step": 11814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781175550429295e-06,
+      "loss": 0.9414,
+      "step": 11815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779574977375358e-06,
+      "loss": 0.8281,
+      "step": 11816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777974588140646e-06,
+      "loss": 0.8047,
+      "step": 11817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7763743827815126e-06,
+      "loss": 0.793,
+      "step": 11818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774774361354306e-06,
+      "loss": 0.8086,
+      "step": 11819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773174523915358e-06,
+      "loss": 0.8672,
+      "step": 11820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7715748705210065e-06,
+      "loss": 0.8203,
+      "step": 11821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.769975401227574e-06,
+      "loss": 0.9492,
+      "step": 11822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768376116091389e-06,
+      "loss": 0.9297,
+      "step": 11823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7667770151687535e-06,
+      "loss": 0.8672,
+      "step": 11824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.765178098515979e-06,
+      "loss": 0.9062,
+      "step": 11825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763579366189367e-06,
+      "loss": 0.7773,
+      "step": 11826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.761980818245214e-06,
+      "loss": 0.8359,
+      "step": 11827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7603824547398006e-06,
+      "loss": 0.8242,
+      "step": 11828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7587842757294105e-06,
+      "loss": 0.8906,
+      "step": 11829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757186281270322e-06,
+      "loss": 0.9023,
+      "step": 11830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.755588471418795e-06,
+      "loss": 0.8203,
+      "step": 11831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753990846231097e-06,
+      "loss": 0.7578,
+      "step": 11832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75239340576348e-06,
+      "loss": 0.8789,
+      "step": 11833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.750796150072197e-06,
+      "loss": 0.8789,
+      "step": 11834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7491990792134836e-06,
+      "loss": 0.8203,
+      "step": 11835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747602193243577e-06,
+      "loss": 0.8594,
+      "step": 11836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746005492218707e-06,
+      "loss": 0.8594,
+      "step": 11837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7444089761951e-06,
+      "loss": 0.793,
+      "step": 11838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742812645228962e-06,
+      "loss": 0.9688,
+      "step": 11839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74121649937651e-06,
+      "loss": 0.8203,
+      "step": 11840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7396205386939445e-06,
+      "loss": 0.9102,
+      "step": 11841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.738024763237464e-06,
+      "loss": 0.8281,
+      "step": 11842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.736429173063254e-06,
+      "loss": 0.7383,
+      "step": 11843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.734833768227498e-06,
+      "loss": 0.793,
+      "step": 11844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.73323854878638e-06,
+      "loss": 0.7812,
+      "step": 11845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731643514796059e-06,
+      "loss": 0.8477,
+      "step": 11846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730048666312706e-06,
+      "loss": 0.8477,
+      "step": 11847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728454003392475e-06,
+      "loss": 1.0312,
+      "step": 11848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7268595260915214e-06,
+      "loss": 0.8203,
+      "step": 11849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7252652344659835e-06,
+      "loss": 0.7812,
+      "step": 11850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723671128571999e-06,
+      "loss": 0.8828,
+      "step": 11851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722077208465702e-06,
+      "loss": 0.8789,
+      "step": 11852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7204834742032175e-06,
+      "loss": 0.8555,
+      "step": 11853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7188899258406595e-06,
+      "loss": 0.8711,
+      "step": 11854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717296563434141e-06,
+      "loss": 0.7461,
+      "step": 11855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715703387039766e-06,
+      "loss": 0.8125,
+      "step": 11856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714110396713638e-06,
+      "loss": 0.8008,
+      "step": 11857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.712517592511841e-06,
+      "loss": 0.9062,
+      "step": 11858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710924974490463e-06,
+      "loss": 0.707,
+      "step": 11859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7093325427055865e-06,
+      "loss": 0.8867,
+      "step": 11860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707740297213277e-06,
+      "loss": 0.8984,
+      "step": 11861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706148238069602e-06,
+      "loss": 0.8828,
+      "step": 11862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704556365330621e-06,
+      "loss": 0.8594,
+      "step": 11863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702964679052392e-06,
+      "loss": 0.875,
+      "step": 11864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.70137317929095e-06,
+      "loss": 0.793,
+      "step": 11865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69978186610234e-06,
+      "loss": 0.9883,
+      "step": 11866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698190739542593e-06,
+      "loss": 0.8789,
+      "step": 11867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6965997996677405e-06,
+      "loss": 0.9766,
+      "step": 11868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695009046533794e-06,
+      "loss": 0.8711,
+      "step": 11869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69341848019677e-06,
+      "loss": 0.8398,
+      "step": 11870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691828100712678e-06,
+      "loss": 0.8828,
+      "step": 11871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69023790813751e-06,
+      "loss": 0.9922,
+      "step": 11872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.688647902527264e-06,
+      "loss": 0.8711,
+      "step": 11873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6870580839379264e-06,
+      "loss": 0.9336,
+      "step": 11874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6854684524254805e-06,
+      "loss": 0.8477,
+      "step": 11875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.683879008045892e-06,
+      "loss": 0.7461,
+      "step": 11876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682289750855131e-06,
+      "loss": 0.7773,
+      "step": 11877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68070068090916e-06,
+      "loss": 0.9648,
+      "step": 11878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679111798263934e-06,
+      "loss": 0.7266,
+      "step": 11879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677523102975393e-06,
+      "loss": 0.875,
+      "step": 11880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.675934595099481e-06,
+      "loss": 0.8164,
+      "step": 11881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6743462746921346e-06,
+      "loss": 0.8047,
+      "step": 11882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672758141809279e-06,
+      "loss": 0.8047,
+      "step": 11883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671170196506833e-06,
+      "loss": 0.8945,
+      "step": 11884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669582438840712e-06,
+      "loss": 0.8906,
+      "step": 11885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667994868866826e-06,
+      "loss": 0.9492,
+      "step": 11886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66640748664107e-06,
+      "loss": 0.8086,
+      "step": 11887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664820292219342e-06,
+      "loss": 0.8594,
+      "step": 11888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663233285657529e-06,
+      "loss": 0.7539,
+      "step": 11889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661646467011516e-06,
+      "loss": 0.8438,
+      "step": 11890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660059836337168e-06,
+      "loss": 0.9258,
+      "step": 11891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658473393690359e-06,
+      "loss": 0.9336,
+      "step": 11892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6568871391269485e-06,
+      "loss": 0.8789,
+      "step": 11893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655301072702797e-06,
+      "loss": 0.8594,
+      "step": 11894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653715194473741e-06,
+      "loss": 0.8594,
+      "step": 11895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65212950449563e-06,
+      "loss": 0.8047,
+      "step": 11896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650544002824294e-06,
+      "loss": 0.8672,
+      "step": 11897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648958689515568e-06,
+      "loss": 0.9805,
+      "step": 11898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647373564625265e-06,
+      "loss": 0.9414,
+      "step": 11899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645788628209202e-06,
+      "loss": 0.9609,
+      "step": 11900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644203880323189e-06,
+      "loss": 0.8867,
+      "step": 11901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642619321023031e-06,
+      "loss": 0.9219,
+      "step": 11902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.641034950364515e-06,
+      "loss": 0.8633,
+      "step": 11903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639450768403432e-06,
+      "loss": 0.9922,
+      "step": 11904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637866775195563e-06,
+      "loss": 0.8672,
+      "step": 11905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.636282970796689e-06,
+      "loss": 0.9336,
+      "step": 11906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634699355262569e-06,
+      "loss": 0.75,
+      "step": 11907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6331159286489695e-06,
+      "loss": 0.8164,
+      "step": 11908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631532691011644e-06,
+      "loss": 0.8477,
+      "step": 11909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629949642406345e-06,
+      "loss": 0.8125,
+      "step": 11910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6283667828888065e-06,
+      "loss": 0.8594,
+      "step": 11911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626784112514768e-06,
+      "loss": 0.8867,
+      "step": 11912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6252016313399565e-06,
+      "loss": 0.8477,
+      "step": 11913
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623619339420098e-06,
+      "loss": 0.9766,
+      "step": 11914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6220372368109e-06,
+      "loss": 0.8594,
+      "step": 11915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.620455323568075e-06,
+      "loss": 1.0703,
+      "step": 11916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618873599747323e-06,
+      "loss": 1.0625,
+      "step": 11917
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617292065404345e-06,
+      "loss": 0.7617,
+      "step": 11918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615710720594819e-06,
+      "loss": 0.8008,
+      "step": 11919
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6141295653744335e-06,
+      "loss": 0.9609,
+      "step": 11920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612548599798861e-06,
+      "loss": 0.8867,
+      "step": 11921
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610967823923774e-06,
+      "loss": 0.8008,
+      "step": 11922
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6093872378048275e-06,
+      "loss": 0.8633,
+      "step": 11923
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607806841497678e-06,
+      "loss": 0.8867,
+      "step": 11924
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606226635057981e-06,
+      "loss": 0.875,
+      "step": 11925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.604646618541366e-06,
+      "loss": 0.8711,
+      "step": 11926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603066792003475e-06,
+      "loss": 0.9023,
+      "step": 11927
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601487155499937e-06,
+      "loss": 0.8594,
+      "step": 11928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599907709086373e-06,
+      "loss": 0.7695,
+      "step": 11929
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598328452818394e-06,
+      "loss": 0.8789,
+      "step": 11930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5967493867516085e-06,
+      "loss": 0.8398,
+      "step": 11931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595170510941621e-06,
+      "loss": 0.8398,
+      "step": 11932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 0.832,
+      "step": 11933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592013330314411e-06,
+      "loss": 0.8789,
+      "step": 11934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590435025608354e-06,
+      "loss": 0.8164,
+      "step": 11935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588856911381433e-06,
+      "loss": 0.7734,
+      "step": 11936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587278987689217e-06,
+      "loss": 0.8203,
+      "step": 11937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585701254587264e-06,
+      "loss": 0.918,
+      "step": 11938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584123712131128e-06,
+      "loss": 0.9141,
+      "step": 11939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582546360376357e-06,
+      "loss": 0.8242,
+      "step": 11940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580969199378499e-06,
+      "loss": 0.8086,
+      "step": 11941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579392229193077e-06,
+      "loss": 0.9648,
+      "step": 11942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577815449875625e-06,
+      "loss": 0.8125,
+      "step": 11943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5762388614816625e-06,
+      "loss": 0.8906,
+      "step": 11944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574662464066708e-06,
+      "loss": 0.8477,
+      "step": 11945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573086257686261e-06,
+      "loss": 0.7461,
+      "step": 11946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571510242395826e-06,
+      "loss": 0.8477,
+      "step": 11947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5699344182508964e-06,
+      "loss": 0.8008,
+      "step": 11948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568358785306965e-06,
+      "loss": 0.9336,
+      "step": 11949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5667833436195016e-06,
+      "loss": 0.9258,
+      "step": 11950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5652080932439865e-06,
+      "loss": 0.7539,
+      "step": 11951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563633034235886e-06,
+      "loss": 0.9297,
+      "step": 11952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562058166650664e-06,
+      "loss": 0.8945,
+      "step": 11953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560483490543765e-06,
+      "loss": 0.7969,
+      "step": 11954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558909005970642e-06,
+      "loss": 0.9023,
+      "step": 11955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.557334712986734e-06,
+      "loss": 0.8516,
+      "step": 11956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555760611647477e-06,
+      "loss": 0.8828,
+      "step": 11957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554186702008292e-06,
+      "loss": 0.9922,
+      "step": 11958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552612984124601e-06,
+      "loss": 0.793,
+      "step": 11959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551039458051818e-06,
+      "loss": 0.793,
+      "step": 11960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549466123845353e-06,
+      "loss": 0.8789,
+      "step": 11961
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547892981560597e-06,
+      "loss": 0.7578,
+      "step": 11962
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546320031252947e-06,
+      "loss": 0.7773,
+      "step": 11963
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54474727297779e-06,
+      "loss": 0.9844,
+      "step": 11964
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543174706790507e-06,
+      "loss": 0.8555,
+      "step": 11965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.541602332746464e-06,
+      "loss": 0.8633,
+      "step": 11966
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540030150901032e-06,
+      "loss": 0.9453,
+      "step": 11967
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538458161309568e-06,
+      "loss": 0.7969,
+      "step": 11968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536886364027428e-06,
+      "loss": 0.9258,
+      "step": 11969
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535314759109951e-06,
+      "loss": 0.9922,
+      "step": 11970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533743346612478e-06,
+      "loss": 0.7773,
+      "step": 11971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532172126590343e-06,
+      "loss": 0.8945,
+      "step": 11972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530601099098873e-06,
+      "loss": 0.8281,
+      "step": 11973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529030264193378e-06,
+      "loss": 0.8945,
+      "step": 11974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527459621929176e-06,
+      "loss": 0.8594,
+      "step": 11975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52588917236157e-06,
+      "loss": 0.8164,
+      "step": 11976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524318915545861e-06,
+      "loss": 0.8359,
+      "step": 11977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522748851537334e-06,
+      "loss": 0.7812,
+      "step": 11978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.521178980391278e-06,
+      "loss": 0.8789,
+      "step": 11979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.519609302162972e-06,
+      "loss": 0.8633,
+      "step": 11980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518039816907681e-06,
+      "loss": 0.7969,
+      "step": 11981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5164705246806715e-06,
+      "loss": 0.9531,
+      "step": 11982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514901425537201e-06,
+      "loss": 0.8477,
+      "step": 11983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513332519532526e-06,
+      "loss": 0.7227,
+      "step": 11984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5117638067218795e-06,
+      "loss": 0.8594,
+      "step": 11985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5101952871605026e-06,
+      "loss": 0.8711,
+      "step": 11986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508626960903627e-06,
+      "loss": 0.9297,
+      "step": 11987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507058828006478e-06,
+      "loss": 0.9219,
+      "step": 11988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5054908885242646e-06,
+      "loss": 0.8125,
+      "step": 11989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503923142512201e-06,
+      "loss": 0.8594,
+      "step": 11990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502355590025489e-06,
+      "loss": 0.9492,
+      "step": 11991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500788231119329e-06,
+      "loss": 0.8203,
+      "step": 11992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499221065848902e-06,
+      "loss": 0.7695,
+      "step": 11993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497654094269396e-06,
+      "loss": 0.8867,
+      "step": 11994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496087316435983e-06,
+      "loss": 0.8047,
+      "step": 11995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494520732403838e-06,
+      "loss": 0.832,
+      "step": 11996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492954342228115e-06,
+      "loss": 0.8672,
+      "step": 11997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491388145963973e-06,
+      "loss": 0.8867,
+      "step": 11998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489822143666559e-06,
+      "loss": 0.9258,
+      "step": 11999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488256335391019e-06,
+      "loss": 0.7812,
+      "step": 12000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486690721192479e-06,
+      "loss": 0.8008,
+      "step": 12001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485125301126072e-06,
+      "loss": 0.9102,
+      "step": 12002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483560075246918e-06,
+      "loss": 0.7695,
+      "step": 12003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481995043610134e-06,
+      "loss": 0.8672,
+      "step": 12004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480430206270822e-06,
+      "loss": 0.832,
+      "step": 12005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478865563284084e-06,
+      "loss": 1.0234,
+      "step": 12006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477301114705014e-06,
+      "loss": 0.8398,
+      "step": 12007
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475736860588703e-06,
+      "loss": 0.9023,
+      "step": 12008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474172800990222e-06,
+      "loss": 0.8047,
+      "step": 12009
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4726089359646495e-06,
+      "loss": 0.9883,
+      "step": 12010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47104526556705e-06,
+      "loss": 0.8047,
+      "step": 12011
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.469481789852487e-06,
+      "loss": 0.8164,
+      "step": 12012
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4679185088760055e-06,
+      "loss": 0.8945,
+      "step": 12013
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466355422692656e-06,
+      "loss": 0.8867,
+      "step": 12014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.464792531357474e-06,
+      "loss": 0.7734,
+      "step": 12015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463229834925499e-06,
+      "loss": 0.7695,
+      "step": 12016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4616673334517455e-06,
+      "loss": 0.7812,
+      "step": 12017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460105026991236e-06,
+      "loss": 0.8359,
+      "step": 12018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458542915598982e-06,
+      "loss": 0.8594,
+      "step": 12019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456980999329993e-06,
+      "loss": 0.8594,
+      "step": 12020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455419278239257e-06,
+      "loss": 0.8984,
+      "step": 12021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4538577523817695e-06,
+      "loss": 0.8125,
+      "step": 12022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452296421812513e-06,
+      "loss": 0.707,
+      "step": 12023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4507352865864686e-06,
+      "loss": 0.9453,
+      "step": 12024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449174346758599e-06,
+      "loss": 0.8594,
+      "step": 12025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.447613602383872e-06,
+      "loss": 0.8359,
+      "step": 12026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446053053517243e-06,
+      "loss": 0.8281,
+      "step": 12027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444492700213664e-06,
+      "loss": 0.793,
+      "step": 12028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4429325425280725e-06,
+      "loss": 0.8047,
+      "step": 12029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441372580515405e-06,
+      "loss": 0.8281,
+      "step": 12030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4398128142305915e-06,
+      "loss": 0.7305,
+      "step": 12031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438253243728558e-06,
+      "loss": 0.8711,
+      "step": 12032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43669386906421e-06,
+      "loss": 0.8125,
+      "step": 12033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435134690292462e-06,
+      "loss": 0.7969,
+      "step": 12034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433575707468214e-06,
+      "loss": 0.8359,
+      "step": 12035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432016920646363e-06,
+      "loss": 0.9688,
+      "step": 12036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4304583298817885e-06,
+      "loss": 0.8086,
+      "step": 12037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428899935229377e-06,
+      "loss": 0.8086,
+      "step": 12038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427341736744003e-06,
+      "loss": 1.0781,
+      "step": 12039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425783734480527e-06,
+      "loss": 0.9258,
+      "step": 12040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424225928493814e-06,
+      "loss": 0.9609,
+      "step": 12041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422668318838713e-06,
+      "loss": 0.8359,
+      "step": 12042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421110905570078e-06,
+      "loss": 0.7227,
+      "step": 12043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419553688742736e-06,
+      "loss": 0.8359,
+      "step": 12044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417996668411525e-06,
+      "loss": 0.8867,
+      "step": 12045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416439844631271e-06,
+      "loss": 0.9258,
+      "step": 12046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414883217456793e-06,
+      "loss": 0.8086,
+      "step": 12047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413326786942898e-06,
+      "loss": 0.8984,
+      "step": 12048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411770553144392e-06,
+      "loss": 0.9023,
+      "step": 12049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4102145161160734e-06,
+      "loss": 0.875,
+      "step": 12050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4086586759127355e-06,
+      "loss": 0.7891,
+      "step": 12051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407103032589156e-06,
+      "loss": 0.8945,
+      "step": 12052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405547586200113e-06,
+      "loss": 0.9414,
+      "step": 12053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403992336800378e-06,
+      "loss": 0.7891,
+      "step": 12054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402437284444716e-06,
+      "loss": 0.8867,
+      "step": 12055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.400882429187877e-06,
+      "loss": 0.9531,
+      "step": 12056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399327771084613e-06,
+      "loss": 0.9141,
+      "step": 12057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397773310189665e-06,
+      "loss": 0.918,
+      "step": 12058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.396219046557771e-06,
+      "loss": 0.8086,
+      "step": 12059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394664980243655e-06,
+      "loss": 0.7773,
+      "step": 12060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.393111111302039e-06,
+      "loss": 0.8906,
+      "step": 12061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391557439787637e-06,
+      "loss": 0.9883,
+      "step": 12062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3900039657551606e-06,
+      "loss": 0.6914,
+      "step": 12063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388450689259304e-06,
+      "loss": 0.8555,
+      "step": 12064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386897610354761e-06,
+      "loss": 0.8438,
+      "step": 12065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385344729096221e-06,
+      "loss": 0.8555,
+      "step": 12066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383792045538367e-06,
+      "loss": 0.9688,
+      "step": 12067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38223955973586e-06,
+      "loss": 0.8516,
+      "step": 12068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380687271743373e-06,
+      "loss": 0.8242,
+      "step": 12069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379135181615565e-06,
+      "loss": 0.7812,
+      "step": 12070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377583289407088e-06,
+      "loss": 0.9219,
+      "step": 12071
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376031595172581e-06,
+      "loss": 0.8672,
+      "step": 12072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374480098966685e-06,
+      "loss": 0.75,
+      "step": 12073
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.37292880084403e-06,
+      "loss": 0.7852,
+      "step": 12074
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371377700859244e-06,
+      "loss": 0.8281,
+      "step": 12075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3698267990669355e-06,
+      "loss": 0.9023,
+      "step": 12076
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368276095521719e-06,
+      "loss": 0.9297,
+      "step": 12077
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366725590278197e-06,
+      "loss": 0.9258,
+      "step": 12078
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365175283390968e-06,
+      "loss": 0.8828,
+      "step": 12079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363625174914613e-06,
+      "loss": 0.8906,
+      "step": 12080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362075264903719e-06,
+      "loss": 0.7617,
+      "step": 12081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36052555341286e-06,
+      "loss": 0.8867,
+      "step": 12082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3589760404966065e-06,
+      "loss": 0.8945,
+      "step": 12083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3574267262095125e-06,
+      "loss": 0.8789,
+      "step": 12084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355877610606136e-06,
+      "loss": 0.8281,
+      "step": 12085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354328693741025e-06,
+      "loss": 0.8516,
+      "step": 12086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352779975668718e-06,
+      "loss": 0.8203,
+      "step": 12087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351231456443747e-06,
+      "loss": 0.7539,
+      "step": 12088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349683136120636e-06,
+      "loss": 0.7656,
+      "step": 12089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348135014753906e-06,
+      "loss": 0.8203,
+      "step": 12090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346587092398073e-06,
+      "loss": 0.9609,
+      "step": 12091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345039369107633e-06,
+      "loss": 0.8711,
+      "step": 12092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3434918449370874e-06,
+      "loss": 0.8633,
+      "step": 12093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341944519940933e-06,
+      "loss": 0.8867,
+      "step": 12094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340397394173643e-06,
+      "loss": 0.8672,
+      "step": 12095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3388504676897e-06,
+      "loss": 0.8594,
+      "step": 12096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337303740543573e-06,
+      "loss": 0.7227,
+      "step": 12097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335757212789727e-06,
+      "loss": 0.7383,
+      "step": 12098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.334210884482612e-06,
+      "loss": 0.8945,
+      "step": 12099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3326647556766796e-06,
+      "loss": 0.8672,
+      "step": 12100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331118826426373e-06,
+      "loss": 0.7539,
+      "step": 12101
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.329573096786127e-06,
+      "loss": 0.8398,
+      "step": 12102
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328027566810364e-06,
+      "loss": 0.957,
+      "step": 12103
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3264822365535084e-06,
+      "loss": 0.7344,
+      "step": 12104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324937106069973e-06,
+      "loss": 0.8906,
+      "step": 12105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.323392175414169e-06,
+      "loss": 0.9062,
+      "step": 12106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321847444640486e-06,
+      "loss": 0.8203,
+      "step": 12107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320302913803322e-06,
+      "loss": 0.7852,
+      "step": 12108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318758582957061e-06,
+      "loss": 0.793,
+      "step": 12109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317214452156086e-06,
+      "loss": 0.9805,
+      "step": 12110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315670521454762e-06,
+      "loss": 1.0156,
+      "step": 12111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.314126790907455e-06,
+      "loss": 0.8438,
+      "step": 12112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3125832605685224e-06,
+      "loss": 0.9492,
+      "step": 12113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311039930492319e-06,
+      "loss": 0.7539,
+      "step": 12114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309496800733178e-06,
+      "loss": 0.8828,
+      "step": 12115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307953871345443e-06,
+      "loss": 0.7891,
+      "step": 12116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306411142383439e-06,
+      "loss": 0.875,
+      "step": 12117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304868613901494e-06,
+      "loss": 0.8945,
+      "step": 12118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303326285953914e-06,
+      "loss": 0.7656,
+      "step": 12119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301784158595011e-06,
+      "loss": 0.7695,
+      "step": 12120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300242231879086e-06,
+      "loss": 0.8281,
+      "step": 12121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298700505860437e-06,
+      "loss": 0.8906,
+      "step": 12122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2971589805933415e-06,
+      "loss": 1.0078,
+      "step": 12123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295617656132084e-06,
+      "loss": 0.9023,
+      "step": 12124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.294076532530936e-06,
+      "loss": 0.918,
+      "step": 12125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2925356098441665e-06,
+      "loss": 0.8828,
+      "step": 12126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290994888126027e-06,
+      "loss": 0.7891,
+      "step": 12127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289454367430772e-06,
+      "loss": 0.8086,
+      "step": 12128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287914047812647e-06,
+      "loss": 0.8672,
+      "step": 12129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28637392932589e-06,
+      "loss": 0.9648,
+      "step": 12130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284834012024726e-06,
+      "loss": 0.7383,
+      "step": 12131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283294295963381e-06,
+      "loss": 0.918,
+      "step": 12132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281754781196069e-06,
+      "loss": 0.7695,
+      "step": 12133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280215467777005e-06,
+      "loss": 0.7891,
+      "step": 12134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278676355760382e-06,
+      "loss": 0.9648,
+      "step": 12135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2771374452003964e-06,
+      "loss": 0.8086,
+      "step": 12136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275598736151239e-06,
+      "loss": 0.7617,
+      "step": 12137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.274060228667091e-06,
+      "loss": 0.9219,
+      "step": 12138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272521922802122e-06,
+      "loss": 0.8828,
+      "step": 12139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270983818610498e-06,
+      "loss": 0.9297,
+      "step": 12140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2694459161463786e-06,
+      "loss": 0.8672,
+      "step": 12141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267908215463921e-06,
+      "loss": 0.8984,
+      "step": 12142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266370716617262e-06,
+      "loss": 0.8125,
+      "step": 12143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264833419660541e-06,
+      "loss": 0.9375,
+      "step": 12144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263296324647892e-06,
+      "loss": 0.7773,
+      "step": 12145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2617594316334395e-06,
+      "loss": 0.8203,
+      "step": 12146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.260222740671295e-06,
+      "loss": 0.832,
+      "step": 12147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258686251815569e-06,
+      "loss": 0.8828,
+      "step": 12148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257149965120369e-06,
+      "loss": 0.8438,
+      "step": 12149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255613880639781e-06,
+      "loss": 0.8281,
+      "step": 12150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254077998427898e-06,
+      "loss": 0.8633,
+      "step": 12151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2525423185388e-06,
+      "loss": 0.7969,
+      "step": 12152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251006841026566e-06,
+      "loss": 0.8477,
+      "step": 12153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249471565945252e-06,
+      "loss": 0.9609,
+      "step": 12154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247936493348924e-06,
+      "loss": 0.8828,
+      "step": 12155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246401623291634e-06,
+      "loss": 0.8477,
+      "step": 12156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2448669558274295e-06,
+      "loss": 0.8555,
+      "step": 12157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243332491010341e-06,
+      "loss": 0.8594,
+      "step": 12158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241798228894404e-06,
+      "loss": 0.8789,
+      "step": 12159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2402641695336424e-06,
+      "loss": 0.8516,
+      "step": 12160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238730312982076e-06,
+      "loss": 0.8516,
+      "step": 12161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237196659293707e-06,
+      "loss": 0.8086,
+      "step": 12162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235663208522542e-06,
+      "loss": 0.875,
+      "step": 12163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234129960722574e-06,
+      "loss": 0.7852,
+      "step": 12164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232596915947797e-06,
+      "loss": 0.8438,
+      "step": 12165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231064074252184e-06,
+      "loss": 0.7734,
+      "step": 12166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229531435689711e-06,
+      "loss": 0.8008,
+      "step": 12167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.227999000314346e-06,
+      "loss": 0.9414,
+      "step": 12168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.226466768180053e-06,
+      "loss": 0.8633,
+      "step": 12169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224934739340776e-06,
+      "loss": 0.9453,
+      "step": 12170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2234029138504625e-06,
+      "loss": 0.793,
+      "step": 12171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2218712917630525e-06,
+      "loss": 0.9023,
+      "step": 12172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220339873132477e-06,
+      "loss": 0.8555,
+      "step": 12173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218808658012657e-06,
+      "loss": 0.7578,
+      "step": 12174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2172776464575095e-06,
+      "loss": 0.832,
+      "step": 12175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.215746838520947e-06,
+      "loss": 0.8867,
+      "step": 12176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214216234256868e-06,
+      "loss": 0.875,
+      "step": 12177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2126858337191675e-06,
+      "loss": 0.8125,
+      "step": 12178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211155636961735e-06,
+      "loss": 0.8867,
+      "step": 12179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209625644038453e-06,
+      "loss": 0.875,
+      "step": 12180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208095855003189e-06,
+      "loss": 0.8789,
+      "step": 12181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206566269909812e-06,
+      "loss": 0.9258,
+      "step": 12182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205036888812183e-06,
+      "loss": 0.9609,
+      "step": 12183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203507711764155e-06,
+      "loss": 0.8398,
+      "step": 12184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201978738819567e-06,
+      "loss": 0.7148,
+      "step": 12185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.200449970032261e-06,
+      "loss": 0.8477,
+      "step": 12186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198921405456065e-06,
+      "loss": 0.8359,
+      "step": 12187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197393045144806e-06,
+      "loss": 0.8398,
+      "step": 12188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.195864889152295e-06,
+      "loss": 0.8086,
+      "step": 12189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194336937532342e-06,
+      "loss": 0.8945,
+      "step": 12190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192809190338753e-06,
+      "loss": 0.9141,
+      "step": 12191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1912816476253145e-06,
+      "loss": 0.9023,
+      "step": 12192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189754309445818e-06,
+      "loss": 0.9062,
+      "step": 12193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188227175854044e-06,
+      "loss": 0.8984,
+      "step": 12194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1867002469037676e-06,
+      "loss": 0.8867,
+      "step": 12195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185173522648748e-06,
+      "loss": 0.8633,
+      "step": 12196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183647003142747e-06,
+      "loss": 0.8984,
+      "step": 12197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182120688439515e-06,
+      "loss": 0.7578,
+      "step": 12198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180594578592799e-06,
+      "loss": 0.8203,
+      "step": 12199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179068673656332e-06,
+      "loss": 0.75,
+      "step": 12200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177542973683841e-06,
+      "loss": 0.7773,
+      "step": 12201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176017478729054e-06,
+      "loss": 1.0703,
+      "step": 12202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174492188845687e-06,
+      "loss": 0.7656,
+      "step": 12203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172967104087441e-06,
+      "loss": 0.9805,
+      "step": 12204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17144222450802e-06,
+      "loss": 0.8281,
+      "step": 12205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169917550161121e-06,
+      "loss": 0.7891,
+      "step": 12206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.168393081100424e-06,
+      "loss": 0.8477,
+      "step": 12207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16686881737961e-06,
+      "loss": 0.875,
+      "step": 12208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165344759052351e-06,
+      "loss": 0.8828,
+      "step": 12209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163820906172315e-06,
+      "loss": 0.875,
+      "step": 12210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162297258793153e-06,
+      "loss": 0.7773,
+      "step": 12211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1607738169685186e-06,
+      "loss": 0.9375,
+      "step": 12212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159250580752053e-06,
+      "loss": 0.8906,
+      "step": 12213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157727550197395e-06,
+      "loss": 0.7695,
+      "step": 12214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156204725358168e-06,
+      "loss": 0.7656,
+      "step": 12215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154682106287996e-06,
+      "loss": 1.0156,
+      "step": 12216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153159693040495e-06,
+      "loss": 0.8047,
+      "step": 12217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151637485669265e-06,
+      "loss": 0.8672,
+      "step": 12218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1501154842279104e-06,
+      "loss": 0.8281,
+      "step": 12219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148593688770019e-06,
+      "loss": 0.7461,
+      "step": 12220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1470720993491844e-06,
+      "loss": 0.8555,
+      "step": 12221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145550716018974e-06,
+      "loss": 1.0234,
+      "step": 12222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144029538832962e-06,
+      "loss": 0.9297,
+      "step": 12223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.14250856784471e-06,
+      "loss": 0.8672,
+      "step": 12224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140987803107781e-06,
+      "loss": 0.8633,
+      "step": 12225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139467244675713e-06,
+      "loss": 0.8633,
+      "step": 12226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1379468926020525e-06,
+      "loss": 0.7734,
+      "step": 12227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.136426746940333e-06,
+      "loss": 0.7461,
+      "step": 12228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134906807744084e-06,
+      "loss": 0.8008,
+      "step": 12229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133387075066819e-06,
+      "loss": 0.8477,
+      "step": 12230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131867548962053e-06,
+      "loss": 0.8789,
+      "step": 12231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1303482294832946e-06,
+      "loss": 0.9609,
+      "step": 12232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1288291166840345e-06,
+      "loss": 0.832,
+      "step": 12233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1273102106177655e-06,
+      "loss": 0.8555,
+      "step": 12234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.125791511337972e-06,
+      "loss": 0.8516,
+      "step": 12235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124273018898132e-06,
+      "loss": 0.832,
+      "step": 12236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122754733351708e-06,
+      "loss": 0.8633,
+      "step": 12237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.121236654752164e-06,
+      "loss": 0.8789,
+      "step": 12238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119718783152955e-06,
+      "loss": 0.8203,
+      "step": 12239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118201118607529e-06,
+      "loss": 0.8867,
+      "step": 12240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116683661169321e-06,
+      "loss": 0.7891,
+      "step": 12241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115166410891765e-06,
+      "loss": 0.8047,
+      "step": 12242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113649367828284e-06,
+      "loss": 0.9375,
+      "step": 12243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112132532032302e-06,
+      "loss": 0.7734,
+      "step": 12244
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110615903557221e-06,
+      "loss": 0.9922,
+      "step": 12245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109099482456446e-06,
+      "loss": 0.9727,
+      "step": 12246
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107583268783376e-06,
+      "loss": 0.9258,
+      "step": 12247
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106067262591399e-06,
+      "loss": 0.8359,
+      "step": 12248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104551463933891e-06,
+      "loss": 0.8242,
+      "step": 12249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103035872864229e-06,
+      "loss": 0.8008,
+      "step": 12250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.101520489435777e-06,
+      "loss": 0.8242,
+      "step": 12251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1000053137019004e-06,
+      "loss": 0.8438,
+      "step": 12252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098490345715943e-06,
+      "loss": 0.9414,
+      "step": 12253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096975585531252e-06,
+      "loss": 0.8242,
+      "step": 12254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095461033201165e-06,
+      "loss": 0.8203,
+      "step": 12255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093946688779014e-06,
+      "loss": 0.8672,
+      "step": 12256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.092432552318117e-06,
+      "loss": 0.7578,
+      "step": 12257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090918623871791e-06,
+      "loss": 0.9023,
+      "step": 12258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089404903493344e-06,
+      "loss": 0.9141,
+      "step": 12259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087891391236079e-06,
+      "loss": 0.8867,
+      "step": 12260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0863780871532835e-06,
+      "loss": 0.8125,
+      "step": 12261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084864991298246e-06,
+      "loss": 0.9609,
+      "step": 12262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083352103724249e-06,
+      "loss": 0.9141,
+      "step": 12263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.081839424484555e-06,
+      "loss": 0.8359,
+      "step": 12264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080326953632433e-06,
+      "loss": 0.9023,
+      "step": 12265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078814691221139e-06,
+      "loss": 1.0,
+      "step": 12266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077302637303926e-06,
+      "loss": 0.7148,
+      "step": 12267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075790791934028e-06,
+      "loss": 0.8789,
+      "step": 12268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074279155164682e-06,
+      "loss": 0.9062,
+      "step": 12269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0727677270491175e-06,
+      "loss": 0.8594,
+      "step": 12270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071256507640556e-06,
+      "loss": 0.8867,
+      "step": 12271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069745496992204e-06,
+      "loss": 0.7617,
+      "step": 12272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068234695157269e-06,
+      "loss": 0.8086,
+      "step": 12273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066724102188949e-06,
+      "loss": 0.7812,
+      "step": 12274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065213718140437e-06,
+      "loss": 0.9375,
+      "step": 12275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063703543064911e-06,
+      "loss": 0.7969,
+      "step": 12276
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062193577015549e-06,
+      "loss": 0.875,
+      "step": 12277
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060683820045518e-06,
+      "loss": 0.8672,
+      "step": 12278
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059174272207984e-06,
+      "loss": 0.7539,
+      "step": 12279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057664933556092e-06,
+      "loss": 0.9336,
+      "step": 12280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056155804142994e-06,
+      "loss": 0.8008,
+      "step": 12281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054646884021827e-06,
+      "loss": 0.8086,
+      "step": 12282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0531381732457265e-06,
+      "loss": 0.8594,
+      "step": 12283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051629671867809e-06,
+      "loss": 0.8828,
+      "step": 12284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050121379941195e-06,
+      "loss": 0.8867,
+      "step": 12285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048613297518994e-06,
+      "loss": 0.9961,
+      "step": 12286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047105424654312e-06,
+      "loss": 0.9219,
+      "step": 12287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045597761400234e-06,
+      "loss": 0.8555,
+      "step": 12288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044090307809854e-06,
+      "loss": 0.8867,
+      "step": 12289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0425830639362485e-06,
+      "loss": 0.8203,
+      "step": 12290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041076029832496e-06,
+      "loss": 0.8398,
+      "step": 12291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039569205551653e-06,
+      "loss": 0.8828,
+      "step": 12292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038062591146783e-06,
+      "loss": 0.8242,
+      "step": 12293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036556186670933e-06,
+      "loss": 0.7422,
+      "step": 12294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0350499921771506e-06,
+      "loss": 0.8086,
+      "step": 12295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033544007718464e-06,
+      "loss": 0.8281,
+      "step": 12296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032038233347906e-06,
+      "loss": 0.8281,
+      "step": 12297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030532669118495e-06,
+      "loss": 0.9297,
+      "step": 12298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029027315083251e-06,
+      "loss": 0.8125,
+      "step": 12299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02752217129517e-06,
+      "loss": 0.9531,
+      "step": 12300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0260172378072536e-06,
+      "loss": 0.9844,
+      "step": 12301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024512514672495e-06,
+      "loss": 0.7578,
+      "step": 12302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023008001943881e-06,
+      "loss": 0.8438,
+      "step": 12303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021503699674379e-06,
+      "loss": 0.8711,
+      "step": 12304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019999607916964e-06,
+      "loss": 0.8516,
+      "step": 12305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018495726724595e-06,
+      "loss": 0.8086,
+      "step": 12306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01699205615023e-06,
+      "loss": 0.8398,
+      "step": 12307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015488596246811e-06,
+      "loss": 0.8047,
+      "step": 12308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013985347067276e-06,
+      "loss": 0.8633,
+      "step": 12309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01248230866456e-06,
+      "loss": 0.8125,
+      "step": 12310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010979481091591e-06,
+      "loss": 0.8164,
+      "step": 12311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009476864401277e-06,
+      "loss": 0.8945,
+      "step": 12312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007974458646533e-06,
+      "loss": 0.875,
+      "step": 12313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006472263880259e-06,
+      "loss": 0.7891,
+      "step": 12314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004970280155355e-06,
+      "loss": 0.8281,
+      "step": 12315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003468507524699e-06,
+      "loss": 0.8086,
+      "step": 12316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001966946041177e-06,
+      "loss": 0.8633,
+      "step": 12317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000465595757662e-06,
+      "loss": 0.9531,
+      "step": 12318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9989644567270134e-06,
+      "loss": 0.8906,
+      "step": 12319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997463529002093e-06,
+      "loss": 0.8047,
+      "step": 12320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995962812635747e-06,
+      "loss": 0.8203,
+      "step": 12321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994462307680827e-06,
+      "loss": 0.875,
+      "step": 12322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992962014190156e-06,
+      "loss": 0.8398,
+      "step": 12323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991461932216568e-06,
+      "loss": 0.9297,
+      "step": 12324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989962061812883e-06,
+      "loss": 0.8203,
+      "step": 12325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988462403031916e-06,
+      "loss": 0.8828,
+      "step": 12326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986962955926467e-06,
+      "loss": 0.7617,
+      "step": 12327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985463720549336e-06,
+      "loss": 0.8828,
+      "step": 12328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983964696953314e-06,
+      "loss": 0.8359,
+      "step": 12329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982465885191188e-06,
+      "loss": 0.7656,
+      "step": 12330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980967285315724e-06,
+      "loss": 0.8516,
+      "step": 12331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979468897379697e-06,
+      "loss": 0.793,
+      "step": 12332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977970721435864e-06,
+      "loss": 0.7891,
+      "step": 12333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976472757536984e-06,
+      "loss": 0.9883,
+      "step": 12334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974975005735796e-06,
+      "loss": 0.9453,
+      "step": 12335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973477466085039e-06,
+      "loss": 0.7461,
+      "step": 12336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9719801386374465e-06,
+      "loss": 0.8203,
+      "step": 12337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970483023445745e-06,
+      "loss": 0.8789,
+      "step": 12338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96898612056264e-06,
+      "loss": 0.8125,
+      "step": 12339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967489430040847e-06,
+      "loss": 0.7695,
+      "step": 12340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.965992951933065e-06,
+      "loss": 0.8438,
+      "step": 12341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964496686291991e-06,
+      "loss": 0.8477,
+      "step": 12342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9630006331703045e-06,
+      "loss": 0.9648,
+      "step": 12343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.961504792620686e-06,
+      "loss": 0.7188,
+      "step": 12344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960009164695808e-06,
+      "loss": 0.8594,
+      "step": 12345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958513749448336e-06,
+      "loss": 0.8906,
+      "step": 12346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957018546930919e-06,
+      "loss": 0.8633,
+      "step": 12347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95552355719621e-06,
+      "loss": 0.8594,
+      "step": 12348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95402878029685e-06,
+      "loss": 0.9336,
+      "step": 12349
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.952534216285474e-06,
+      "loss": 0.8359,
+      "step": 12350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951039865214702e-06,
+      "loss": 0.918,
+      "step": 12351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949545727137156e-06,
+      "loss": 0.8359,
+      "step": 12352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948051802105447e-06,
+      "loss": 0.8086,
+      "step": 12353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946558090172182e-06,
+      "loss": 0.8672,
+      "step": 12354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945064591389949e-06,
+      "loss": 0.957,
+      "step": 12355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943571305811341e-06,
+      "loss": 0.8242,
+      "step": 12356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942078233488939e-06,
+      "loss": 0.7812,
+      "step": 12357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940585374475317e-06,
+      "loss": 0.9141,
+      "step": 12358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939092728823038e-06,
+      "loss": 0.9141,
+      "step": 12359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937600296584662e-06,
+      "loss": 0.8398,
+      "step": 12360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.936108077812741e-06,
+      "loss": 0.9414,
+      "step": 12361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934616072559818e-06,
+      "loss": 0.8125,
+      "step": 12362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933124280878426e-06,
+      "loss": 0.832,
+      "step": 12363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931632702821095e-06,
+      "loss": 0.8711,
+      "step": 12364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930141338440347e-06,
+      "loss": 0.9258,
+      "step": 12365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928650187788696e-06,
+      "loss": 0.9609,
+      "step": 12366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927159250918644e-06,
+      "loss": 0.9141,
+      "step": 12367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925668527882691e-06,
+      "loss": 0.8711,
+      "step": 12368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924178018733328e-06,
+      "loss": 0.9102,
+      "step": 12369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922687723523041e-06,
+      "loss": 0.9219,
+      "step": 12370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921197642304299e-06,
+      "loss": 0.9375,
+      "step": 12371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919707775129574e-06,
+      "loss": 0.7461,
+      "step": 12372
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91821812205133e-06,
+      "loss": 0.8438,
+      "step": 12373
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916728683122012e-06,
+      "loss": 0.832,
+      "step": 12374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91523945839407e-06,
+      "loss": 0.8281,
+      "step": 12375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9137504479199405e-06,
+      "loss": 0.8203,
+      "step": 12376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91226165175206e-06,
+      "loss": 0.7344,
+      "step": 12377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910773069942841e-06,
+      "loss": 0.793,
+      "step": 12378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9092847025447055e-06,
+      "loss": 0.7266,
+      "step": 12379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907796549610059e-06,
+      "loss": 0.8789,
+      "step": 12380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.906308611191306e-06,
+      "loss": 0.8828,
+      "step": 12381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.904820887340832e-06,
+      "loss": 0.7812,
+      "step": 12382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903333378111025e-06,
+      "loss": 0.7695,
+      "step": 12383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901846083554264e-06,
+      "loss": 0.8555,
+      "step": 12384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90035900372292e-06,
+      "loss": 0.7773,
+      "step": 12385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89887213866935e-06,
+      "loss": 0.8008,
+      "step": 12386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897385488445914e-06,
+      "loss": 0.8828,
+      "step": 12387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895899053104956e-06,
+      "loss": 0.8984,
+      "step": 12388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89441283269882e-06,
+      "loss": 0.8672,
+      "step": 12389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892926827279832e-06,
+      "loss": 0.8906,
+      "step": 12390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8914410369003205e-06,
+      "loss": 0.7266,
+      "step": 12391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8899554616126015e-06,
+      "loss": 0.875,
+      "step": 12392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888470101468988e-06,
+      "loss": 0.9375,
+      "step": 12393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886984956521774e-06,
+      "loss": 0.8984,
+      "step": 12394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885500026823259e-06,
+      "loss": 0.9062,
+      "step": 12395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884015312425728e-06,
+      "loss": 0.9531,
+      "step": 12396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882530813381465e-06,
+      "loss": 0.8398,
+      "step": 12397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881046529742734e-06,
+      "loss": 0.8828,
+      "step": 12398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879562461561802e-06,
+      "loss": 0.918,
+      "step": 12399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878078608890926e-06,
+      "loss": 0.8203,
+      "step": 12400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876594971782358e-06,
+      "loss": 0.8672,
+      "step": 12401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875111550288332e-06,
+      "loss": 0.8555,
+      "step": 12402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873628344461085e-06,
+      "loss": 0.8438,
+      "step": 12403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872145354352843e-06,
+      "loss": 0.8203,
+      "step": 12404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870662580015828e-06,
+      "loss": 0.8125,
+      "step": 12405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8691800215022435e-06,
+      "loss": 0.8555,
+      "step": 12406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867697678864296e-06,
+      "loss": 0.8945,
+      "step": 12407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866215552154184e-06,
+      "loss": 0.7773,
+      "step": 12408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864733641424093e-06,
+      "loss": 0.7812,
+      "step": 12409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863251946726202e-06,
+      "loss": 0.9805,
+      "step": 12410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861770468112684e-06,
+      "loss": 0.7812,
+      "step": 12411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860289205635706e-06,
+      "loss": 0.7773,
+      "step": 12412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858808159347428e-06,
+      "loss": 0.7617,
+      "step": 12413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857327329299993e-06,
+      "loss": 0.8203,
+      "step": 12414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855846715545547e-06,
+      "loss": 0.9336,
+      "step": 12415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854366318136225e-06,
+      "loss": 0.9648,
+      "step": 12416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.852886137124156e-06,
+      "loss": 0.75,
+      "step": 12417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851406172561456e-06,
+      "loss": 0.832,
+      "step": 12418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849926424500237e-06,
+      "loss": 0.8125,
+      "step": 12419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848446892992604e-06,
+      "loss": 0.8594,
+      "step": 12420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846967578090657e-06,
+      "loss": 0.8594,
+      "step": 12421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845488479846478e-06,
+      "loss": 0.75,
+      "step": 12422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844009598312154e-06,
+      "loss": 0.75,
+      "step": 12423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842530933539757e-06,
+      "loss": 0.7969,
+      "step": 12424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841052485581355e-06,
+      "loss": 0.8047,
+      "step": 12425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839574254489e-06,
+      "loss": 0.8555,
+      "step": 12426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8380962403147495e-06,
+      "loss": 0.7969,
+      "step": 12427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836618443110643e-06,
+      "loss": 0.8594,
+      "step": 12428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83514086292872e-06,
+      "loss": 0.8477,
+      "step": 12429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8336634998210045e-06,
+      "loss": 0.918,
+      "step": 12430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832186353839517e-06,
+      "loss": 0.7695,
+      "step": 12431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830709425036273e-06,
+      "loss": 0.8438,
+      "step": 12432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829232713463274e-06,
+      "loss": 0.8242,
+      "step": 12433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827756219172518e-06,
+      "loss": 0.8398,
+      "step": 12434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8262799422159955e-06,
+      "loss": 0.8984,
+      "step": 12435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824803882645692e-06,
+      "loss": 0.7852,
+      "step": 12436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823328040513575e-06,
+      "loss": 0.7773,
+      "step": 12437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821852415871614e-06,
+      "loss": 0.8203,
+      "step": 12438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8203770087717685e-06,
+      "loss": 0.7422,
+      "step": 12439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818901819265993e-06,
+      "loss": 0.8125,
+      "step": 12440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817426847406226e-06,
+      "loss": 0.8789,
+      "step": 12441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815952093244405e-06,
+      "loss": 0.9023,
+      "step": 12442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8144775568324576e-06,
+      "loss": 0.8906,
+      "step": 12443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8130032382223104e-06,
+      "loss": 0.9297,
+      "step": 12444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8115291374658694e-06,
+      "loss": 0.8164,
+      "step": 12445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8100552546150406e-06,
+      "loss": 1.0078,
+      "step": 12446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808581589721725e-06,
+      "loss": 0.9727,
+      "step": 12447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807108142837813e-06,
+      "loss": 0.8867,
+      "step": 12448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805634914015183e-06,
+      "loss": 0.8867,
+      "step": 12449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804161903305711e-06,
+      "loss": 0.8945,
+      "step": 12450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802689110761265e-06,
+      "loss": 0.8672,
+      "step": 12451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801216536433707e-06,
+      "loss": 0.9414,
+      "step": 12452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7997441803748836e-06,
+      "loss": 0.7578,
+      "step": 12453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7982720426366394e-06,
+      "loss": 0.8398,
+      "step": 12454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796800123270813e-06,
+      "loss": 0.9414,
+      "step": 12455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795328422329234e-06,
+      "loss": 0.8789,
+      "step": 12456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7938569398637195e-06,
+      "loss": 0.7734,
+      "step": 12457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7923856759260837e-06,
+      "loss": 0.875,
+      "step": 12458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.790914630568132e-06,
+      "loss": 0.8086,
+      "step": 12459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.789443803841668e-06,
+      "loss": 0.8438,
+      "step": 12460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7879731957984724e-06,
+      "loss": 0.7695,
+      "step": 12461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.786502806490332e-06,
+      "loss": 0.7891,
+      "step": 12462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7850326359690214e-06,
+      "loss": 0.8516,
+      "step": 12463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7835626842863105e-06,
+      "loss": 0.9531,
+      "step": 12464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782092951493953e-06,
+      "loss": 0.7422,
+      "step": 12465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7806234376437023e-06,
+      "loss": 0.9062,
+      "step": 12466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7791541427873036e-06,
+      "loss": 0.7969,
+      "step": 12467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7776850669764965e-06,
+      "loss": 0.6562,
+      "step": 12468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.776216210263001e-06,
+      "loss": 0.8828,
+      "step": 12469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774747572698543e-06,
+      "loss": 0.7891,
+      "step": 12470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7732791543348333e-06,
+      "loss": 0.8125,
+      "step": 12471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7718109552235837e-06,
+      "loss": 0.8242,
+      "step": 12472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.770342975416482e-06,
+      "loss": 1.0,
+      "step": 12473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768875214965223e-06,
+      "loss": 0.8281,
+      "step": 12474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767407673921489e-06,
+      "loss": 1.0,
+      "step": 12475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7659403523369574e-06,
+      "loss": 0.875,
+      "step": 12476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7644732502632873e-06,
+      "loss": 0.9414,
+      "step": 12477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.763006367752142e-06,
+      "loss": 0.7852,
+      "step": 12478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.761539704855173e-06,
+      "loss": 0.8672,
+      "step": 12479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7600732616240254e-06,
+      "loss": 0.7969,
+      "step": 12480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75860703811033e-06,
+      "loss": 0.8984,
+      "step": 12481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.757141034365718e-06,
+      "loss": 0.8789,
+      "step": 12482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7556752504418083e-06,
+      "loss": 0.8672,
+      "step": 12483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7542096863902177e-06,
+      "loss": 0.7969,
+      "step": 12484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7527443422625444e-06,
+      "loss": 0.875,
+      "step": 12485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7512792181103872e-06,
+      "loss": 0.8711,
+      "step": 12486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7498143139853414e-06,
+      "loss": 0.9609,
+      "step": 12487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748349629938981e-06,
+      "loss": 0.8281,
+      "step": 12488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.746885166022882e-06,
+      "loss": 0.8125,
+      "step": 12489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745420922288611e-06,
+      "loss": 0.8594,
+      "step": 12490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7439568987877294e-06,
+      "loss": 0.8828,
+      "step": 12491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742493095571782e-06,
+      "loss": 0.8047,
+      "step": 12492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7410295126923145e-06,
+      "loss": 0.8828,
+      "step": 12493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7395661502008607e-06,
+      "loss": 0.7656,
+      "step": 12494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7381030081489534e-06,
+      "loss": 0.7969,
+      "step": 12495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7366400865881025e-06,
+      "loss": 0.957,
+      "step": 12496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.735177385569826e-06,
+      "loss": 0.918,
+      "step": 12497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7337149051456267e-06,
+      "loss": 0.918,
+      "step": 12498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.732252645367004e-06,
+      "loss": 0.957,
+      "step": 12499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7307906062854395e-06,
+      "loss": 0.8047,
+      "step": 12500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7293287879524165e-06,
+      "loss": 0.8555,
+      "step": 12501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7278671904194098e-06,
+      "loss": 0.9883,
+      "step": 12502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7264058137378866e-06,
+      "loss": 0.8047,
+      "step": 12503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7249446579592972e-06,
+      "loss": 0.9062,
+      "step": 12504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723483723135096e-06,
+      "loss": 0.8242,
+      "step": 12505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.722023009316723e-06,
+      "loss": 0.9258,
+      "step": 12506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7205625165556167e-06,
+      "loss": 0.8125,
+      "step": 12507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7191022449031965e-06,
+      "loss": 0.7969,
+      "step": 12508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7176421944108844e-06,
+      "loss": 0.8516,
+      "step": 12509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.71618236513009e-06,
+      "loss": 0.7773,
+      "step": 12510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.71472275711222e-06,
+      "loss": 0.8828,
+      "step": 12511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.713263370408663e-06,
+      "loss": 0.8398,
+      "step": 12512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.711804205070809e-06,
+      "loss": 0.7812,
+      "step": 12513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.710345261150038e-06,
+      "loss": 0.9375,
+      "step": 12514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7088865386977246e-06,
+      "loss": 0.918,
+      "step": 12515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.707428037765227e-06,
+      "loss": 0.8203,
+      "step": 12516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7059697584039034e-06,
+      "loss": 0.7969,
+      "step": 12517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7045117006651033e-06,
+      "loss": 0.8906,
+      "step": 12518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.703053864600169e-06,
+      "loss": 0.9648,
+      "step": 12519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7015962502604275e-06,
+      "loss": 0.8438,
+      "step": 12520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7001388576972074e-06,
+      "loss": 0.9297,
+      "step": 12521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6986816869618268e-06,
+      "loss": 0.8672,
+      "step": 12522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6972247381055917e-06,
+      "loss": 0.9023,
+      "step": 12523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6957680111798034e-06,
+      "loss": 0.8125,
+      "step": 12524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.694311506235758e-06,
+      "loss": 0.8203,
+      "step": 12525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692855223324744e-06,
+      "loss": 0.8555,
+      "step": 12526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6913991624980317e-06,
+      "loss": 0.8945,
+      "step": 12527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.689943323806896e-06,
+      "loss": 0.7539,
+      "step": 12528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.688487707302598e-06,
+      "loss": 0.8711,
+      "step": 12529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6870323130363972e-06,
+      "loss": 0.875,
+      "step": 12530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.685577141059532e-06,
+      "loss": 0.9102,
+      "step": 12531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.684122191423244e-06,
+      "loss": 0.9492,
+      "step": 12532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6826674641787663e-06,
+      "loss": 0.9023,
+      "step": 12533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.681212959377325e-06,
+      "loss": 0.8828,
+      "step": 12534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.679758677070128e-06,
+      "loss": 0.832,
+      "step": 12535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6783046173083858e-06,
+      "loss": 0.8711,
+      "step": 12536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.676850780143302e-06,
+      "loss": 0.8203,
+      "step": 12537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.675397165626062e-06,
+      "loss": 0.8594,
+      "step": 12538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.673943773807853e-06,
+      "loss": 0.8281,
+      "step": 12539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6724906047398513e-06,
+      "loss": 0.8906,
+      "step": 12540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6710376584732277e-06,
+      "loss": 0.832,
+      "step": 12541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6695849350591374e-06,
+      "loss": 0.8906,
+      "step": 12542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6681324345487357e-06,
+      "loss": 0.7656,
+      "step": 12543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6666801569931677e-06,
+      "loss": 0.8594,
+      "step": 12544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6652281024435723e-06,
+      "loss": 0.8711,
+      "step": 12545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6637762709510737e-06,
+      "loss": 0.8828,
+      "step": 12546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.662324662566795e-06,
+      "loss": 0.8086,
+      "step": 12547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660873277341851e-06,
+      "loss": 0.8398,
+      "step": 12548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.659422115327349e-06,
+      "loss": 0.8672,
+      "step": 12549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.657971176574382e-06,
+      "loss": 0.8672,
+      "step": 12550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.656520461134041e-06,
+      "loss": 0.9375,
+      "step": 12551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6550699690574133e-06,
+      "loss": 0.7305,
+      "step": 12552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6536197003955643e-06,
+      "loss": 0.9141,
+      "step": 12553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6521696551995644e-06,
+      "loss": 0.8672,
+      "step": 12554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650719833520472e-06,
+      "loss": 0.8281,
+      "step": 12555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.649270235409341e-06,
+      "loss": 0.9219,
+      "step": 12556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6478208609172074e-06,
+      "loss": 0.9102,
+      "step": 12557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646371710095108e-06,
+      "loss": 1.0156,
+      "step": 12558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6449227829940714e-06,
+      "loss": 0.9688,
+      "step": 12559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.643474079665119e-06,
+      "loss": 0.8555,
+      "step": 12560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.642025600159256e-06,
+      "loss": 0.9062,
+      "step": 12561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6405773445274884e-06,
+      "loss": 0.9141,
+      "step": 12562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.639129312820814e-06,
+      "loss": 0.8047,
+      "step": 12563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6376815050902146e-06,
+      "loss": 0.8828,
+      "step": 12564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636233921386673e-06,
+      "loss": 0.8359,
+      "step": 12565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63478656176116e-06,
+      "loss": 0.8125,
+      "step": 12566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6333394262646437e-06,
+      "loss": 0.8242,
+      "step": 12567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6318925149480735e-06,
+      "loss": 0.8516,
+      "step": 12568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630445827862399e-06,
+      "loss": 0.8008,
+      "step": 12569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.628999365058562e-06,
+      "loss": 0.9922,
+      "step": 12570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6275531265874976e-06,
+      "loss": 0.8438,
+      "step": 12571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6261071125001237e-06,
+      "loss": 0.8672,
+      "step": 12572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6246613228473594e-06,
+      "loss": 0.8555,
+      "step": 12573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.623215757680113e-06,
+      "loss": 0.957,
+      "step": 12574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6217704170492896e-06,
+      "loss": 0.8633,
+      "step": 12575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.620325301005774e-06,
+      "loss": 0.8555,
+      "step": 12576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6188804096004546e-06,
+      "loss": 0.8906,
+      "step": 12577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6174357428842123e-06,
+      "loss": 0.8594,
+      "step": 12578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6159913009079083e-06,
+      "loss": 0.9727,
+      "step": 12579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6145470837224074e-06,
+      "loss": 0.8398,
+      "step": 12580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6131030913785624e-06,
+      "loss": 0.75,
+      "step": 12581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6116593239272237e-06,
+      "loss": 0.8438,
+      "step": 12582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6102157814192196e-06,
+      "loss": 0.793,
+      "step": 12583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6087724639053836e-06,
+      "loss": 0.8633,
+      "step": 12584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6073293714365363e-06,
+      "loss": 0.9141,
+      "step": 12585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6058865040634972e-06,
+      "loss": 0.8594,
+      "step": 12586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6044438618370624e-06,
+      "loss": 0.8945,
+      "step": 12587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.603001444808034e-06,
+      "loss": 0.8398,
+      "step": 12588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6015592530272015e-06,
+      "loss": 0.875,
+      "step": 12589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6001172865453506e-06,
+      "loss": 0.8711,
+      "step": 12590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5986755454132484e-06,
+      "loss": 0.832,
+      "step": 12591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.597234029681663e-06,
+      "loss": 0.7773,
+      "step": 12592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595792739401354e-06,
+      "loss": 0.9297,
+      "step": 12593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5943516746230733e-06,
+      "loss": 0.957,
+      "step": 12594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5929108353975574e-06,
+      "loss": 0.8203,
+      "step": 12595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591470221775544e-06,
+      "loss": 0.8164,
+      "step": 12596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590029833807761e-06,
+      "loss": 0.9453,
+      "step": 12597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588589671544922e-06,
+      "loss": 0.9219,
+      "step": 12598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5871497350377404e-06,
+      "loss": 0.8828,
+      "step": 12599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.585710024336917e-06,
+      "loss": 0.7305,
+      "step": 12600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.584270539493151e-06,
+      "loss": 0.8047,
+      "step": 12601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5828312805571217e-06,
+      "loss": 0.8008,
+      "step": 12602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5813922475795116e-06,
+      "loss": 0.8281,
+      "step": 12603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5799534406109904e-06,
+      "loss": 0.8984,
+      "step": 12604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5785148597022245e-06,
+      "loss": 0.7852,
+      "step": 12605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5770765049038624e-06,
+      "loss": 0.7812,
+      "step": 12606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5756383762665524e-06,
+      "loss": 0.8203,
+      "step": 12607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574200473840935e-06,
+      "loss": 0.9727,
+      "step": 12608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5727627976776437e-06,
+      "loss": 0.7891,
+      "step": 12609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5713253478272947e-06,
+      "loss": 0.8242,
+      "step": 12610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.569888124340506e-06,
+      "loss": 0.8281,
+      "step": 12611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5684511272678836e-06,
+      "loss": 0.8945,
+      "step": 12612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5670143566600312e-06,
+      "loss": 0.9062,
+      "step": 12613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5655778125675334e-06,
+      "loss": 0.7773,
+      "step": 12614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5641414950409745e-06,
+      "loss": 0.8008,
+      "step": 12615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5627054041309305e-06,
+      "loss": 0.8125,
+      "step": 12616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.561269539887973e-06,
+      "loss": 0.8203,
+      "step": 12617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5598339023626514e-06,
+      "loss": 0.9258,
+      "step": 12618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5583984916055225e-06,
+      "loss": 0.9297,
+      "step": 12619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5569633076671282e-06,
+      "loss": 0.9492,
+      "step": 12620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5555283505980063e-06,
+      "loss": 0.8203,
+      "step": 12621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.55409362044868e-06,
+      "loss": 0.9609,
+      "step": 12622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5526591172696687e-06,
+      "loss": 0.8398,
+      "step": 12623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5512248411114846e-06,
+      "loss": 0.9102,
+      "step": 12624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.549790792024634e-06,
+      "loss": 0.8281,
+      "step": 12625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5483569700596055e-06,
+      "loss": 0.832,
+      "step": 12626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5469233752668896e-06,
+      "loss": 0.9141,
+      "step": 12627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5454900076969654e-06,
+      "loss": 0.8672,
+      "step": 12628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.544056867400306e-06,
+      "loss": 0.7812,
+      "step": 12629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.54262395442737e-06,
+      "loss": 0.9375,
+      "step": 12630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5411912688286144e-06,
+      "loss": 0.9844,
+      "step": 12631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5397588106544865e-06,
+      "loss": 0.8672,
+      "step": 12632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.538326579955429e-06,
+      "loss": 0.8594,
+      "step": 12633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5368945767818662e-06,
+      "loss": 0.9844,
+      "step": 12634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.535462801184224e-06,
+      "loss": 0.8672,
+      "step": 12635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534031253212917e-06,
+      "loss": 0.9297,
+      "step": 12636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.532599932918358e-06,
+      "loss": 0.9023,
+      "step": 12637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531168840350937e-06,
+      "loss": 0.7812,
+      "step": 12638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5297379755610484e-06,
+      "loss": 0.7734,
+      "step": 12639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5283073385990762e-06,
+      "loss": 0.8867,
+      "step": 12640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.526876929515398e-06,
+      "loss": 0.8086,
+      "step": 12641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5254467483603747e-06,
+      "loss": 0.9414,
+      "step": 12642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5240167951843683e-06,
+      "loss": 0.9258,
+      "step": 12643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5225870700377284e-06,
+      "loss": 0.8828,
+      "step": 12644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5211575729708036e-06,
+      "loss": 0.9219,
+      "step": 12645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.519728304033919e-06,
+      "loss": 0.8867,
+      "step": 12646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.518299263277408e-06,
+      "loss": 0.8906,
+      "step": 12647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5168704507515862e-06,
+      "loss": 0.8047,
+      "step": 12648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5154418665067704e-06,
+      "loss": 0.8828,
+      "step": 12649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5140135105932536e-06,
+      "loss": 0.9023,
+      "step": 12650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5125853830613367e-06,
+      "loss": 0.7266,
+      "step": 12651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511157483961305e-06,
+      "loss": 0.8516,
+      "step": 12652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5097298133434398e-06,
+      "loss": 0.8594,
+      "step": 12653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5083023712580056e-06,
+      "loss": 0.875,
+      "step": 12654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5068751577552686e-06,
+      "loss": 0.8945,
+      "step": 12655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5054481728854863e-06,
+      "loss": 0.9883,
+      "step": 12656
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5040214166988975e-06,
+      "loss": 0.8438,
+      "step": 12657
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502594889245744e-06,
+      "loss": 0.7031,
+      "step": 12658
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5011685905762562e-06,
+      "loss": 0.7969,
+      "step": 12659
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49974252074066e-06,
+      "loss": 0.8945,
+      "step": 12660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4983166797891633e-06,
+      "loss": 0.8867,
+      "step": 12661
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4968910677719746e-06,
+      "loss": 0.8633,
+      "step": 12662
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.495465684739292e-06,
+      "loss": 0.8516,
+      "step": 12663
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4940405307413095e-06,
+      "loss": 0.8672,
+      "step": 12664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4926156058282013e-06,
+      "loss": 0.8477,
+      "step": 12665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.491190910050145e-06,
+      "loss": 0.9023,
+      "step": 12666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4897664434573063e-06,
+      "loss": 0.8789,
+      "step": 12667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4883422060998463e-06,
+      "loss": 0.8281,
+      "step": 12668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.486918198027908e-06,
+      "loss": 0.8281,
+      "step": 12669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.485494419291636e-06,
+      "loss": 0.9336,
+      "step": 12670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484070869941164e-06,
+      "loss": 1.0547,
+      "step": 12671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.48264755002662e-06,
+      "loss": 0.75,
+      "step": 12672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481224459598116e-06,
+      "loss": 0.8945,
+      "step": 12673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4798015987057633e-06,
+      "loss": 0.8477,
+      "step": 12674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4783789673996637e-06,
+      "loss": 0.8789,
+      "step": 12675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.476956565729913e-06,
+      "loss": 0.8555,
+      "step": 12676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4755343937465902e-06,
+      "loss": 0.8477,
+      "step": 12677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4741124514997747e-06,
+      "loss": 0.8281,
+      "step": 12678
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472690739039536e-06,
+      "loss": 0.7695,
+      "step": 12679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4712692564159376e-06,
+      "loss": 0.7891,
+      "step": 12680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469848003679026e-06,
+      "loss": 0.8633,
+      "step": 12681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4684269808788485e-06,
+      "loss": 0.9258,
+      "step": 12682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4670061880654426e-06,
+      "loss": 0.875,
+      "step": 12683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4655856252888376e-06,
+      "loss": 0.8633,
+      "step": 12684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4641652925990498e-06,
+      "loss": 0.8555,
+      "step": 12685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4627451900460927e-06,
+      "loss": 0.9531,
+      "step": 12686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.461325317679971e-06,
+      "loss": 0.7578,
+      "step": 12687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4599056755506845e-06,
+      "loss": 0.8477,
+      "step": 12688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458486263708213e-06,
+      "loss": 0.8555,
+      "step": 12689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4570670822025397e-06,
+      "loss": 0.8555,
+      "step": 12690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4556481310836375e-06,
+      "loss": 0.8711,
+      "step": 12691
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.454229410401473e-06,
+      "loss": 0.7734,
+      "step": 12692
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4528109202059933e-06,
+      "loss": 0.8008,
+      "step": 12693
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.7695,
+      "step": 12694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4499746314748827e-06,
+      "loss": 0.8711,
+      "step": 12695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448556833039124e-06,
+      "loss": 0.7969,
+      "step": 12696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.447139265289793e-06,
+      "loss": 0.7852,
+      "step": 12697
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.445721928276805e-06,
+      "loss": 0.8203,
+      "step": 12698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4443048220500674e-06,
+      "loss": 1.0156,
+      "step": 12699
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4428879466594835e-06,
+      "loss": 0.9648,
+      "step": 12700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.441471302154934e-06,
+      "loss": 0.8398,
+      "step": 12701
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440054888586307e-06,
+      "loss": 0.8047,
+      "step": 12702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4386387060034767e-06,
+      "loss": 0.8281,
+      "step": 12703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437222754456312e-06,
+      "loss": 0.8789,
+      "step": 12704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4358070339946625e-06,
+      "loss": 0.9609,
+      "step": 12705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.434391544668383e-06,
+      "loss": 0.9609,
+      "step": 12706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432976286527314e-06,
+      "loss": 0.7656,
+      "step": 12707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4315612596212932e-06,
+      "loss": 0.8516,
+      "step": 12708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.430146464000138e-06,
+      "loss": 0.7812,
+      "step": 12709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42873189971367e-06,
+      "loss": 0.7578,
+      "step": 12710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427317566811702e-06,
+      "loss": 0.8984,
+      "step": 12711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.425903465344026e-06,
+      "loss": 0.8047,
+      "step": 12712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4244895953604405e-06,
+      "loss": 0.7539,
+      "step": 12713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4230759569107285e-06,
+      "loss": 0.7812,
+      "step": 12714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4216625500446696e-06,
+      "loss": 0.9102,
+      "step": 12715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.420249374812027e-06,
+      "loss": 0.8438,
+      "step": 12716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4188364312625623e-06,
+      "loss": 0.7891,
+      "step": 12717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.417423719446028e-06,
+      "loss": 0.8477,
+      "step": 12718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4160112394121715e-06,
+      "loss": 0.9805,
+      "step": 12719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.414598991210721e-06,
+      "loss": 0.8789,
+      "step": 12720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413186974891409e-06,
+      "loss": 0.7812,
+      "step": 12721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411775190503953e-06,
+      "loss": 1.0156,
+      "step": 12722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4103636380980686e-06,
+      "loss": 0.7969,
+      "step": 12723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4089523177234517e-06,
+      "loss": 0.7148,
+      "step": 12724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4075412294297993e-06,
+      "loss": 0.7852,
+      "step": 12725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4061303732667995e-06,
+      "loss": 0.9102,
+      "step": 12726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404719749284133e-06,
+      "loss": 0.7969,
+      "step": 12727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4033093575314645e-06,
+      "loss": 0.875,
+      "step": 12728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4018991980584582e-06,
+      "loss": 0.8359,
+      "step": 12729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4004892709147695e-06,
+      "loss": 0.7969,
+      "step": 12730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.399079576150046e-06,
+      "loss": 0.8516,
+      "step": 12731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397670113813919e-06,
+      "loss": 0.793,
+      "step": 12732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3962608839560208e-06,
+      "loss": 0.8906,
+      "step": 12733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.394851886625974e-06,
+      "loss": 0.8203,
+      "step": 12734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393443121873393e-06,
+      "loss": 0.7969,
+      "step": 12735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3920345897478766e-06,
+      "loss": 0.8672,
+      "step": 12736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.390626290299026e-06,
+      "loss": 1.0078,
+      "step": 12737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.389218223576428e-06,
+      "loss": 0.9414,
+      "step": 12738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3878103896296677e-06,
+      "loss": 0.9062,
+      "step": 12739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3864027885083085e-06,
+      "loss": 0.9531,
+      "step": 12740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3849954202619184e-06,
+      "loss": 0.9062,
+      "step": 12741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3835882849400536e-06,
+      "loss": 1.1484,
+      "step": 12742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3821813825922632e-06,
+      "loss": 0.8711,
+      "step": 12743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3807747132680825e-06,
+      "loss": 0.8906,
+      "step": 12744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3793682770170434e-06,
+      "loss": 0.9062,
+      "step": 12745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3779620738886698e-06,
+      "loss": 0.7812,
+      "step": 12746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.376556103932479e-06,
+      "loss": 0.8438,
+      "step": 12747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3751503671979715e-06,
+      "loss": 0.9336,
+      "step": 12748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.373744863734647e-06,
+      "loss": 0.9258,
+      "step": 12749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.372339593591998e-06,
+      "loss": 0.8281,
+      "step": 12750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.370934556819507e-06,
+      "loss": 0.8906,
+      "step": 12751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.369529753466644e-06,
+      "loss": 0.8359,
+      "step": 12752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3681251835828743e-06,
+      "loss": 0.9883,
+      "step": 12753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.366720847217657e-06,
+      "loss": 0.8945,
+      "step": 12754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3653167444204447e-06,
+      "loss": 0.8594,
+      "step": 12755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.363912875240669e-06,
+      "loss": 0.8359,
+      "step": 12756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362509239727768e-06,
+      "loss": 0.8633,
+      "step": 12757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3611058379311647e-06,
+      "loss": 0.9531,
+      "step": 12758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359702669900278e-06,
+      "loss": 0.8945,
+      "step": 12759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3582997356845103e-06,
+      "loss": 0.8281,
+      "step": 12760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.356897035333263e-06,
+      "loss": 0.832,
+      "step": 12761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3554945688959284e-06,
+      "loss": 0.7812,
+      "step": 12762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3540923364218926e-06,
+      "loss": 0.8828,
+      "step": 12763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3526903379605236e-06,
+      "loss": 0.9219,
+      "step": 12764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.351288573561191e-06,
+      "loss": 0.8516,
+      "step": 12765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3498870432732566e-06,
+      "loss": 0.8164,
+      "step": 12766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.348485747146064e-06,
+      "loss": 0.8164,
+      "step": 12767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3470846852289584e-06,
+      "loss": 0.9062,
+      "step": 12768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3456838575712723e-06,
+      "loss": 0.8203,
+      "step": 12769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3442832642223353e-06,
+      "loss": 0.8789,
+      "step": 12770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.342882905231458e-06,
+      "loss": 0.832,
+      "step": 12771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.341482780647952e-06,
+      "loss": 0.7227,
+      "step": 12772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340082890521118e-06,
+      "loss": 0.7734,
+      "step": 12773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.338683234900252e-06,
+      "loss": 0.7539,
+      "step": 12774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.337283813834631e-06,
+      "loss": 0.8672,
+      "step": 12775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3358846273735334e-06,
+      "loss": 0.8047,
+      "step": 12776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3344856755662282e-06,
+      "loss": 0.8867,
+      "step": 12777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3330869584619775e-06,
+      "loss": 0.9219,
+      "step": 12778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3316884761100244e-06,
+      "loss": 0.8125,
+      "step": 12779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3302902285596163e-06,
+      "loss": 0.9062,
+      "step": 12780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328892215859987e-06,
+      "loss": 0.957,
+      "step": 12781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.327494438060367e-06,
+      "loss": 0.8242,
+      "step": 12782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.326096895209966e-06,
+      "loss": 0.7969,
+      "step": 12783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324699587357998e-06,
+      "loss": 0.9531,
+      "step": 12784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3233025145536644e-06,
+      "loss": 0.7812,
+      "step": 12785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.321905676846161e-06,
+      "loss": 0.918,
+      "step": 12786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.320509074284667e-06,
+      "loss": 0.8945,
+      "step": 12787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3191127069183614e-06,
+      "loss": 0.9375,
+      "step": 12788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3177165747964124e-06,
+      "loss": 0.8477,
+      "step": 12789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3163206779679836e-06,
+      "loss": 0.8047,
+      "step": 12790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.31492501648222e-06,
+      "loss": 0.8438,
+      "step": 12791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.313529590388268e-06,
+      "loss": 0.8828,
+      "step": 12792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.312134399735264e-06,
+      "loss": 0.8398,
+      "step": 12793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310739444572336e-06,
+      "loss": 0.9492,
+      "step": 12794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3093447249485975e-06,
+      "loss": 0.9062,
+      "step": 12795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3079502409131615e-06,
+      "loss": 0.7461,
+      "step": 12796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30655599251513e-06,
+      "loss": 0.9922,
+      "step": 12797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3051619798036004e-06,
+      "loss": 0.8555,
+      "step": 12798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.303768202827651e-06,
+      "loss": 0.8984,
+      "step": 12799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3023746616363617e-06,
+      "loss": 0.9102,
+      "step": 12800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3009813562788017e-06,
+      "loss": 0.7695,
+      "step": 12801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2995882868040352e-06,
+      "loss": 0.8867,
+      "step": 12802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.298195453261106e-06,
+      "loss": 0.8242,
+      "step": 12803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2968028556990637e-06,
+      "loss": 0.8945,
+      "step": 12804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2954104941669417e-06,
+      "loss": 0.8594,
+      "step": 12805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2940183687137717e-06,
+      "loss": 0.9297,
+      "step": 12806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2926264793885667e-06,
+      "loss": 0.8594,
+      "step": 12807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2912348262403382e-06,
+      "loss": 0.8945,
+      "step": 12808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2898434093180897e-06,
+      "loss": 0.8555,
+      "step": 12809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2884522286708186e-06,
+      "loss": 0.8359,
+      "step": 12810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2870612843475046e-06,
+      "loss": 0.8711,
+      "step": 12811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2856705763971273e-06,
+      "loss": 0.9062,
+      "step": 12812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.284280104868656e-06,
+      "loss": 0.8359,
+      "step": 12813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2828898698110535e-06,
+      "loss": 0.8281,
+      "step": 12814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.281499871273267e-06,
+      "loss": 0.8477,
+      "step": 12815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2801101093042444e-06,
+      "loss": 0.832,
+      "step": 12816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278720583952919e-06,
+      "loss": 0.7578,
+      "step": 12817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2773312952682236e-06,
+      "loss": 0.8633,
+      "step": 12818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2759422432990694e-06,
+      "loss": 0.8359,
+      "step": 12819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2745534280943713e-06,
+      "loss": 0.8789,
+      "step": 12820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.273164849703031e-06,
+      "loss": 0.8711,
+      "step": 12821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.271776508173945e-06,
+      "loss": 0.7539,
+      "step": 12822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.270388403555994e-06,
+      "loss": 0.9336,
+      "step": 12823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269000535898057e-06,
+      "loss": 0.7344,
+      "step": 12824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267612905249008e-06,
+      "loss": 0.9062,
+      "step": 12825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.266225511657699e-06,
+      "loss": 0.7773,
+      "step": 12826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264838355172988e-06,
+      "loss": 0.8984,
+      "step": 12827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2634514358437163e-06,
+      "loss": 0.9531,
+      "step": 12828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2620647537187244e-06,
+      "loss": 0.8477,
+      "step": 12829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.260678308846833e-06,
+      "loss": 0.8945,
+      "step": 12830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259292101276864e-06,
+      "loss": 0.957,
+      "step": 12831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2579061310576275e-06,
+      "loss": 0.8477,
+      "step": 12832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.256520398237929e-06,
+      "loss": 0.75,
+      "step": 12833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255134902866556e-06,
+      "loss": 0.7852,
+      "step": 12834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253749644992297e-06,
+      "loss": 0.8359,
+      "step": 12835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2523646246639294e-06,
+      "loss": 0.7969,
+      "step": 12836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250979841930224e-06,
+      "loss": 0.8008,
+      "step": 12837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.249595296839937e-06,
+      "loss": 0.9453,
+      "step": 12838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2482109894418213e-06,
+      "loss": 0.8281,
+      "step": 12839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2468269197846213e-06,
+      "loss": 0.9141,
+      "step": 12840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2454430879170762e-06,
+      "loss": 0.7969,
+      "step": 12841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2440594938879056e-06,
+      "loss": 0.8789,
+      "step": 12842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2426761377458316e-06,
+      "loss": 0.8633,
+      "step": 12843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241293019539563e-06,
+      "loss": 0.9727,
+      "step": 12844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.239910139317807e-06,
+      "loss": 0.8828,
+      "step": 12845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.238527497129249e-06,
+      "loss": 0.8438,
+      "step": 12846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2371450930225767e-06,
+      "loss": 0.9023,
+      "step": 12847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2357629270464685e-06,
+      "loss": 0.832,
+      "step": 12848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2343809992495945e-06,
+      "loss": 0.9883,
+      "step": 12849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232999309680609e-06,
+      "loss": 0.8008,
+      "step": 12850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.231617858388165e-06,
+      "loss": 0.8125,
+      "step": 12851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2302366454209066e-06,
+      "loss": 0.8047,
+      "step": 12852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228855670827472e-06,
+      "loss": 0.8711,
+      "step": 12853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2274749346564805e-06,
+      "loss": 0.9141,
+      "step": 12854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.226094436956554e-06,
+      "loss": 0.8359,
+      "step": 12855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2247141777763e-06,
+      "loss": 0.8945,
+      "step": 12856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223334157164325e-06,
+      "loss": 0.7344,
+      "step": 12857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.221954375169214e-06,
+      "loss": 0.8555,
+      "step": 12858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2205748318395545e-06,
+      "loss": 0.7305,
+      "step": 12859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.219195527223923e-06,
+      "loss": 0.8633,
+      "step": 12860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2178164613708897e-06,
+      "loss": 0.8594,
+      "step": 12861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2164376343290073e-06,
+      "loss": 0.9219,
+      "step": 12862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.215059046146829e-06,
+      "loss": 0.7852,
+      "step": 12863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213680696872897e-06,
+      "loss": 0.8281,
+      "step": 12864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.21230258655575e-06,
+      "loss": 0.8438,
+      "step": 12865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.210924715243906e-06,
+      "loss": 0.8828,
+      "step": 12866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2095470829858856e-06,
+      "loss": 0.9023,
+      "step": 12867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2081696898301983e-06,
+      "loss": 0.8672,
+      "step": 12868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2067925358253414e-06,
+      "loss": 0.7656,
+      "step": 12869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2054156210198075e-06,
+      "loss": 0.8477,
+      "step": 12870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.204038945462079e-06,
+      "loss": 0.7812,
+      "step": 12871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.202662509200637e-06,
+      "loss": 0.7734,
+      "step": 12872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2012863122839386e-06,
+      "loss": 0.8828,
+      "step": 12873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1999103547604473e-06,
+      "loss": 0.7812,
+      "step": 12874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1985346366786117e-06,
+      "loss": 0.8398,
+      "step": 12875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1971591580868754e-06,
+      "loss": 0.7773,
+      "step": 12876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.195783919033667e-06,
+      "loss": 0.8555,
+      "step": 12877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.194408919567411e-06,
+      "loss": 0.8086,
+      "step": 12878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1930341597365253e-06,
+      "loss": 0.8359,
+      "step": 12879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1916596395894207e-06,
+      "loss": 0.8633,
+      "step": 12880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1902853591744876e-06,
+      "loss": 0.8516,
+      "step": 12881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1889113185401223e-06,
+      "loss": 0.8555,
+      "step": 12882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.187537517734708e-06,
+      "loss": 0.9297,
+      "step": 12883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.186163956806614e-06,
+      "loss": 0.8594,
+      "step": 12884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.184790635804207e-06,
+      "loss": 0.7305,
+      "step": 12885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.183417554775844e-06,
+      "loss": 0.8984,
+      "step": 12886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1820447137698772e-06,
+      "loss": 0.8789,
+      "step": 12887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.180672112834641e-06,
+      "loss": 0.8516,
+      "step": 12888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1792997520184663e-06,
+      "loss": 0.8359,
+      "step": 12889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1779276313696804e-06,
+      "loss": 0.8008,
+      "step": 12890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1765557509365986e-06,
+      "loss": 0.8086,
+      "step": 12891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1751841107675205e-06,
+      "loss": 0.8711,
+      "step": 12892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1738127109107473e-06,
+      "loss": 0.9102,
+      "step": 12893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.172441551414569e-06,
+      "loss": 0.8867,
+      "step": 12894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171070632327268e-06,
+      "loss": 0.8672,
+      "step": 12895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1696999536971107e-06,
+      "loss": 0.875,
+      "step": 12896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1683295155723637e-06,
+      "loss": 0.7617,
+      "step": 12897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1669593180012857e-06,
+      "loss": 0.875,
+      "step": 12898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1655893610321174e-06,
+      "loss": 0.8047,
+      "step": 12899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1642196447131003e-06,
+      "loss": 0.7695,
+      "step": 12900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.162850169092464e-06,
+      "loss": 0.9648,
+      "step": 12901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1614809342184327e-06,
+      "loss": 0.832,
+      "step": 12902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1601119401392134e-06,
+      "loss": 0.8984,
+      "step": 12903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1587431869030126e-06,
+      "loss": 0.8516,
+      "step": 12904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.157374674558028e-06,
+      "loss": 0.793,
+      "step": 12905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1560064031524485e-06,
+      "loss": 0.9102,
+      "step": 12906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1546383727344476e-06,
+      "loss": 0.8867,
+      "step": 12907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1532705833521993e-06,
+      "loss": 0.9258,
+      "step": 12908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1519030350538683e-06,
+      "loss": 0.7812,
+      "step": 12909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1505357278876014e-06,
+      "loss": 0.8789,
+      "step": 12910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149168661901547e-06,
+      "loss": 0.8516,
+      "step": 12911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1478018371438422e-06,
+      "loss": 0.8672,
+      "step": 12912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1464352536626173e-06,
+      "loss": 0.8203,
+      "step": 12913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.145068911505985e-06,
+      "loss": 0.8359,
+      "step": 12914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143702810722061e-06,
+      "loss": 0.7578,
+      "step": 12915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1423369513589474e-06,
+      "loss": 0.8633,
+      "step": 12916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140971333464741e-06,
+      "loss": 0.7578,
+      "step": 12917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1396059570875217e-06,
+      "loss": 0.8828,
+      "step": 12918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.138240822275368e-06,
+      "loss": 0.8594,
+      "step": 12919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1368759290763497e-06,
+      "loss": 0.7617,
+      "step": 12920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.13551127753853e-06,
+      "loss": 0.8594,
+      "step": 12921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.134146867709954e-06,
+      "loss": 0.7578,
+      "step": 12922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1327826996386666e-06,
+      "loss": 0.8164,
+      "step": 12923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1314187733727076e-06,
+      "loss": 0.8906,
+      "step": 12924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1300550889600945e-06,
+      "loss": 0.918,
+      "step": 12925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1286916464488506e-06,
+      "loss": 0.8789,
+      "step": 12926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1273284458869814e-06,
+      "loss": 0.832,
+      "step": 12927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.125965487322492e-06,
+      "loss": 0.9258,
+      "step": 12928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.12460277080337e-06,
+      "loss": 0.7891,
+      "step": 12929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.123240296377599e-06,
+      "loss": 0.8047,
+      "step": 12930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1218780640931535e-06,
+      "loss": 0.793,
+      "step": 12931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1205160739980058e-06,
+      "loss": 0.7969,
+      "step": 12932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119154326140106e-06,
+      "loss": 0.8984,
+      "step": 12933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1177928205674056e-06,
+      "loss": 0.8008,
+      "step": 12934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.116431557327847e-06,
+      "loss": 0.8984,
+      "step": 12935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1150705364693633e-06,
+      "loss": 0.9336,
+      "step": 12936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1137097580398745e-06,
+      "loss": 0.7734,
+      "step": 12937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1123492220872964e-06,
+      "loss": 0.7422,
+      "step": 12938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1109889286595405e-06,
+      "loss": 0.8516,
+      "step": 12939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1096288778044982e-06,
+      "loss": 0.7617,
+      "step": 12940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.108269069570061e-06,
+      "loss": 0.8516,
+      "step": 12941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1069095040041107e-06,
+      "loss": 0.8398,
+      "step": 12942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105550181154524e-06,
+      "loss": 0.9297,
+      "step": 12943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.104191101069156e-06,
+      "loss": 0.918,
+      "step": 12944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1028322637958675e-06,
+      "loss": 0.8906,
+      "step": 12945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.101473669382503e-06,
+      "loss": 0.8203,
+      "step": 12946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1001153178769063e-06,
+      "loss": 0.7578,
+      "step": 12947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.098757209326898e-06,
+      "loss": 1.0312,
+      "step": 12948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097399343780305e-06,
+      "loss": 0.8438,
+      "step": 12949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096041721284938e-06,
+      "loss": 0.8789,
+      "step": 12950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0946843418886062e-06,
+      "loss": 0.8047,
+      "step": 12951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.093327205639096e-06,
+      "loss": 0.9766,
+      "step": 12952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091970312584199e-06,
+      "loss": 0.7969,
+      "step": 12953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.090613662771693e-06,
+      "loss": 0.8281,
+      "step": 12954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089257256249352e-06,
+      "loss": 0.8281,
+      "step": 12955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.08790109306493e-06,
+      "loss": 0.8906,
+      "step": 12956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086545173266182e-06,
+      "loss": 0.6992,
+      "step": 12957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085189496900853e-06,
+      "loss": 0.7734,
+      "step": 12958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083834064016682e-06,
+      "loss": 0.8711,
+      "step": 12959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.082478874661388e-06,
+      "loss": 0.9141,
+      "step": 12960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0811239288826943e-06,
+      "loss": 0.8867,
+      "step": 12961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0797692267283095e-06,
+      "loss": 0.9766,
+      "step": 12962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0784147682459375e-06,
+      "loss": 0.8281,
+      "step": 12963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0770605534832663e-06,
+      "loss": 0.9023,
+      "step": 12964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0757065824879816e-06,
+      "loss": 0.7461,
+      "step": 12965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0743528553077596e-06,
+      "loss": 0.7539,
+      "step": 12966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0729993719902696e-06,
+      "loss": 0.7617,
+      "step": 12967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0716461325831636e-06,
+      "loss": 0.9141,
+      "step": 12968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.070293137134094e-06,
+      "loss": 1.0391,
+      "step": 12969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.068940385690704e-06,
+      "loss": 1.0078,
+      "step": 12970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0675878783006263e-06,
+      "loss": 0.8477,
+      "step": 12971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0662356150114804e-06,
+      "loss": 1.0156,
+      "step": 12972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0648835958708845e-06,
+      "loss": 0.7969,
+      "step": 12973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0635318209264455e-06,
+      "loss": 0.8984,
+      "step": 12974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0621802902257626e-06,
+      "loss": 0.8359,
+      "step": 12975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060829003816421e-06,
+      "loss": 0.8789,
+      "step": 12976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0594779617460045e-06,
+      "loss": 0.8594,
+      "step": 12977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0581271640620857e-06,
+      "loss": 0.9062,
+      "step": 12978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.05677661081223e-06,
+      "loss": 0.9727,
+      "step": 12979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055426302043988e-06,
+      "loss": 0.8828,
+      "step": 12980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0540762378049072e-06,
+      "loss": 0.918,
+      "step": 12981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0527264181425263e-06,
+      "loss": 0.8555,
+      "step": 12982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0513768431043787e-06,
+      "loss": 0.8398,
+      "step": 12983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0500275127379773e-06,
+      "loss": 0.8867,
+      "step": 12984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0486784270908387e-06,
+      "loss": 0.8164,
+      "step": 12985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0473295862104657e-06,
+      "loss": 0.7969,
+      "step": 12986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0459809901443547e-06,
+      "loss": 0.7812,
+      "step": 12987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.044632638939988e-06,
+      "loss": 0.875,
+      "step": 12988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0432845326448444e-06,
+      "loss": 0.7148,
+      "step": 12989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041936671306395e-06,
+      "loss": 0.8477,
+      "step": 12990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0405890549720995e-06,
+      "loss": 0.8984,
+      "step": 12991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0392416836894066e-06,
+      "loss": 0.8477,
+      "step": 12992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0378945575057615e-06,
+      "loss": 0.8398,
+      "step": 12993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.036547676468602e-06,
+      "loss": 0.8008,
+      "step": 12994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035201040625347e-06,
+      "loss": 0.7617,
+      "step": 12995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0338546500234165e-06,
+      "loss": 0.8711,
+      "step": 12996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032508504710221e-06,
+      "loss": 0.7578,
+      "step": 12997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.03116260473316e-06,
+      "loss": 0.9414,
+      "step": 12998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.029816950139621e-06,
+      "loss": 0.8789,
+      "step": 12999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0284715409769894e-06,
+      "loss": 0.8203,
+      "step": 13000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0271263772926398e-06,
+      "loss": 0.8711,
+      "step": 13001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0257814591339384e-06,
+      "loss": 0.7969,
+      "step": 13002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024436786548237e-06,
+      "loss": 0.9961,
+      "step": 13003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0230923595828864e-06,
+      "loss": 0.8242,
+      "step": 13004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0217481782852264e-06,
+      "loss": 0.9844,
+      "step": 13005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0204042427025904e-06,
+      "loss": 0.8203,
+      "step": 13006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0190605528822946e-06,
+      "loss": 0.793,
+      "step": 13007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0177171088716563e-06,
+      "loss": 0.7773,
+      "step": 13008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.016373910717978e-06,
+      "loss": 0.7812,
+      "step": 13009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0150309584685598e-06,
+      "loss": 0.9062,
+      "step": 13010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0136882521706833e-06,
+      "loss": 0.8633,
+      "step": 13011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0123457918716313e-06,
+      "loss": 0.8789,
+      "step": 13012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011003577618672e-06,
+      "loss": 0.7734,
+      "step": 13013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0096616094590712e-06,
+      "loss": 0.8438,
+      "step": 13014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.008319887440074e-06,
+      "loss": 0.9453,
+      "step": 13015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0069784116089306e-06,
+      "loss": 0.7812,
+      "step": 13016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005637182012873e-06,
+      "loss": 0.8477,
+      "step": 13017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0042961986991335e-06,
+      "loss": 0.8867,
+      "step": 13018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0029554617149225e-06,
+      "loss": 0.8516,
+      "step": 13019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0016149711074537e-06,
+      "loss": 0.957,
+      "step": 13020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0002747269239265e-06,
+      "loss": 0.7695,
+      "step": 13021
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9989347292115366e-06,
+      "loss": 0.7344,
+      "step": 13022
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997594978017463e-06,
+      "loss": 0.7734,
+      "step": 13023
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9962554733888806e-06,
+      "loss": 0.8984,
+      "step": 13024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.994916215372957e-06,
+      "loss": 1.0156,
+      "step": 13025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993577204016852e-06,
+      "loss": 0.8945,
+      "step": 13026
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9922384393677094e-06,
+      "loss": 0.8867,
+      "step": 13027
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9908999214726707e-06,
+      "loss": 0.7969,
+      "step": 13028
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989561650378867e-06,
+      "loss": 0.9297,
+      "step": 13029
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9882236261334263e-06,
+      "loss": 0.9102,
+      "step": 13030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.986885848783454e-06,
+      "loss": 0.7734,
+      "step": 13031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.985548318376059e-06,
+      "loss": 0.8711,
+      "step": 13032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9842110349583387e-06,
+      "loss": 0.9414,
+      "step": 13033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9828739985773823e-06,
+      "loss": 0.7578,
+      "step": 13034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9815372092802654e-06,
+      "loss": 0.8438,
+      "step": 13035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9802006671140583e-06,
+      "loss": 0.918,
+      "step": 13036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.978864372125825e-06,
+      "loss": 0.8516,
+      "step": 13037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9775283243626196e-06,
+      "loss": 0.7734,
+      "step": 13038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9761925238714826e-06,
+      "loss": 0.9727,
+      "step": 13039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974856970699451e-06,
+      "loss": 0.7852,
+      "step": 13040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9735216648935526e-06,
+      "loss": 0.8828,
+      "step": 13041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9721866065008087e-06,
+      "loss": 0.8633,
+      "step": 13042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9708517955682227e-06,
+      "loss": 1.0,
+      "step": 13043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9695172321427966e-06,
+      "loss": 0.9141,
+      "step": 13044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.968182916271525e-06,
+      "loss": 0.9531,
+      "step": 13045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9668488480013924e-06,
+      "loss": 0.8008,
+      "step": 13046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9655150273793676e-06,
+      "loss": 0.9375,
+      "step": 13047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9641814544524204e-06,
+      "loss": 0.875,
+      "step": 13048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9628481292675106e-06,
+      "loss": 0.8398,
+      "step": 13049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9615150518715795e-06,
+      "loss": 0.75,
+      "step": 13050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.960182222311572e-06,
+      "loss": 0.7969,
+      "step": 13051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958849640634417e-06,
+      "loss": 0.8516,
+      "step": 13052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9575173068870412e-06,
+      "loss": 0.7969,
+      "step": 13053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.956185221116352e-06,
+      "loss": 0.9961,
+      "step": 13054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9548533833692562e-06,
+      "loss": 0.8359,
+      "step": 13055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953521793692651e-06,
+      "loss": 0.8203,
+      "step": 13056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9521904521334254e-06,
+      "loss": 0.8711,
+      "step": 13057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.950859358738453e-06,
+      "loss": 0.8242,
+      "step": 13058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9495285135546057e-06,
+      "loss": 0.832,
+      "step": 13059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9481979166287467e-06,
+      "loss": 0.9453,
+      "step": 13060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.94686756800773e-06,
+      "loss": 0.875,
+      "step": 13061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.945537467738393e-06,
+      "loss": 0.8711,
+      "step": 13062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9442076158675725e-06,
+      "loss": 0.8047,
+      "step": 13063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9428780124420975e-06,
+      "loss": 0.8555,
+      "step": 13064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.941548657508786e-06,
+      "loss": 0.7734,
+      "step": 13065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9402195511144427e-06,
+      "loss": 0.8867,
+      "step": 13066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938890693305868e-06,
+      "loss": 0.9453,
+      "step": 13067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9375620841298545e-06,
+      "loss": 0.8281,
+      "step": 13068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9362337236331884e-06,
+      "loss": 0.7266,
+      "step": 13069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934905611862635e-06,
+      "loss": 0.8789,
+      "step": 13070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933577748864964e-06,
+      "loss": 0.8711,
+      "step": 13071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9322501346869315e-06,
+      "loss": 0.8516,
+      "step": 13072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.930922769375286e-06,
+      "loss": 0.9414,
+      "step": 13073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.929595652976762e-06,
+      "loss": 0.8672,
+      "step": 13074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.928268785538093e-06,
+      "loss": 0.7578,
+      "step": 13075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9269421671059973e-06,
+      "loss": 0.8945,
+      "step": 13076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9256157977271926e-06,
+      "loss": 0.8594,
+      "step": 13077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.924289677448374e-06,
+      "loss": 0.7969,
+      "step": 13078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9229638063162423e-06,
+      "loss": 0.9844,
+      "step": 13079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9216381843774812e-06,
+      "loss": 0.7734,
+      "step": 13080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9203128116787726e-06,
+      "loss": 0.8555,
+      "step": 13081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9189876882667766e-06,
+      "loss": 0.9883,
+      "step": 13082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.917662814188158e-06,
+      "loss": 0.8047,
+      "step": 13083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.916338189489566e-06,
+      "loss": 0.9062,
+      "step": 13084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9150138142176464e-06,
+      "loss": 0.9648,
+      "step": 13085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.913689688419028e-06,
+      "loss": 0.9023,
+      "step": 13086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9123658121403355e-06,
+      "loss": 0.7891,
+      "step": 13087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9110421854281868e-06,
+      "loss": 0.6953,
+      "step": 13088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9097188083291904e-06,
+      "loss": 0.8516,
+      "step": 13089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9083956808899406e-06,
+      "loss": 0.7578,
+      "step": 13090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.907072803157026e-06,
+      "loss": 0.8867,
+      "step": 13091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.905750175177031e-06,
+      "loss": 0.8555,
+      "step": 13092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9044277969965285e-06,
+      "loss": 1.0781,
+      "step": 13093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903105668662076e-06,
+      "loss": 0.8711,
+      "step": 13094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9017837902202304e-06,
+      "loss": 0.8203,
+      "step": 13095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9004621617175367e-06,
+      "loss": 0.8203,
+      "step": 13096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8991407832005356e-06,
+      "loss": 0.9141,
+      "step": 13097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.897819654715748e-06,
+      "loss": 0.8008,
+      "step": 13098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8964987763096964e-06,
+      "loss": 0.8828,
+      "step": 13099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89517814802889e-06,
+      "loss": 0.8477,
+      "step": 13100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8938577699198345e-06,
+      "loss": 0.9883,
+      "step": 13101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8925376420290153e-06,
+      "loss": 0.9102,
+      "step": 13102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.891217764402919e-06,
+      "loss": 0.8594,
+      "step": 13103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8898981370880253e-06,
+      "loss": 0.8906,
+      "step": 13104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8885787601307923e-06,
+      "loss": 0.7852,
+      "step": 13105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8872596335776827e-06,
+      "loss": 0.8203,
+      "step": 13106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.885940757475142e-06,
+      "loss": 0.8789,
+      "step": 13107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8846221318696142e-06,
+      "loss": 0.9336,
+      "step": 13108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.883303756807525e-06,
+      "loss": 0.8984,
+      "step": 13109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.881985632335298e-06,
+      "loss": 0.8477,
+      "step": 13110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8806677584993483e-06,
+      "loss": 0.8125,
+      "step": 13111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8793501353460807e-06,
+      "loss": 0.7852,
+      "step": 13112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8780327629218864e-06,
+      "loss": 0.7656,
+      "step": 13113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8767156412731544e-06,
+      "loss": 0.9023,
+      "step": 13114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.875398770446264e-06,
+      "loss": 0.8555,
+      "step": 13115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874082150487586e-06,
+      "loss": 0.832,
+      "step": 13116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8727657814434752e-06,
+      "loss": 0.9453,
+      "step": 13117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8714496633602838e-06,
+      "loss": 0.8828,
+      "step": 13118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.870133796284358e-06,
+      "loss": 0.9062,
+      "step": 13119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.868818180262033e-06,
+      "loss": 0.832,
+      "step": 13120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8675028153396266e-06,
+      "loss": 0.8398,
+      "step": 13121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8661877015634585e-06,
+      "loss": 0.9102,
+      "step": 13122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8648728389798365e-06,
+      "loss": 0.8125,
+      "step": 13123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8635582276350606e-06,
+      "loss": 0.7734,
+      "step": 13124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.862243867575416e-06,
+      "loss": 0.8164,
+      "step": 13125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.860929758847185e-06,
+      "loss": 0.832,
+      "step": 13126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8596159014966396e-06,
+      "loss": 0.8945,
+      "step": 13127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8583022955700466e-06,
+      "loss": 0.7695,
+      "step": 13128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.856988941113653e-06,
+      "loss": 0.8203,
+      "step": 13129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.855675838173708e-06,
+      "loss": 0.7852,
+      "step": 13130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8543629867964485e-06,
+      "loss": 0.7578,
+      "step": 13131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8530503870281024e-06,
+      "loss": 0.7734,
+      "step": 13132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8517380389148852e-06,
+      "loss": 0.7812,
+      "step": 13133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8504259425030077e-06,
+      "loss": 0.8867,
+      "step": 13134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.849114097838672e-06,
+      "loss": 0.8984,
+      "step": 13135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8478025049680734e-06,
+      "loss": 0.8281,
+      "step": 13136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.846491163937388e-06,
+      "loss": 0.8047,
+      "step": 13137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845180074792794e-06,
+      "loss": 0.9336,
+      "step": 13138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8438692375804557e-06,
+      "loss": 0.8125,
+      "step": 13139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8425586523465344e-06,
+      "loss": 0.8672,
+      "step": 13140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841248319137171e-06,
+      "loss": 0.9414,
+      "step": 13141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8399382379985072e-06,
+      "loss": 0.9492,
+      "step": 13142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8386284089766726e-06,
+      "loss": 0.8086,
+      "step": 13143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8373188321177923e-06,
+      "loss": 0.8203,
+      "step": 13144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.836009507467973e-06,
+      "loss": 0.8828,
+      "step": 13145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.834700435073319e-06,
+      "loss": 0.9141,
+      "step": 13146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8333916149799258e-06,
+      "loss": 0.8555,
+      "step": 13147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8320830472338823e-06,
+      "loss": 0.8516,
+      "step": 13148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8307747318812594e-06,
+      "loss": 0.832,
+      "step": 13149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8294666689681273e-06,
+      "loss": 0.8711,
+      "step": 13150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828158858540545e-06,
+      "loss": 0.8984,
+      "step": 13151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8268513006445653e-06,
+      "loss": 0.7852,
+      "step": 13152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8255439953262232e-06,
+      "loss": 0.8438,
+      "step": 13153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8242369426315554e-06,
+      "loss": 0.8711,
+      "step": 13154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8229301426065836e-06,
+      "loss": 0.8867,
+      "step": 13155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821623595297326e-06,
+      "loss": 0.8125,
+      "step": 13156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820317300749782e-06,
+      "loss": 0.8203,
+      "step": 13157
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8190112590099504e-06,
+      "loss": 0.8047,
+      "step": 13158
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.817705470123825e-06,
+      "loss": 0.9844,
+      "step": 13159
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816399934137375e-06,
+      "loss": 0.9297,
+      "step": 13160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8150946510965737e-06,
+      "loss": 0.7695,
+      "step": 13161
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8137896210473847e-06,
+      "loss": 0.8242,
+      "step": 13162
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.812484844035761e-06,
+      "loss": 0.9414,
+      "step": 13163
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8111803201076406e-06,
+      "loss": 0.7422,
+      "step": 13164
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8098760493089606e-06,
+      "loss": 0.7617,
+      "step": 13165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.808572031685647e-06,
+      "loss": 0.7852,
+      "step": 13166
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8072682672836184e-06,
+      "loss": 0.7422,
+      "step": 13167
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8059647561487758e-06,
+      "loss": 0.8945,
+      "step": 13168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804661498327023e-06,
+      "loss": 0.9062,
+      "step": 13169
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.803358493864248e-06,
+      "loss": 0.8828,
+      "step": 13170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8020557428063343e-06,
+      "loss": 0.9219,
+      "step": 13171
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8007532451991494e-06,
+      "loss": 0.7852,
+      "step": 13172
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.799451001088558e-06,
+      "loss": 0.9023,
+      "step": 13173
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7981490105204133e-06,
+      "loss": 0.7227,
+      "step": 13174
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.796847273540566e-06,
+      "loss": 0.8672,
+      "step": 13175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7955457901948447e-06,
+      "loss": 0.7617,
+      "step": 13176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7942445605290802e-06,
+      "loss": 0.8633,
+      "step": 13177
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7929435845890895e-06,
+      "loss": 0.7734,
+      "step": 13178
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791642862420686e-06,
+      "loss": 0.7812,
+      "step": 13179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.790342394069664e-06,
+      "loss": 0.8438,
+      "step": 13180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7890421795818177e-06,
+      "loss": 0.7344,
+      "step": 13181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.787742219002931e-06,
+      "loss": 0.8047,
+      "step": 13182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7864425123787785e-06,
+      "loss": 0.8125,
+      "step": 13183
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.78514305975512e-06,
+      "loss": 0.8828,
+      "step": 13184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783843861177714e-06,
+      "loss": 0.9531,
+      "step": 13185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782544916692308e-06,
+      "loss": 0.8125,
+      "step": 13186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7812462263446415e-06,
+      "loss": 0.7891,
+      "step": 13187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7799477901804385e-06,
+      "loss": 0.8594,
+      "step": 13188
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7786496082454218e-06,
+      "loss": 0.8516,
+      "step": 13189
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777351680585302e-06,
+      "loss": 0.7188,
+      "step": 13190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.776054007245784e-06,
+      "loss": 0.7773,
+      "step": 13191
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7747565882725546e-06,
+      "loss": 0.8672,
+      "step": 13192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7734594237113024e-06,
+      "loss": 1.0156,
+      "step": 13193
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7721625136077014e-06,
+      "loss": 0.832,
+      "step": 13194
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.770865858007421e-06,
+      "loss": 0.8281,
+      "step": 13195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769569456956113e-06,
+      "loss": 0.8242,
+      "step": 13196
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7682733104994273e-06,
+      "loss": 0.957,
+      "step": 13197
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7669774186830046e-06,
+      "loss": 0.8867,
+      "step": 13198
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765681781552477e-06,
+      "loss": 0.7891,
+      "step": 13199
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7643863991534613e-06,
+      "loss": 0.8789,
+      "step": 13200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7630912715315717e-06,
+      "loss": 0.8438,
+      "step": 13201
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7617963987324127e-06,
+      "loss": 0.9414,
+      "step": 13202
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760501780801581e-06,
+      "loss": 0.9922,
+      "step": 13203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7592074177846573e-06,
+      "loss": 0.9336,
+      "step": 13204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.757913309727219e-06,
+      "loss": 0.8125,
+      "step": 13205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756619456674835e-06,
+      "loss": 0.8906,
+      "step": 13206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.755325858673067e-06,
+      "loss": 0.7773,
+      "step": 13207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7540325157674586e-06,
+      "loss": 0.832,
+      "step": 13208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7527394280035523e-06,
+      "loss": 0.8789,
+      "step": 13209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.751446595426881e-06,
+      "loss": 0.8828,
+      "step": 13210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.75015401808297e-06,
+      "loss": 0.918,
+      "step": 13211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7488616960173265e-06,
+      "loss": 0.8711,
+      "step": 13212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7475696292754585e-06,
+      "loss": 0.8359,
+      "step": 13213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7462778179028636e-06,
+      "loss": 0.7852,
+      "step": 13214
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.744986261945024e-06,
+      "loss": 0.8711,
+      "step": 13215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7436949614474205e-06,
+      "loss": 0.7773,
+      "step": 13216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7424039164555204e-06,
+      "loss": 1.0156,
+      "step": 13217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7411131270147873e-06,
+      "loss": 0.8047,
+      "step": 13218
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.739822593170666e-06,
+      "loss": 0.8281,
+      "step": 13219
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7385323149686006e-06,
+      "loss": 0.8438,
+      "step": 13220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7372422924540244e-06,
+      "loss": 0.8594,
+      "step": 13221
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735952525672363e-06,
+      "loss": 0.7461,
+      "step": 13222
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7346630146690266e-06,
+      "loss": 0.7734,
+      "step": 13223
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7333737594894226e-06,
+      "loss": 0.8438,
+      "step": 13224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.732084760178948e-06,
+      "loss": 0.8906,
+      "step": 13225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.730796016782995e-06,
+      "loss": 0.7695,
+      "step": 13226
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7295075293469332e-06,
+      "loss": 0.8203,
+      "step": 13227
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.728219297916138e-06,
+      "loss": 0.7344,
+      "step": 13228
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7269313225359715e-06,
+      "loss": 0.7969,
+      "step": 13229
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.72564360325178e-06,
+      "loss": 0.8828,
+      "step": 13230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.724356140108908e-06,
+      "loss": 0.8828,
+      "step": 13231
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7230689331526905e-06,
+      "loss": 0.8477,
+      "step": 13232
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7217819824284543e-06,
+      "loss": 0.8555,
+      "step": 13233
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7204952879815096e-06,
+      "loss": 0.7383,
+      "step": 13234
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7192088498571647e-06,
+      "loss": 0.8008,
+      "step": 13235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7179226681007176e-06,
+      "loss": 0.7969,
+      "step": 13236
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7166367427574593e-06,
+      "loss": 0.8359,
+      "step": 13237
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.715351073872665e-06,
+      "loss": 0.8594,
+      "step": 13238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7140656614916063e-06,
+      "loss": 0.7656,
+      "step": 13239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7127805056595447e-06,
+      "loss": 0.9023,
+      "step": 13240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.711495606421737e-06,
+      "loss": 0.8359,
+      "step": 13241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7102109638234185e-06,
+      "loss": 0.7969,
+      "step": 13242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7089265779098283e-06,
+      "loss": 0.9375,
+      "step": 13243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.707642448726193e-06,
+      "loss": 0.8203,
+      "step": 13244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7063585763177235e-06,
+      "loss": 0.8945,
+      "step": 13245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.70507496072963e-06,
+      "loss": 0.9102,
+      "step": 13246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7037916020071097e-06,
+      "loss": 0.7695,
+      "step": 13247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7025085001953566e-06,
+      "loss": 0.8438,
+      "step": 13248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7012256553395434e-06,
+      "loss": 0.7461,
+      "step": 13249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6999430674848436e-06,
+      "loss": 0.9297,
+      "step": 13250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6986607366764193e-06,
+      "loss": 1.0156,
+      "step": 13251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.697378662959428e-06,
+      "loss": 0.9102,
+      "step": 13252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696096846379005e-06,
+      "loss": 0.7695,
+      "step": 13253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6948152869802892e-06,
+      "loss": 0.8398,
+      "step": 13254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6935339848084097e-06,
+      "loss": 0.9219,
+      "step": 13255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6922529399084773e-06,
+      "loss": 0.7656,
+      "step": 13256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6909721523256015e-06,
+      "loss": 0.8359,
+      "step": 13257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6896916221048806e-06,
+      "loss": 0.832,
+      "step": 13258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.688411349291409e-06,
+      "loss": 0.8086,
+      "step": 13259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6871313339302596e-06,
+      "loss": 0.8359,
+      "step": 13260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.685851576066507e-06,
+      "loss": 1.0156,
+      "step": 13261
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6845720757452134e-06,
+      "loss": 0.7539,
+      "step": 13262
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683292833011434e-06,
+      "loss": 0.7969,
+      "step": 13263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6820138479102087e-06,
+      "loss": 0.875,
+      "step": 13264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680735120486575e-06,
+      "loss": 0.7695,
+      "step": 13265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.679456650785558e-06,
+      "loss": 0.7812,
+      "step": 13266
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678178438852179e-06,
+      "loss": 0.8672,
+      "step": 13267
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6769004847314386e-06,
+      "loss": 0.7617,
+      "step": 13268
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6756227884683393e-06,
+      "loss": 0.7461,
+      "step": 13269
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674345350107873e-06,
+      "loss": 0.7734,
+      "step": 13270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6730681696950146e-06,
+      "loss": 0.8867,
+      "step": 13271
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6717912472747386e-06,
+      "loss": 0.8359,
+      "step": 13272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6705145828920085e-06,
+      "loss": 0.7656,
+      "step": 13273
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6692381765917785e-06,
+      "loss": 0.9102,
+      "step": 13274
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6679620284189887e-06,
+      "loss": 0.7852,
+      "step": 13275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.666686138418576e-06,
+      "loss": 0.8711,
+      "step": 13276
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6654105066354675e-06,
+      "loss": 0.7734,
+      "step": 13277
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6641351331145827e-06,
+      "loss": 0.9883,
+      "step": 13278
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6628600179008226e-06,
+      "loss": 0.9531,
+      "step": 13279
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661585161039091e-06,
+      "loss": 0.8516,
+      "step": 13280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6603105625742754e-06,
+      "loss": 0.8672,
+      "step": 13281
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6590362225512602e-06,
+      "loss": 0.8633,
+      "step": 13282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.657762141014911e-06,
+      "loss": 0.9297,
+      "step": 13283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.656488318010094e-06,
+      "loss": 0.8516,
+      "step": 13284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.655214753581664e-06,
+      "loss": 0.9727,
+      "step": 13285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6539414477744607e-06,
+      "loss": 0.9805,
+      "step": 13286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6526684006333204e-06,
+      "loss": 0.9297,
+      "step": 13287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6513956122030705e-06,
+      "loss": 0.8945,
+      "step": 13288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6501230825285294e-06,
+      "loss": 0.8906,
+      "step": 13289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.648850811654501e-06,
+      "loss": 0.8164,
+      "step": 13290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.647578799625785e-06,
+      "loss": 0.8906,
+      "step": 13291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646307046487172e-06,
+      "loss": 0.793,
+      "step": 13292
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6450355522834446e-06,
+      "loss": 0.8672,
+      "step": 13293
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6437643170593686e-06,
+      "loss": 0.8594,
+      "step": 13294
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6424933408597085e-06,
+      "loss": 0.957,
+      "step": 13295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.641222623729218e-06,
+      "loss": 0.8828,
+      "step": 13296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6399521657126437e-06,
+      "loss": 0.8672,
+      "step": 13297
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.638681966854716e-06,
+      "loss": 1.0156,
+      "step": 13298
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6374120272001603e-06,
+      "loss": 0.8516,
+      "step": 13299
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6361423467936954e-06,
+      "loss": 0.7695,
+      "step": 13300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6348729256800323e-06,
+      "loss": 0.832,
+      "step": 13301
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6336037639038602e-06,
+      "loss": 0.8594,
+      "step": 13302
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.632334861509875e-06,
+      "loss": 0.7812,
+      "step": 13303
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.631066218542754e-06,
+      "loss": 0.7539,
+      "step": 13304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629797835047173e-06,
+      "loss": 0.8086,
+      "step": 13305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6285297110677853e-06,
+      "loss": 0.957,
+      "step": 13306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6272618466492496e-06,
+      "loss": 0.8906,
+      "step": 13307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625994241836207e-06,
+      "loss": 0.8281,
+      "step": 13308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6247268966732953e-06,
+      "loss": 0.832,
+      "step": 13309
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6234598112051344e-06,
+      "loss": 0.9102,
+      "step": 13310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622192985476343e-06,
+      "loss": 0.7773,
+      "step": 13311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6209264195315276e-06,
+      "loss": 0.9141,
+      "step": 13312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61966011341529e-06,
+      "loss": 0.875,
+      "step": 13313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6183940671722107e-06,
+      "loss": 0.8125,
+      "step": 13314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6171282808468745e-06,
+      "loss": 0.8711,
+      "step": 13315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61586275448385e-06,
+      "loss": 0.9922,
+      "step": 13316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6145974881277026e-06,
+      "loss": 0.9375,
+      "step": 13317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613332481822977e-06,
+      "loss": 0.8711,
+      "step": 13318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61206773561422e-06,
+      "loss": 0.8359,
+      "step": 13319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6108032495459656e-06,
+      "loss": 0.9141,
+      "step": 13320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6095390236627393e-06,
+      "loss": 0.9219,
+      "step": 13321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.608275058009053e-06,
+      "loss": 0.7617,
+      "step": 13322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607011352629414e-06,
+      "loss": 0.7891,
+      "step": 13323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6057479075683203e-06,
+      "loss": 0.8242,
+      "step": 13324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.604484722870263e-06,
+      "loss": 0.7617,
+      "step": 13325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6032217985797136e-06,
+      "loss": 0.8555,
+      "step": 13326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6019591347411454e-06,
+      "loss": 0.8555,
+      "step": 13327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6006967313990216e-06,
+      "loss": 0.8203,
+      "step": 13328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5994345885977877e-06,
+      "loss": 0.8125,
+      "step": 13329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598172706381887e-06,
+      "loss": 0.8242,
+      "step": 13330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5969110847957557e-06,
+      "loss": 0.8203,
+      "step": 13331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5956497238838173e-06,
+      "loss": 1.0156,
+      "step": 13332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5943886236904814e-06,
+      "loss": 0.7734,
+      "step": 13333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5931277842601566e-06,
+      "loss": 0.8281,
+      "step": 13334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5918672056372386e-06,
+      "loss": 0.8867,
+      "step": 13335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5906068878661174e-06,
+      "loss": 0.8281,
+      "step": 13336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.589346830991165e-06,
+      "loss": 0.957,
+      "step": 13337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588087035056752e-06,
+      "loss": 0.8477,
+      "step": 13338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5868275001072396e-06,
+      "loss": 0.8555,
+      "step": 13339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5855682261869796e-06,
+      "loss": 0.875,
+      "step": 13340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584309213340306e-06,
+      "loss": 0.8828,
+      "step": 13341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5830504616115557e-06,
+      "loss": 0.8633,
+      "step": 13342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58179197104505e-06,
+      "loss": 0.7891,
+      "step": 13343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580533741685106e-06,
+      "loss": 0.8281,
+      "step": 13344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579275773576022e-06,
+      "loss": 0.8242,
+      "step": 13345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578018066762095e-06,
+      "loss": 1.0312,
+      "step": 13346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5767606212876115e-06,
+      "loss": 0.918,
+      "step": 13347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575503437196851e-06,
+      "loss": 0.8281,
+      "step": 13348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5742465145340744e-06,
+      "loss": 0.8633,
+      "step": 13349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572989853343545e-06,
+      "loss": 0.7891,
+      "step": 13350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5717334536695084e-06,
+      "loss": 0.793,
+      "step": 13351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.570477315556209e-06,
+      "loss": 0.9688,
+      "step": 13352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5692214390478733e-06,
+      "loss": 0.8672,
+      "step": 13353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.567965824188722e-06,
+      "loss": 0.8945,
+      "step": 13354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5667104710229697e-06,
+      "loss": 0.9023,
+      "step": 13355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5654553795948224e-06,
+      "loss": 0.9375,
+      "step": 13356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5642005499484657e-06,
+      "loss": 0.8516,
+      "step": 13357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.562945982128089e-06,
+      "loss": 0.793,
+      "step": 13358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5616916761778674e-06,
+      "loss": 0.8047,
+      "step": 13359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.56043763214197e-06,
+      "loss": 0.8594,
+      "step": 13360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559183850064546e-06,
+      "loss": 0.7969,
+      "step": 13361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5579303299897486e-06,
+      "loss": 1.0234,
+      "step": 13362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.556677071961714e-06,
+      "loss": 0.8516,
+      "step": 13363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.555424076024575e-06,
+      "loss": 1.0156,
+      "step": 13364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5541713422224457e-06,
+      "loss": 0.8633,
+      "step": 13365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55291887059944e-06,
+      "loss": 0.8555,
+      "step": 13366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551666661199659e-06,
+      "loss": 0.8086,
+      "step": 13367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5504147140671985e-06,
+      "loss": 0.8008,
+      "step": 13368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5491630292461343e-06,
+      "loss": 0.9258,
+      "step": 13369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5479116067805444e-06,
+      "loss": 0.8008,
+      "step": 13370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546660446714493e-06,
+      "loss": 1.0078,
+      "step": 13371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545409549092038e-06,
+      "loss": 0.9922,
+      "step": 13372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5441589139572197e-06,
+      "loss": 0.8477,
+      "step": 13373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5429085413540778e-06,
+      "loss": 0.8086,
+      "step": 13374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541658431326639e-06,
+      "loss": 0.9414,
+      "step": 13375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5404085839189263e-06,
+      "loss": 0.957,
+      "step": 13376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.539158999174942e-06,
+      "loss": 0.832,
+      "step": 13377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.537909677138689e-06,
+      "loss": 0.9297,
+      "step": 13378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5366606178541573e-06,
+      "loss": 0.7852,
+      "step": 13379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5354118213653324e-06,
+      "loss": 0.7734,
+      "step": 13380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534163287716178e-06,
+      "loss": 0.9141,
+      "step": 13381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5329150169506634e-06,
+      "loss": 0.957,
+      "step": 13382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5316670091127383e-06,
+      "loss": 0.8906,
+      "step": 13383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.530419264246353e-06,
+      "loss": 0.7891,
+      "step": 13384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5291717823954354e-06,
+      "loss": 0.6953,
+      "step": 13385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527924563603914e-06,
+      "loss": 0.9023,
+      "step": 13386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5266776079157083e-06,
+      "loss": 0.8594,
+      "step": 13387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.52543091537472e-06,
+      "loss": 0.9961,
+      "step": 13388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524184486024851e-06,
+      "loss": 0.9258,
+      "step": 13389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5229383199099876e-06,
+      "loss": 0.7422,
+      "step": 13390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.521692417074013e-06,
+      "loss": 0.7969,
+      "step": 13391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.520446777560793e-06,
+      "loss": 0.8594,
+      "step": 13392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.519201401414191e-06,
+      "loss": 0.8516,
+      "step": 13393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5179562886780563e-06,
+      "loss": 0.9141,
+      "step": 13394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5167114393962366e-06,
+      "loss": 0.8555,
+      "step": 13395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.515466853612558e-06,
+      "loss": 0.8516,
+      "step": 13396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514222531370848e-06,
+      "loss": 0.875,
+      "step": 13397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512978472714921e-06,
+      "loss": 0.8359,
+      "step": 13398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5117346776885843e-06,
+      "loss": 0.8867,
+      "step": 13399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5104911463356286e-06,
+      "loss": 0.7578,
+      "step": 13400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.509247878699843e-06,
+      "loss": 0.8906,
+      "step": 13401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5080048748250063e-06,
+      "loss": 0.8008,
+      "step": 13402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5067621347548876e-06,
+      "loss": 0.8633,
+      "step": 13403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.505519658533242e-06,
+      "loss": 0.7344,
+      "step": 13404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.504277446203819e-06,
+      "loss": 1.0312,
+      "step": 13405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503035497810361e-06,
+      "loss": 0.7539,
+      "step": 13406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5017938133966022e-06,
+      "loss": 0.7656,
+      "step": 13407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.500552393006257e-06,
+      "loss": 0.8438,
+      "step": 13408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4993112366830406e-06,
+      "loss": 0.8906,
+      "step": 13409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4980703444706567e-06,
+      "loss": 0.7617,
+      "step": 13410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496829716412803e-06,
+      "loss": 0.7344,
+      "step": 13411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.495589352553156e-06,
+      "loss": 0.8242,
+      "step": 13412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494349252935394e-06,
+      "loss": 0.8633,
+      "step": 13413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4931094176031845e-06,
+      "loss": 0.7773,
+      "step": 13414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4918698466001854e-06,
+      "loss": 0.8086,
+      "step": 13415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4906305399700393e-06,
+      "loss": 0.7656,
+      "step": 13416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489391497756386e-06,
+      "loss": 0.875,
+      "step": 13417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4881527200028543e-06,
+      "loss": 0.8086,
+      "step": 13418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4869142067530673e-06,
+      "loss": 0.6836,
+      "step": 13419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485675958050627e-06,
+      "loss": 0.9219,
+      "step": 13420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4844379739391388e-06,
+      "loss": 0.8125,
+      "step": 13421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4832002544621938e-06,
+      "loss": 0.875,
+      "step": 13422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4819627996633767e-06,
+      "loss": 0.7617,
+      "step": 13423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4807256095862542e-06,
+      "loss": 0.8711,
+      "step": 13424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4794886842743927e-06,
+      "loss": 0.8594,
+      "step": 13425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478252023771346e-06,
+      "loss": 0.9609,
+      "step": 13426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4770156281206625e-06,
+      "loss": 0.7383,
+      "step": 13427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47577949736587e-06,
+      "loss": 0.8164,
+      "step": 13428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4745436315505e-06,
+      "loss": 0.7617,
+      "step": 13429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4733080307180677e-06,
+      "loss": 0.8438,
+      "step": 13430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4720726949120845e-06,
+      "loss": 0.7812,
+      "step": 13431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4708376241760412e-06,
+      "loss": 0.8281,
+      "step": 13432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469602818553429e-06,
+      "loss": 0.8203,
+      "step": 13433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4683682780877305e-06,
+      "loss": 0.7461,
+      "step": 13434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4671340028224156e-06,
+      "loss": 0.9297,
+      "step": 13435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4658999928009407e-06,
+      "loss": 0.8203,
+      "step": 13436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464666248066759e-06,
+      "loss": 0.9102,
+      "step": 13437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4634327686633143e-06,
+      "loss": 0.8164,
+      "step": 13438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4621995546340395e-06,
+      "loss": 0.9023,
+      "step": 13439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.460966606022355e-06,
+      "loss": 0.8047,
+      "step": 13440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4597339228716765e-06,
+      "loss": 0.7266,
+      "step": 13441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4585015052254112e-06,
+      "loss": 0.7188,
+      "step": 13442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4572693531269498e-06,
+      "loss": 0.8828,
+      "step": 13443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4560374666196796e-06,
+      "loss": 0.9297,
+      "step": 13444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.454805845746978e-06,
+      "loss": 0.9688,
+      "step": 13445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4535744905522165e-06,
+      "loss": 0.8281,
+      "step": 13446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.452343401078744e-06,
+      "loss": 0.8359,
+      "step": 13447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4511125773699153e-06,
+      "loss": 0.7578,
+      "step": 13448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4498820194690675e-06,
+      "loss": 0.9453,
+      "step": 13449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4486517274195344e-06,
+      "loss": 0.8516,
+      "step": 13450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4474217012646308e-06,
+      "loss": 0.8398,
+      "step": 13451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4461919410476696e-06,
+      "loss": 0.9766,
+      "step": 13452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4449624468119537e-06,
+      "loss": 0.7773,
+      "step": 13453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4437332186007768e-06,
+      "loss": 0.8047,
+      "step": 13454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 1.0625,
+      "step": 13455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441275560425154e-06,
+      "loss": 1.0312,
+      "step": 13456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440047130547246e-06,
+      "loss": 0.7773,
+      "step": 13457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.438818966866956e-06,
+      "loss": 0.8633,
+      "step": 13458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4375910694275207e-06,
+      "loss": 0.7578,
+      "step": 13459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4363634382721803e-06,
+      "loss": 0.8086,
+      "step": 13460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.435136073444161e-06,
+      "loss": 0.9414,
+      "step": 13461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.433908974986684e-06,
+      "loss": 0.8516,
+      "step": 13462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4326821429429516e-06,
+      "loss": 0.8945,
+      "step": 13463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4314555773561643e-06,
+      "loss": 0.7852,
+      "step": 13464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4302292782695125e-06,
+      "loss": 0.8203,
+      "step": 13465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429003245726178e-06,
+      "loss": 1.0156,
+      "step": 13466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4277774797693253e-06,
+      "loss": 0.793,
+      "step": 13467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42655198044212e-06,
+      "loss": 0.8086,
+      "step": 13468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.425326747787712e-06,
+      "loss": 1.0469,
+      "step": 13469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4241017818492486e-06,
+      "loss": 0.8906,
+      "step": 13470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4228770826698546e-06,
+      "loss": 0.9102,
+      "step": 13471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4216526502926575e-06,
+      "loss": 0.7656,
+      "step": 13472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4204284847607717e-06,
+      "loss": 0.8672,
+      "step": 13473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.419204586117305e-06,
+      "loss": 0.7812,
+      "step": 13474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.417980954405347e-06,
+      "loss": 0.7812,
+      "step": 13475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.416757589667985e-06,
+      "loss": 0.9375,
+      "step": 13476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.415534491948298e-06,
+      "loss": 0.7812,
+      "step": 13477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4143116612893547e-06,
+      "loss": 0.8125,
+      "step": 13478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413089097734207e-06,
+      "loss": 0.7461,
+      "step": 13479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411866801325906e-06,
+      "loss": 0.793,
+      "step": 13480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.410644772107491e-06,
+      "loss": 0.8594,
+      "step": 13481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409423010121994e-06,
+      "loss": 0.7227,
+      "step": 13482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4082015154124307e-06,
+      "loss": 0.9102,
+      "step": 13483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406980288021813e-06,
+      "loss": 0.8398,
+      "step": 13484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4057593279931434e-06,
+      "loss": 0.7812,
+      "step": 13485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4045386353694166e-06,
+      "loss": 0.8281,
+      "step": 13486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.403318210193608e-06,
+      "loss": 0.8633,
+      "step": 13487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4020980525086945e-06,
+      "loss": 0.9609,
+      "step": 13488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4008781623576405e-06,
+      "loss": 0.8281,
+      "step": 13489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3996585397834014e-06,
+      "loss": 0.8711,
+      "step": 13490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.398439184828918e-06,
+      "loss": 0.7773,
+      "step": 13491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3972200975371274e-06,
+      "loss": 0.8047,
+      "step": 13492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3960012779509557e-06,
+      "loss": 0.8281,
+      "step": 13493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3947827261133226e-06,
+      "loss": 0.8828,
+      "step": 13494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393564442067129e-06,
+      "loss": 0.6914,
+      "step": 13495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392346425855275e-06,
+      "loss": 0.8203,
+      "step": 13496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3911286775206542e-06,
+      "loss": 0.8633,
+      "step": 13497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3899111971061362e-06,
+      "loss": 0.8789,
+      "step": 13498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3886939846545963e-06,
+      "loss": 0.9297,
+      "step": 13499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387477040208892e-06,
+      "loss": 0.8438,
+      "step": 13500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3862603638118775e-06,
+      "loss": 0.7305,
+      "step": 13501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.385043955506389e-06,
+      "loss": 0.9648,
+      "step": 13502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.38382781533526e-06,
+      "loss": 0.8438,
+      "step": 13503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3826119433413133e-06,
+      "loss": 0.875,
+      "step": 13504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3813963395673636e-06,
+      "loss": 0.6719,
+      "step": 13505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.38018100405621e-06,
+      "loss": 0.7969,
+      "step": 13506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378965936850647e-06,
+      "loss": 0.9219,
+      "step": 13507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3777511379934604e-06,
+      "loss": 0.7422,
+      "step": 13508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3765366075274287e-06,
+      "loss": 0.9375,
+      "step": 13509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.37532234549531e-06,
+      "loss": 0.8867,
+      "step": 13510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3741083519398646e-06,
+      "loss": 0.8438,
+      "step": 13511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.372894626903838e-06,
+      "loss": 0.8359,
+      "step": 13512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.37168117042997e-06,
+      "loss": 0.8984,
+      "step": 13513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3704679825609845e-06,
+      "loss": 0.8633,
+      "step": 13514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3692550633396006e-06,
+      "loss": 0.9062,
+      "step": 13515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.368042412808528e-06,
+      "loss": 0.8867,
+      "step": 13516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3668300310104686e-06,
+      "loss": 0.8711,
+      "step": 13517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3656179179881066e-06,
+      "loss": 0.7969,
+      "step": 13518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3644060737841247e-06,
+      "loss": 0.7656,
+      "step": 13519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3631944984411947e-06,
+      "loss": 0.8242,
+      "step": 13520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.361983192001981e-06,
+      "loss": 0.7539,
+      "step": 13521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360772154509129e-06,
+      "loss": 0.9102,
+      "step": 13522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3595613860052835e-06,
+      "loss": 0.9219,
+      "step": 13523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3583508865330794e-06,
+      "loss": 0.9844,
+      "step": 13524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357140656135142e-06,
+      "loss": 0.8281,
+      "step": 13525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3559306948540785e-06,
+      "loss": 0.8086,
+      "step": 13526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3547210027324987e-06,
+      "loss": 0.8984,
+      "step": 13527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3535115798129972e-06,
+      "loss": 0.832,
+      "step": 13528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3523024261381613e-06,
+      "loss": 0.8047,
+      "step": 13529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3510935417505622e-06,
+      "loss": 0.8945,
+      "step": 13530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.349884926692769e-06,
+      "loss": 0.8242,
+      "step": 13531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3486765810073397e-06,
+      "loss": 0.8906,
+      "step": 13532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.347468504736825e-06,
+      "loss": 0.7812,
+      "step": 13533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.346260697923756e-06,
+      "loss": 0.7891,
+      "step": 13534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.345053160610665e-06,
+      "loss": 0.8047,
+      "step": 13535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.343845892840072e-06,
+      "loss": 0.9141,
+      "step": 13536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.342638894654489e-06,
+      "loss": 0.8398,
+      "step": 13537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341432166096411e-06,
+      "loss": 0.7578,
+      "step": 13538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.34022570720833e-06,
+      "loss": 0.8359,
+      "step": 13539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33901951803273e-06,
+      "loss": 0.8086,
+      "step": 13540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3378135986120832e-06,
+      "loss": 0.7969,
+      "step": 13541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3366079489888473e-06,
+      "loss": 0.8086,
+      "step": 13542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3354025692054784e-06,
+      "loss": 0.6992,
+      "step": 13543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.334197459304418e-06,
+      "loss": 0.8359,
+      "step": 13544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332992619328105e-06,
+      "loss": 0.8789,
+      "step": 13545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3317880493189558e-06,
+      "loss": 0.7773,
+      "step": 13546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3305837493193894e-06,
+      "loss": 0.7578,
+      "step": 13547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3293797193718114e-06,
+      "loss": 0.9727,
+      "step": 13548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3281759595186194e-06,
+      "loss": 0.8477,
+      "step": 13549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3269724698021935e-06,
+      "loss": 0.7578,
+      "step": 13550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.325769250264914e-06,
+      "loss": 0.8242,
+      "step": 13551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3245663009491494e-06,
+      "loss": 0.8555,
+      "step": 13552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3233636218972587e-06,
+      "loss": 0.9414,
+      "step": 13553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322161213151585e-06,
+      "loss": 0.8086,
+      "step": 13554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3209590747544686e-06,
+      "loss": 0.8008,
+      "step": 13555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3197572067482433e-06,
+      "loss": 0.8516,
+      "step": 13556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3185556091752214e-06,
+      "loss": 0.7695,
+      "step": 13557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3173542820777174e-06,
+      "loss": 0.8242,
+      "step": 13558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31615322549803e-06,
+      "loss": 0.7578,
+      "step": 13559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.314952439478455e-06,
+      "loss": 0.9219,
+      "step": 13560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.313751924061267e-06,
+      "loss": 0.8164,
+      "step": 13561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3125516792887415e-06,
+      "loss": 0.9531,
+      "step": 13562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.311351705203141e-06,
+      "loss": 0.9219,
+      "step": 13563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3101520018467214e-06,
+      "loss": 0.7422,
+      "step": 13564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3089525692617197e-06,
+      "loss": 0.8867,
+      "step": 13565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.307753407490373e-06,
+      "loss": 0.8047,
+      "step": 13566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.306554516574905e-06,
+      "loss": 0.8711,
+      "step": 13567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.305355896557534e-06,
+      "loss": 0.7305,
+      "step": 13568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.30415754748046e-06,
+      "loss": 0.8867,
+      "step": 13569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3029594693858816e-06,
+      "loss": 0.9062,
+      "step": 13570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3017616623159833e-06,
+      "loss": 0.7734,
+      "step": 13571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3005641263129475e-06,
+      "loss": 0.7617,
+      "step": 13572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2993668614189322e-06,
+      "loss": 0.8008,
+      "step": 13573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2981698676761e-06,
+      "loss": 0.7773,
+      "step": 13574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296973145126601e-06,
+      "loss": 0.7812,
+      "step": 13575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.295776693812569e-06,
+      "loss": 0.7617,
+      "step": 13576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2945805137761345e-06,
+      "loss": 0.9141,
+      "step": 13577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2933846050594167e-06,
+      "loss": 0.8477,
+      "step": 13578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2921889677045296e-06,
+      "loss": 0.8633,
+      "step": 13579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290993601753566e-06,
+      "loss": 0.7109,
+      "step": 13580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2897985072486205e-06,
+      "loss": 0.7461,
+      "step": 13581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.288603684231775e-06,
+      "loss": 1.0547,
+      "step": 13582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287409132745102e-06,
+      "loss": 0.8906,
+      "step": 13583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2862148528306593e-06,
+      "loss": 0.8398,
+      "step": 13584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285020844530502e-06,
+      "loss": 0.8516,
+      "step": 13585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2838271078866714e-06,
+      "loss": 0.7852,
+      "step": 13586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2826336429412067e-06,
+      "loss": 0.8477,
+      "step": 13587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2814404497361233e-06,
+      "loss": 0.875,
+      "step": 13588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2802475283134384e-06,
+      "loss": 0.8125,
+      "step": 13589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2790548787151613e-06,
+      "loss": 0.8516,
+      "step": 13590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2778625009832788e-06,
+      "loss": 0.8477,
+      "step": 13591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.276670395159782e-06,
+      "loss": 0.8672,
+      "step": 13592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.275478561286645e-06,
+      "loss": 0.8164,
+      "step": 13593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274286999405837e-06,
+      "loss": 0.7109,
+      "step": 13594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2730957095593097e-06,
+      "loss": 0.8477,
+      "step": 13595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.271904691789012e-06,
+      "loss": 0.7852,
+      "step": 13596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270713946136883e-06,
+      "loss": 0.8242,
+      "step": 13597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.269523472644852e-06,
+      "loss": 0.832,
+      "step": 13598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2683332713548336e-06,
+      "loss": 0.8516,
+      "step": 13599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2671433423087386e-06,
+      "loss": 0.8281,
+      "step": 13600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265953685548468e-06,
+      "loss": 0.9023,
+      "step": 13601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.264764301115907e-06,
+      "loss": 0.8516,
+      "step": 13602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2635751890529378e-06,
+      "loss": 0.9062,
+      "step": 13603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262386349401431e-06,
+      "loss": 0.7852,
+      "step": 13604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2611977822032505e-06,
+      "loss": 0.8555,
+      "step": 13605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260009487500243e-06,
+      "loss": 0.8906,
+      "step": 13606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.258821465334251e-06,
+      "loss": 0.8125,
+      "step": 13607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2576337157471083e-06,
+      "loss": 0.8086,
+      "step": 13608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2564462387806386e-06,
+      "loss": 0.8203,
+      "step": 13609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25525903447665e-06,
+      "loss": 0.8047,
+      "step": 13610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2540721028769497e-06,
+      "loss": 0.7812,
+      "step": 13611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2528854440233296e-06,
+      "loss": 0.7188,
+      "step": 13612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251699057957577e-06,
+      "loss": 0.832,
+      "step": 13613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2505129447214614e-06,
+      "loss": 0.9531,
+      "step": 13614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249327104356751e-06,
+      "loss": 0.8828,
+      "step": 13615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2481415369052017e-06,
+      "loss": 0.8711,
+      "step": 13616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2469562424085555e-06,
+      "loss": 0.9688,
+      "step": 13617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.245771220908551e-06,
+      "loss": 0.8203,
+      "step": 13618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2445864724469146e-06,
+      "loss": 1.0234,
+      "step": 13619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2434019970653655e-06,
+      "loss": 0.8086,
+      "step": 13620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2422177948056044e-06,
+      "loss": 0.9375,
+      "step": 13621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2410338657093335e-06,
+      "loss": 0.875,
+      "step": 13622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2398502098182407e-06,
+      "loss": 0.7891,
+      "step": 13623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2386668271740054e-06,
+      "loss": 0.8516,
+      "step": 13624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2374837178182917e-06,
+      "loss": 0.8594,
+      "step": 13625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2363008817927625e-06,
+      "loss": 0.9023,
+      "step": 13626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.235118319139066e-06,
+      "loss": 0.8633,
+      "step": 13627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2339360298988443e-06,
+      "loss": 0.8281,
+      "step": 13628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.232754014113724e-06,
+      "loss": 0.8203,
+      "step": 13629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2315722718253276e-06,
+      "loss": 0.832,
+      "step": 13630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2303908030752674e-06,
+      "loss": 0.8047,
+      "step": 13631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229209607905142e-06,
+      "loss": 0.8125,
+      "step": 13632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.228028686356544e-06,
+      "loss": 0.8438,
+      "step": 13633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2268480384710545e-06,
+      "loss": 0.8516,
+      "step": 13634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225667664290252e-06,
+      "loss": 0.8125,
+      "step": 13635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224487563855691e-06,
+      "loss": 0.9609,
+      "step": 13636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2233077372089274e-06,
+      "loss": 0.8672,
+      "step": 13637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2221281843915067e-06,
+      "loss": 0.8281,
+      "step": 13638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.220948905444964e-06,
+      "loss": 0.9141,
+      "step": 13639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2197699004108186e-06,
+      "loss": 0.7812,
+      "step": 13640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.218591169330587e-06,
+      "loss": 0.7344,
+      "step": 13641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2174127122457756e-06,
+      "loss": 0.9531,
+      "step": 13642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2162345291978816e-06,
+      "loss": 0.8203,
+      "step": 13643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.215056620228385e-06,
+      "loss": 0.8125,
+      "step": 13644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2138789853787648e-06,
+      "loss": 0.7969,
+      "step": 13645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2127016246904864e-06,
+      "loss": 1.0078,
+      "step": 13646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.211524538205011e-06,
+      "loss": 0.8164,
+      "step": 13647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2103477259637783e-06,
+      "loss": 0.9336,
+      "step": 13648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209171188008231e-06,
+      "loss": 0.875,
+      "step": 13649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207994924379794e-06,
+      "loss": 0.7617,
+      "step": 13650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.20681893511989e-06,
+      "loss": 0.8516,
+      "step": 13651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.205643220269921e-06,
+      "loss": 0.8047,
+      "step": 13652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2044677798712877e-06,
+      "loss": 0.9961,
+      "step": 13653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.203292613965381e-06,
+      "loss": 0.9258,
+      "step": 13654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.202117722593582e-06,
+      "loss": 0.8711,
+      "step": 13655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2009431057972553e-06,
+      "loss": 0.832,
+      "step": 13656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1997687636177635e-06,
+      "loss": 0.9219,
+      "step": 13657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1985946960964565e-06,
+      "loss": 0.7383,
+      "step": 13658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1974209032746797e-06,
+      "loss": 0.8242,
+      "step": 13659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196247385193756e-06,
+      "loss": 0.7617,
+      "step": 13660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1950741418950115e-06,
+      "loss": 0.8398,
+      "step": 13661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1939011734197567e-06,
+      "loss": 0.8281,
+      "step": 13662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1927284798092975e-06,
+      "loss": 0.875,
+      "step": 13663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.191556061104919e-06,
+      "loss": 0.8008,
+      "step": 13664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1903839173479093e-06,
+      "loss": 0.9258,
+      "step": 13665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189212048579542e-06,
+      "loss": 0.8711,
+      "step": 13666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1880404548410762e-06,
+      "loss": 0.875,
+      "step": 13667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186869136173766e-06,
+      "loss": 0.9062,
+      "step": 13668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1856980926188587e-06,
+      "loss": 0.7891,
+      "step": 13669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184527324217588e-06,
+      "loss": 0.7227,
+      "step": 13670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1833568310111753e-06,
+      "loss": 0.7852,
+      "step": 13671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1821866130408377e-06,
+      "loss": 0.8008,
+      "step": 13672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18101667034778e-06,
+      "loss": 0.9297,
+      "step": 13673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.179847002973201e-06,
+      "loss": 0.7969,
+      "step": 13674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1786776109582796e-06,
+      "loss": 0.7578,
+      "step": 13675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.177508494344196e-06,
+      "loss": 0.6914,
+      "step": 13676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176339653172116e-06,
+      "loss": 0.918,
+      "step": 13677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1751710874832e-06,
+      "loss": 0.7656,
+      "step": 13678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1740027973185883e-06,
+      "loss": 0.8555,
+      "step": 13679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1728347827194206e-06,
+      "loss": 0.8672,
+      "step": 13680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1716670437268274e-06,
+      "loss": 0.75,
+      "step": 13681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170499580381925e-06,
+      "loss": 0.7422,
+      "step": 13682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169332392725819e-06,
+      "loss": 0.75,
+      "step": 13683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.168165480799611e-06,
+      "loss": 0.9688,
+      "step": 13684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1669988446443877e-06,
+      "loss": 0.8281,
+      "step": 13685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165832484301232e-06,
+      "loss": 0.8164,
+      "step": 13686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164666399811208e-06,
+      "loss": 0.7617,
+      "step": 13687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1635005912153783e-06,
+      "loss": 0.7305,
+      "step": 13688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.162335058554792e-06,
+      "loss": 0.7969,
+      "step": 13689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1611698018704917e-06,
+      "loss": 0.9688,
+      "step": 13690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.160004821203504e-06,
+      "loss": 0.8281,
+      "step": 13691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1588401165948503e-06,
+      "loss": 0.7656,
+      "step": 13692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.157675688085543e-06,
+      "loss": 0.875,
+      "step": 13693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1565115357165865e-06,
+      "loss": 0.8242,
+      "step": 13694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1553476595289667e-06,
+      "loss": 0.9922,
+      "step": 13695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1541840595636678e-06,
+      "loss": 0.8086,
+      "step": 13696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1530207358616615e-06,
+      "loss": 0.7812,
+      "step": 13697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1518576884639144e-06,
+      "loss": 0.7422,
+      "step": 13698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1506949174113722e-06,
+      "loss": 0.7969,
+      "step": 13699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.149532422744981e-06,
+      "loss": 0.8281,
+      "step": 13700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1483702045056752e-06,
+      "loss": 1.0234,
+      "step": 13701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1472082627343793e-06,
+      "loss": 0.7109,
+      "step": 13702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1460465974720026e-06,
+      "loss": 0.9453,
+      "step": 13703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.144885208759452e-06,
+      "loss": 0.8945,
+      "step": 13704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1437240966376215e-06,
+      "loss": 0.9414,
+      "step": 13705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1425632611473978e-06,
+      "loss": 0.8945,
+      "step": 13706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1414027023296515e-06,
+      "loss": 0.8789,
+      "step": 13707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1402424202252494e-06,
+      "loss": 0.8477,
+      "step": 13708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.139082414875047e-06,
+      "loss": 0.8594,
+      "step": 13709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.137922686319892e-06,
+      "loss": 0.8594,
+      "step": 13710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136763234600616e-06,
+      "loss": 0.7734,
+      "step": 13711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1356040597580474e-06,
+      "loss": 1.0078,
+      "step": 13712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.134445161833003e-06,
+      "loss": 0.7383,
+      "step": 13713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1332865408662905e-06,
+      "loss": 0.9258,
+      "step": 13714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132128196898704e-06,
+      "loss": 0.8438,
+      "step": 13715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1309701299710305e-06,
+      "loss": 0.7773,
+      "step": 13716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1298123401240478e-06,
+      "loss": 0.8672,
+      "step": 13717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1286548273985288e-06,
+      "loss": 0.9609,
+      "step": 13718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1274975918352226e-06,
+      "loss": 0.875,
+      "step": 13719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1263406334748816e-06,
+      "loss": 0.9102,
+      "step": 13720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1251839523582465e-06,
+      "loss": 0.8242,
+      "step": 13721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.12402754852604e-06,
+      "loss": 0.9023,
+      "step": 13722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122871422018984e-06,
+      "loss": 0.9023,
+      "step": 13723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1217155728777884e-06,
+      "loss": 0.7812,
+      "step": 13724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1205600011431525e-06,
+      "loss": 0.7578,
+      "step": 13725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1194047068557645e-06,
+      "loss": 0.8867,
+      "step": 13726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1182496900563022e-06,
+      "loss": 0.9102,
+      "step": 13727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.117094950785439e-06,
+      "loss": 0.7344,
+      "step": 13728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1159404890838365e-06,
+      "loss": 0.8359,
+      "step": 13729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114786304992139e-06,
+      "loss": 1.0625,
+      "step": 13730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.11363239855099e-06,
+      "loss": 0.8047,
+      "step": 13731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1124787698010218e-06,
+      "loss": 0.875,
+      "step": 13732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1113254187828568e-06,
+      "loss": 0.8008,
+      "step": 13733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1101723455371003e-06,
+      "loss": 0.8789,
+      "step": 13734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1090195501043587e-06,
+      "loss": 0.9727,
+      "step": 13735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1078670325252214e-06,
+      "loss": 0.8008,
+      "step": 13736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1067147928402753e-06,
+      "loss": 0.7695,
+      "step": 13737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1055628310900842e-06,
+      "loss": 0.8789,
+      "step": 13738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.104411147315216e-06,
+      "loss": 0.8164,
+      "step": 13739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1032597415562218e-06,
+      "loss": 0.8945,
+      "step": 13740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1021086138536474e-06,
+      "loss": 0.7383,
+      "step": 13741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1009577642480193e-06,
+      "loss": 0.8164,
+      "step": 13742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099807192779866e-06,
+      "loss": 0.793,
+      "step": 13743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0986568994896985e-06,
+      "loss": 0.8477,
+      "step": 13744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0975068844180237e-06,
+      "loss": 0.8516,
+      "step": 13745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09635714760533e-06,
+      "loss": 0.8633,
+      "step": 13746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0952076890921056e-06,
+      "loss": 0.8555,
+      "step": 13747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0940585089188226e-06,
+      "loss": 0.8867,
+      "step": 13748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0929096071259493e-06,
+      "loss": 0.8438,
+      "step": 13749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.091760983753933e-06,
+      "loss": 0.9141,
+      "step": 13750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0906126388432245e-06,
+      "loss": 0.957,
+      "step": 13751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0894645724342567e-06,
+      "loss": 0.8242,
+      "step": 13752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0883167845674577e-06,
+      "loss": 0.8828,
+      "step": 13753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0871692752832374e-06,
+      "loss": 0.8672,
+      "step": 13754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0860220446220046e-06,
+      "loss": 0.8516,
+      "step": 13755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0848750926241544e-06,
+      "loss": 0.8555,
+      "step": 13756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.083728419330077e-06,
+      "loss": 0.8633,
+      "step": 13757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08258202478014e-06,
+      "loss": 0.9844,
+      "step": 13758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0814359090147153e-06,
+      "loss": 0.8203,
+      "step": 13759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0802900720741592e-06,
+      "loss": 0.8398,
+      "step": 13760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0791445139988197e-06,
+      "loss": 0.8516,
+      "step": 13761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077999234829029e-06,
+      "loss": 0.7812,
+      "step": 13762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0768542346051167e-06,
+      "loss": 0.8438,
+      "step": 13763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0757095133674e-06,
+      "loss": 0.7656,
+      "step": 13764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0745650711561895e-06,
+      "loss": 0.9023,
+      "step": 13765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0734209080117763e-06,
+      "loss": 0.8711,
+      "step": 13766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0722770239744515e-06,
+      "loss": 0.9141,
+      "step": 13767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0711334190844934e-06,
+      "loss": 0.9023,
+      "step": 13768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0699900933821724e-06,
+      "loss": 0.8828,
+      "step": 13769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0688470469077425e-06,
+      "loss": 0.7578,
+      "step": 13770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067704279701452e-06,
+      "loss": 0.8008,
+      "step": 13771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.066561791803543e-06,
+      "loss": 0.9023,
+      "step": 13772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0654195832542434e-06,
+      "loss": 0.8867,
+      "step": 13773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0642776540937703e-06,
+      "loss": 0.8281,
+      "step": 13774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0631360043623327e-06,
+      "loss": 0.8359,
+      "step": 13775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.061994634100131e-06,
+      "loss": 0.8477,
+      "step": 13776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0608535433473565e-06,
+      "loss": 0.8164,
+      "step": 13777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0597127321441857e-06,
+      "loss": 1.0156,
+      "step": 13778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0585722005307873e-06,
+      "loss": 0.7852,
+      "step": 13779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057431948547327e-06,
+      "loss": 0.8203,
+      "step": 13780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0562919762339495e-06,
+      "loss": 0.8164,
+      "step": 13781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055152283630795e-06,
+      "loss": 0.8398,
+      "step": 13782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.054012870777996e-06,
+      "loss": 0.7773,
+      "step": 13783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052873737715675e-06,
+      "loss": 0.7734,
+      "step": 13784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0517348844839367e-06,
+      "loss": 1.0234,
+      "step": 13785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0505963111228864e-06,
+      "loss": 0.8398,
+      "step": 13786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0494580176726133e-06,
+      "loss": 0.7188,
+      "step": 13787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0483200041732022e-06,
+      "loss": 0.7773,
+      "step": 13788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047182270664717e-06,
+      "loss": 0.8555,
+      "step": 13789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046044817187225e-06,
+      "loss": 0.9844,
+      "step": 13790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0449076437807746e-06,
+      "loss": 0.8945,
+      "step": 13791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043770750485412e-06,
+      "loss": 1.0312,
+      "step": 13792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0426341373411618e-06,
+      "loss": 0.8086,
+      "step": 13793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0414978043880507e-06,
+      "loss": 0.8359,
+      "step": 13794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04036175166609e-06,
+      "loss": 0.8477,
+      "step": 13795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0392259792152846e-06,
+      "loss": 0.793,
+      "step": 13796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03809048707562e-06,
+      "loss": 0.9766,
+      "step": 13797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0369552752870834e-06,
+      "loss": 0.8164,
+      "step": 13798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035820343889646e-06,
+      "loss": 0.9805,
+      "step": 13799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.034685692923274e-06,
+      "loss": 0.7812,
+      "step": 13800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033551322427915e-06,
+      "loss": 0.918,
+      "step": 13801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0324172324435133e-06,
+      "loss": 0.9336,
+      "step": 13802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.031283423010003e-06,
+      "loss": 0.8789,
+      "step": 13803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030149894167309e-06,
+      "loss": 0.8516,
+      "step": 13804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.029016645955342e-06,
+      "loss": 0.7188,
+      "step": 13805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0278836784140043e-06,
+      "loss": 0.832,
+      "step": 13806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.026750991583192e-06,
+      "loss": 0.8672,
+      "step": 13807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.02561858550279e-06,
+      "loss": 0.8516,
+      "step": 13808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.024486460212667e-06,
+      "loss": 0.9219,
+      "step": 13809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0233546157526906e-06,
+      "loss": 0.918,
+      "step": 13810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.022223052162714e-06,
+      "loss": 0.875,
+      "step": 13811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0210917694825836e-06,
+      "loss": 0.8594,
+      "step": 13812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0199607677521295e-06,
+      "loss": 0.9453,
+      "step": 13813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0188300470111778e-06,
+      "loss": 0.793,
+      "step": 13814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0176996072995423e-06,
+      "loss": 0.8867,
+      "step": 13815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016569448657032e-06,
+      "loss": 0.8828,
+      "step": 13816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.015439571123434e-06,
+      "loss": 0.7383,
+      "step": 13817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0143099747385377e-06,
+      "loss": 0.7734,
+      "step": 13818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0131806595421166e-06,
+      "loss": 0.7891,
+      "step": 13819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0120516255739384e-06,
+      "loss": 0.8867,
+      "step": 13820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.010922872873754e-06,
+      "loss": 0.7305,
+      "step": 13821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.009794401481311e-06,
+      "loss": 0.6875,
+      "step": 13822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008666211436342e-06,
+      "loss": 0.7383,
+      "step": 13823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0075383027785788e-06,
+      "loss": 0.7461,
+      "step": 13824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0064106755477286e-06,
+      "loss": 0.7344,
+      "step": 13825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0052833297835005e-06,
+      "loss": 0.7969,
+      "step": 13826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.004156265525591e-06,
+      "loss": 0.8789,
+      "step": 13827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0030294828136867e-06,
+      "loss": 0.8633,
+      "step": 13828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.001902981687459e-06,
+      "loss": 0.9609,
+      "step": 13829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.000776762186577e-06,
+      "loss": 0.8984,
+      "step": 13830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999650824350695e-06,
+      "loss": 0.875,
+      "step": 13831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.998525168219464e-06,
+      "loss": 0.9414,
+      "step": 13832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997399793832512e-06,
+      "loss": 0.8945,
+      "step": 13833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9962747012294704e-06,
+      "loss": 0.8047,
+      "step": 13834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9951498904499554e-06,
+      "loss": 0.7812,
+      "step": 13835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9940253615335713e-06,
+      "loss": 0.8828,
+      "step": 13836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992901114519915e-06,
+      "loss": 0.7773,
+      "step": 13837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9917771494485727e-06,
+      "loss": 0.8008,
+      "step": 13838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990653466359125e-06,
+      "loss": 0.7773,
+      "step": 13839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.989530065291132e-06,
+      "loss": 0.7422,
+      "step": 13840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9884069462841536e-06,
+      "loss": 0.8516,
+      "step": 13841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9872841093777363e-06,
+      "loss": 0.7969,
+      "step": 13842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98616155461142e-06,
+      "loss": 0.8555,
+      "step": 13843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9850392820247266e-06,
+      "loss": 0.9414,
+      "step": 13844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9839172916571747e-06,
+      "loss": 0.8164,
+      "step": 13845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9827955835482714e-06,
+      "loss": 0.8242,
+      "step": 13846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9816741577375177e-06,
+      "loss": 0.7695,
+      "step": 13847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9805530142643935e-06,
+      "loss": 0.8164,
+      "step": 13848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97943215316838e-06,
+      "loss": 0.8711,
+      "step": 13849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9783115744889447e-06,
+      "loss": 0.8477,
+      "step": 13850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9771912782655455e-06,
+      "loss": 0.8242,
+      "step": 13851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.976071264537628e-06,
+      "loss": 0.75,
+      "step": 13852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.974951533344629e-06,
+      "loss": 0.8789,
+      "step": 13853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9738320847259762e-06,
+      "loss": 0.8906,
+      "step": 13854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9727129187210925e-06,
+      "loss": 0.8281,
+      "step": 13855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9715940353693775e-06,
+      "loss": 0.8281,
+      "step": 13856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970475434710233e-06,
+      "loss": 0.832,
+      "step": 13857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969357116783045e-06,
+      "loss": 0.8359,
+      "step": 13858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9682390816271947e-06,
+      "loss": 0.8633,
+      "step": 13859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.967121329282046e-06,
+      "loss": 0.9062,
+      "step": 13860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9660038597869567e-06,
+      "loss": 0.9062,
+      "step": 13861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9648866731812768e-06,
+      "loss": 0.8555,
+      "step": 13862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9637697695043455e-06,
+      "loss": 0.8828,
+      "step": 13863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962653148795487e-06,
+      "loss": 0.8867,
+      "step": 13864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961536811094019e-06,
+      "loss": 0.7734,
+      "step": 13865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960420756439253e-06,
+      "loss": 0.8711,
+      "step": 13866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.959304984870487e-06,
+      "loss": 0.8008,
+      "step": 13867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9581894964270043e-06,
+      "loss": 0.7812,
+      "step": 13868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9570742911480867e-06,
+      "loss": 0.8203,
+      "step": 13869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9559593690730006e-06,
+      "loss": 0.8555,
+      "step": 13870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9548447302410077e-06,
+      "loss": 0.9648,
+      "step": 13871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.953730374691353e-06,
+      "loss": 0.8672,
+      "step": 13872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9526163024632726e-06,
+      "loss": 1.0,
+      "step": 13873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9515025135959986e-06,
+      "loss": 0.793,
+      "step": 13874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95038900812875e-06,
+      "loss": 0.9453,
+      "step": 13875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9492757861007314e-06,
+      "loss": 0.8789,
+      "step": 13876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9481628475511417e-06,
+      "loss": 0.8438,
+      "step": 13877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9470501925191708e-06,
+      "loss": 0.9688,
+      "step": 13878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9459378210439984e-06,
+      "loss": 0.793,
+      "step": 13879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9448257331647878e-06,
+      "loss": 0.832,
+      "step": 13880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9437139289207018e-06,
+      "loss": 0.7969,
+      "step": 13881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9426024083508865e-06,
+      "loss": 0.8789,
+      "step": 13882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941491171494483e-06,
+      "loss": 0.9219,
+      "step": 13883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940380218390615e-06,
+      "loss": 0.7734,
+      "step": 13884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.939269549078404e-06,
+      "loss": 0.7773,
+      "step": 13885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938159163596959e-06,
+      "loss": 0.8242,
+      "step": 13886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937049061985378e-06,
+      "loss": 0.8633,
+      "step": 13887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9359392442827472e-06,
+      "loss": 0.8867,
+      "step": 13888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9348297105281455e-06,
+      "loss": 0.8281,
+      "step": 13889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9337204607606465e-06,
+      "loss": 0.8438,
+      "step": 13890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9326114950193e-06,
+      "loss": 0.8828,
+      "step": 13891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.93150281334316e-06,
+      "loss": 0.8594,
+      "step": 13892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.930394415771263e-06,
+      "loss": 0.9062,
+      "step": 13893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9292863023426413e-06,
+      "loss": 0.8555,
+      "step": 13894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.928178473096307e-06,
+      "loss": 0.707,
+      "step": 13895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927070928071272e-06,
+      "loss": 0.8594,
+      "step": 13896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925963667306533e-06,
+      "loss": 0.7852,
+      "step": 13897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9248566908410836e-06,
+      "loss": 1.0078,
+      "step": 13898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.923749998713894e-06,
+      "loss": 0.9102,
+      "step": 13899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9226435909639383e-06,
+      "loss": 0.75,
+      "step": 13900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9215374676301724e-06,
+      "loss": 0.8594,
+      "step": 13901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920431628751548e-06,
+      "loss": 1.0156,
+      "step": 13902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9193260743669973e-06,
+      "loss": 0.8125,
+      "step": 13903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9182208045154536e-06,
+      "loss": 0.8477,
+      "step": 13904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917115819235833e-06,
+      "loss": 0.832,
+      "step": 13905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916011118567046e-06,
+      "loss": 0.9531,
+      "step": 13906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914906702547987e-06,
+      "loss": 0.7812,
+      "step": 13907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913802571217548e-06,
+      "loss": 0.832,
+      "step": 13908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9126987246146032e-06,
+      "loss": 0.8438,
+      "step": 13909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911595162778027e-06,
+      "loss": 0.832,
+      "step": 13910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9104918857466713e-06,
+      "loss": 0.8594,
+      "step": 13911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9093888935593854e-06,
+      "loss": 0.9258,
+      "step": 13912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9082861862550095e-06,
+      "loss": 0.8633,
+      "step": 13913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9071837638723722e-06,
+      "loss": 0.8008,
+      "step": 13914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.906081626450288e-06,
+      "loss": 0.7969,
+      "step": 13915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9049797740275655e-06,
+      "loss": 0.8594,
+      "step": 13916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.903878206643005e-06,
+      "loss": 0.7617,
+      "step": 13917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9027769243353956e-06,
+      "loss": 0.9219,
+      "step": 13918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9016759271435104e-06,
+      "loss": 0.9727,
+      "step": 13919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9005752151061185e-06,
+      "loss": 0.8672,
+      "step": 13920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8994747882619824e-06,
+      "loss": 0.7773,
+      "step": 13921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.898374646649843e-06,
+      "loss": 0.957,
+      "step": 13922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.897274790308442e-06,
+      "loss": 0.8164,
+      "step": 13923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.896175219276506e-06,
+      "loss": 0.8086,
+      "step": 13924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8950759335927548e-06,
+      "loss": 0.832,
+      "step": 13925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8939769332958913e-06,
+      "loss": 0.7852,
+      "step": 13926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8928782184246164e-06,
+      "loss": 0.7734,
+      "step": 13927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8917797890176158e-06,
+      "loss": 0.793,
+      "step": 13928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8906816451135712e-06,
+      "loss": 0.9141,
+      "step": 13929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8895837867511435e-06,
+      "loss": 0.9258,
+      "step": 13930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8884862139689941e-06,
+      "loss": 0.7539,
+      "step": 13931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8873889268057689e-06,
+      "loss": 0.7344,
+      "step": 13932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8862919253001077e-06,
+      "loss": 0.7305,
+      "step": 13933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8851952094906322e-06,
+      "loss": 0.8789,
+      "step": 13934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8840987794159637e-06,
+      "loss": 0.875,
+      "step": 13935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8830026351147102e-06,
+      "loss": 0.8086,
+      "step": 13936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.881906776625464e-06,
+      "loss": 0.7812,
+      "step": 13937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8808112039868143e-06,
+      "loss": 0.7266,
+      "step": 13938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.879715917237338e-06,
+      "loss": 0.8242,
+      "step": 13939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8786209164156056e-06,
+      "loss": 0.7812,
+      "step": 13940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8775262015601659e-06,
+      "loss": 0.8633,
+      "step": 13941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8764317727095705e-06,
+      "loss": 0.9727,
+      "step": 13942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875337629902355e-06,
+      "loss": 0.8047,
+      "step": 13943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874243773177049e-06,
+      "loss": 0.9141,
+      "step": 13944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8731502025721626e-06,
+      "loss": 0.8008,
+      "step": 13945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.872056918126206e-06,
+      "loss": 0.8984,
+      "step": 13946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8709639198776774e-06,
+      "loss": 0.875,
+      "step": 13947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8698712078650582e-06,
+      "loss": 0.8516,
+      "step": 13948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8687787821268255e-06,
+      "loss": 0.7695,
+      "step": 13949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8676866427014472e-06,
+      "loss": 0.8359,
+      "step": 13950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8665947896273807e-06,
+      "loss": 0.7227,
+      "step": 13951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.865503222943067e-06,
+      "loss": 0.8281,
+      "step": 13952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8644119426869456e-06,
+      "loss": 0.8086,
+      "step": 13953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.863320948897439e-06,
+      "loss": 0.8281,
+      "step": 13954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.862230241612969e-06,
+      "loss": 0.8125,
+      "step": 13955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8611398208719333e-06,
+      "loss": 0.8867,
+      "step": 13956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8600496867127316e-06,
+      "loss": 0.9453,
+      "step": 13957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8589598391737485e-06,
+      "loss": 0.7617,
+      "step": 13958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8578702782933612e-06,
+      "loss": 0.8984,
+      "step": 13959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.856781004109931e-06,
+      "loss": 0.8828,
+      "step": 13960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8556920166618143e-06,
+      "loss": 0.8281,
+      "step": 13961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8546033159873588e-06,
+      "loss": 0.8984,
+      "step": 13962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8535149021248943e-06,
+      "loss": 0.7578,
+      "step": 13963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8524267751127488e-06,
+      "loss": 0.9844,
+      "step": 13964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8513389349892364e-06,
+      "loss": 0.8438,
+      "step": 13965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8502513817926638e-06,
+      "loss": 0.9023,
+      "step": 13966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8491641155613204e-06,
+      "loss": 0.9219,
+      "step": 13967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8480771363334938e-06,
+      "loss": 0.7852,
+      "step": 13968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8469904441474573e-06,
+      "loss": 0.8711,
+      "step": 13969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8459040390414785e-06,
+      "loss": 0.9141,
+      "step": 13970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8448179210538065e-06,
+      "loss": 0.7617,
+      "step": 13971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8437320902226862e-06,
+      "loss": 0.8359,
+      "step": 13972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8426465465863518e-06,
+      "loss": 0.8438,
+      "step": 13973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8415612901830303e-06,
+      "loss": 0.9102,
+      "step": 13974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8404763210509302e-06,
+      "loss": 0.8359,
+      "step": 13975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8393916392282574e-06,
+      "loss": 0.8477,
+      "step": 13976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838307244753207e-06,
+      "loss": 0.7305,
+      "step": 13977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837223137663957e-06,
+      "loss": 0.8789,
+      "step": 13978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8361393179986841e-06,
+      "loss": 0.9766,
+      "step": 13979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.835055785795551e-06,
+      "loss": 0.9688,
+      "step": 13980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8339725410927134e-06,
+      "loss": 0.8047,
+      "step": 13981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.832889583928308e-06,
+      "loss": 0.8594,
+      "step": 13982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8318069143404694e-06,
+      "loss": 0.9375,
+      "step": 13983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8307245323673217e-06,
+      "loss": 0.8125,
+      "step": 13984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.829642438046979e-06,
+      "loss": 0.8633,
+      "step": 13985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8285606314175385e-06,
+      "loss": 0.7852,
+      "step": 13986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8274791125170955e-06,
+      "loss": 0.7656,
+      "step": 13987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82639788138373e-06,
+      "loss": 0.8633,
+      "step": 13988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8253169380555191e-06,
+      "loss": 0.8633,
+      "step": 13989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8242362825705173e-06,
+      "loss": 0.8242,
+      "step": 13990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8231559149667788e-06,
+      "loss": 0.8984,
+      "step": 13991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8220758352823464e-06,
+      "loss": 0.7891,
+      "step": 13992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8209960435552531e-06,
+      "loss": 0.7891,
+      "step": 13993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8199165398235141e-06,
+      "loss": 0.7695,
+      "step": 13994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8188373241251444e-06,
+      "loss": 0.7383,
+      "step": 13995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8177583964981437e-06,
+      "loss": 0.8203,
+      "step": 13996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8166797569805062e-06,
+      "loss": 0.8516,
+      "step": 13997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8156014056102067e-06,
+      "loss": 0.8867,
+      "step": 13998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8145233424252185e-06,
+      "loss": 0.8672,
+      "step": 13999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.813445567463501e-06,
+      "loss": 0.7539,
+      "step": 14000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8123680807630084e-06,
+      "loss": 0.7266,
+      "step": 14001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.811290882361675e-06,
+      "loss": 0.8594,
+      "step": 14002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8102139722974321e-06,
+      "loss": 0.8125,
+      "step": 14003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8091373506082033e-06,
+      "loss": 0.8359,
+      "step": 14004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8080610173318936e-06,
+      "loss": 0.75,
+      "step": 14005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8069849725064026e-06,
+      "loss": 0.707,
+      "step": 14006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8059092161696202e-06,
+      "loss": 0.7969,
+      "step": 14007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8048337483594292e-06,
+      "loss": 0.8438,
+      "step": 14008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8037585691136927e-06,
+      "loss": 0.9375,
+      "step": 14009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8026836784702716e-06,
+      "loss": 0.7656,
+      "step": 14010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8016090764670158e-06,
+      "loss": 1.0156,
+      "step": 14011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8005347631417636e-06,
+      "loss": 0.8359,
+      "step": 14012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7994607385323415e-06,
+      "loss": 0.832,
+      "step": 14013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.798387002676567e-06,
+      "loss": 0.7344,
+      "step": 14014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.79731355561225e-06,
+      "loss": 0.8438,
+      "step": 14015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7962403973771903e-06,
+      "loss": 0.7656,
+      "step": 14016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7951675280091697e-06,
+      "loss": 0.9688,
+      "step": 14017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.794094947545969e-06,
+      "loss": 0.8008,
+      "step": 14018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.793022656025356e-06,
+      "loss": 0.7969,
+      "step": 14019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7919506534850883e-06,
+      "loss": 0.8711,
+      "step": 14020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.790878939962909e-06,
+      "loss": 0.8047,
+      "step": 14021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7898075154965566e-06,
+      "loss": 0.8086,
+      "step": 14022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7887363801237589e-06,
+      "loss": 0.832,
+      "step": 14023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7876655338822336e-06,
+      "loss": 0.8125,
+      "step": 14024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7865949768096813e-06,
+      "loss": 0.75,
+      "step": 14025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7855247089438022e-06,
+      "loss": 0.8672,
+      "step": 14026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7844547303222813e-06,
+      "loss": 0.9414,
+      "step": 14027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7833850409827968e-06,
+      "loss": 0.7656,
+      "step": 14028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7823156409630083e-06,
+      "loss": 0.7852,
+      "step": 14029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7812465303005743e-06,
+      "loss": 0.8008,
+      "step": 14030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7801777090331396e-06,
+      "loss": 0.8438,
+      "step": 14031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7791091771983427e-06,
+      "loss": 0.9102,
+      "step": 14032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7780409348338012e-06,
+      "loss": 0.8672,
+      "step": 14033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776972981977133e-06,
+      "loss": 0.9375,
+      "step": 14034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7759053186659424e-06,
+      "loss": 0.7734,
+      "step": 14035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7748379449378261e-06,
+      "loss": 0.7383,
+      "step": 14036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7737708608303627e-06,
+      "loss": 0.7617,
+      "step": 14037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.772704066381129e-06,
+      "loss": 0.8125,
+      "step": 14038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7716375616276881e-06,
+      "loss": 0.8359,
+      "step": 14039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7705713466075959e-06,
+      "loss": 0.8125,
+      "step": 14040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7695054213583896e-06,
+      "loss": 0.8828,
+      "step": 14041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7684397859176061e-06,
+      "loss": 0.7891,
+      "step": 14042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7673744403227666e-06,
+      "loss": 0.7969,
+      "step": 14043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7663093846113866e-06,
+      "loss": 0.9258,
+      "step": 14044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7652446188209637e-06,
+      "loss": 0.8047,
+      "step": 14045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7641801429889916e-06,
+      "loss": 0.9688,
+      "step": 14046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7631159571529533e-06,
+      "loss": 0.7891,
+      "step": 14047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7620520613503223e-06,
+      "loss": 0.7891,
+      "step": 14048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7609884556185552e-06,
+      "loss": 0.7617,
+      "step": 14049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7599251399951056e-06,
+      "loss": 0.8203,
+      "step": 14050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7588621145174145e-06,
+      "loss": 0.8828,
+      "step": 14051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7577993792229153e-06,
+      "loss": 0.8203,
+      "step": 14052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7567369341490225e-06,
+      "loss": 0.7969,
+      "step": 14053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7556747793331508e-06,
+      "loss": 0.8359,
+      "step": 14054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7546129148126988e-06,
+      "loss": 0.8828,
+      "step": 14055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7535513406250592e-06,
+      "loss": 0.9297,
+      "step": 14056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7524900568076075e-06,
+      "loss": 0.7305,
+      "step": 14057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.751429063397715e-06,
+      "loss": 0.8242,
+      "step": 14058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7503683604327426e-06,
+      "loss": 0.8047,
+      "step": 14059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7493079479500362e-06,
+      "loss": 0.9141,
+      "step": 14060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.748247825986935e-06,
+      "loss": 0.7578,
+      "step": 14061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.747187994580769e-06,
+      "loss": 0.832,
+      "step": 14062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7461284537688595e-06,
+      "loss": 0.875,
+      "step": 14063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7450692035885075e-06,
+      "loss": 0.8867,
+      "step": 14064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7440102440770146e-06,
+      "loss": 0.9531,
+      "step": 14065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7429515752716675e-06,
+      "loss": 0.8164,
+      "step": 14066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7418931972097474e-06,
+      "loss": 0.7227,
+      "step": 14067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.740835109928516e-06,
+      "loss": 0.8789,
+      "step": 14068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7397773134652329e-06,
+      "loss": 0.8711,
+      "step": 14069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7387198078571433e-06,
+      "loss": 0.7383,
+      "step": 14070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737662593141488e-06,
+      "loss": 0.9258,
+      "step": 14071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7366056693554868e-06,
+      "loss": 0.7461,
+      "step": 14072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7355490365363592e-06,
+      "loss": 0.8828,
+      "step": 14073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7344926947213092e-06,
+      "loss": 0.8203,
+      "step": 14074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.733436643947537e-06,
+      "loss": 1.0469,
+      "step": 14075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7323808842522206e-06,
+      "loss": 0.7891,
+      "step": 14076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7313254156725389e-06,
+      "loss": 0.8906,
+      "step": 14077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7302702382456571e-06,
+      "loss": 0.8281,
+      "step": 14078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7292153520087296e-06,
+      "loss": 0.9453,
+      "step": 14079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7281607569988979e-06,
+      "loss": 0.8047,
+      "step": 14080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7271064532532977e-06,
+      "loss": 0.9648,
+      "step": 14081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7260524408090517e-06,
+      "loss": 0.8359,
+      "step": 14082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.724998719703278e-06,
+      "loss": 0.8086,
+      "step": 14083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7239452899730725e-06,
+      "loss": 0.793,
+      "step": 14084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.722892151655532e-06,
+      "loss": 0.9609,
+      "step": 14085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.721839304787738e-06,
+      "loss": 0.8359,
+      "step": 14086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7207867494067664e-06,
+      "loss": 0.8555,
+      "step": 14087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7197344855496745e-06,
+      "loss": 0.918,
+      "step": 14088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7186825132535156e-06,
+      "loss": 0.8047,
+      "step": 14089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7176308325553314e-06,
+      "loss": 0.8672,
+      "step": 14090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7165794434921567e-06,
+      "loss": 0.8594,
+      "step": 14091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7155283461010064e-06,
+      "loss": 0.8008,
+      "step": 14092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7144775404188952e-06,
+      "loss": 0.793,
+      "step": 14093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7134270264828212e-06,
+      "loss": 0.8242,
+      "step": 14094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7123768043297794e-06,
+      "loss": 0.8867,
+      "step": 14095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7113268739967437e-06,
+      "loss": 0.8477,
+      "step": 14096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7102772355206865e-06,
+      "loss": 0.7148,
+      "step": 14097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7092278889385661e-06,
+      "loss": 0.8477,
+      "step": 14098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081788342873361e-06,
+      "loss": 0.875,
+      "step": 14099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071300716039297e-06,
+      "loss": 0.8906,
+      "step": 14100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060816009252768e-06,
+      "loss": 0.7422,
+      "step": 14101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050334222882969e-06,
+      "loss": 0.8203,
+      "step": 14102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7039855357299006e-06,
+      "loss": 0.8555,
+      "step": 14103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029379412869796e-06,
+      "loss": 1.0703,
+      "step": 14104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701890638996424e-06,
+      "loss": 0.7148,
+      "step": 14105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7008436288951124e-06,
+      "loss": 0.832,
+      "step": 14106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699796911019912e-06,
+      "loss": 0.7031,
+      "step": 14107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6987504854076753e-06,
+      "loss": 0.8594,
+      "step": 14108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6977043520952518e-06,
+      "loss": 0.9102,
+      "step": 14109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6966585111194756e-06,
+      "loss": 0.7734,
+      "step": 14110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6956129625171768e-06,
+      "loss": 0.7852,
+      "step": 14111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6945677063251654e-06,
+      "loss": 0.8711,
+      "step": 14112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6935227425802481e-06,
+      "loss": 0.9141,
+      "step": 14113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6924780713192235e-06,
+      "loss": 0.8438,
+      "step": 14114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6914336925788687e-06,
+      "loss": 0.7773,
+      "step": 14115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6903896063959635e-06,
+      "loss": 0.8477,
+      "step": 14116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6893458128072692e-06,
+      "loss": 0.9375,
+      "step": 14117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6883023118495434e-06,
+      "loss": 0.707,
+      "step": 14118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6872591035595242e-06,
+      "loss": 0.7773,
+      "step": 14119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6862161879739469e-06,
+      "loss": 0.8711,
+      "step": 14120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.685173565129533e-06,
+      "loss": 0.832,
+      "step": 14121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684131235062999e-06,
+      "loss": 0.9688,
+      "step": 14122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.683089197811042e-06,
+      "loss": 0.918,
+      "step": 14123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6820474534103549e-06,
+      "loss": 0.8633,
+      "step": 14124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6810060018976204e-06,
+      "loss": 0.8672,
+      "step": 14125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6799648433095116e-06,
+      "loss": 0.8203,
+      "step": 14126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6789239776826838e-06,
+      "loss": 0.7305,
+      "step": 14127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6778834050537907e-06,
+      "loss": 0.8633,
+      "step": 14128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.676843125459472e-06,
+      "loss": 0.9023,
+      "step": 14129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6758031389363605e-06,
+      "loss": 0.9102,
+      "step": 14130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6747634455210704e-06,
+      "loss": 0.8984,
+      "step": 14131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6737240452502136e-06,
+      "loss": 0.8242,
+      "step": 14132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6726849381603883e-06,
+      "loss": 0.7773,
+      "step": 14133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6716461242881865e-06,
+      "loss": 0.8789,
+      "step": 14134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6706076036701812e-06,
+      "loss": 0.8477,
+      "step": 14135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6695693763429433e-06,
+      "loss": 0.7969,
+      "step": 14136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6685314423430288e-06,
+      "loss": 0.8477,
+      "step": 14137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6674938017069898e-06,
+      "loss": 0.707,
+      "step": 14138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6664564544713557e-06,
+      "loss": 0.7734,
+      "step": 14139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6654194006726577e-06,
+      "loss": 0.8398,
+      "step": 14140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6643826403474106e-06,
+      "loss": 0.7461,
+      "step": 14141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6633461735321232e-06,
+      "loss": 1.0469,
+      "step": 14142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6623100002632864e-06,
+      "loss": 0.9727,
+      "step": 14143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6612741205773874e-06,
+      "loss": 0.918,
+      "step": 14144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660238534510903e-06,
+      "loss": 0.8203,
+      "step": 14145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6592032421002978e-06,
+      "loss": 0.8359,
+      "step": 14146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6581682433820224e-06,
+      "loss": 0.8125,
+      "step": 14147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6571335383925224e-06,
+      "loss": 0.8945,
+      "step": 14148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6560991271682324e-06,
+      "loss": 0.8633,
+      "step": 14149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6550650097455777e-06,
+      "loss": 0.8906,
+      "step": 14150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.654031186160966e-06,
+      "loss": 0.9258,
+      "step": 14151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6529976564508022e-06,
+      "loss": 0.7617,
+      "step": 14152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6519644206514784e-06,
+      "loss": 0.707,
+      "step": 14153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.650931478799379e-06,
+      "loss": 0.9336,
+      "step": 14154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6498988309308717e-06,
+      "loss": 0.8516,
+      "step": 14155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.648866477082318e-06,
+      "loss": 0.9297,
+      "step": 14156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6478344172900706e-06,
+      "loss": 0.8281,
+      "step": 14157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6468026515904712e-06,
+      "loss": 0.8672,
+      "step": 14158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645771180019845e-06,
+      "loss": 0.8477,
+      "step": 14159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6447400026145143e-06,
+      "loss": 0.875,
+      "step": 14160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.643709119410789e-06,
+      "loss": 0.8359,
+      "step": 14161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.642678530444971e-06,
+      "loss": 0.9297,
+      "step": 14162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6416482357533414e-06,
+      "loss": 0.7695,
+      "step": 14163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6406182353721844e-06,
+      "loss": 0.8242,
+      "step": 14164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6395885293377655e-06,
+      "loss": 0.8242,
+      "step": 14165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6385591176863448e-06,
+      "loss": 0.8203,
+      "step": 14166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6375300004541661e-06,
+      "loss": 0.9258,
+      "step": 14167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6365011776774663e-06,
+      "loss": 0.8828,
+      "step": 14168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6354726493924745e-06,
+      "loss": 0.7812,
+      "step": 14169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6344444156354077e-06,
+      "loss": 0.9141,
+      "step": 14170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6334164764424665e-06,
+      "loss": 0.9219,
+      "step": 14171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6323888318498494e-06,
+      "loss": 0.9609,
+      "step": 14172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.631361481893744e-06,
+      "loss": 0.8047,
+      "step": 14173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6303344266103195e-06,
+      "loss": 0.9297,
+      "step": 14174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6293076660357421e-06,
+      "loss": 0.8633,
+      "step": 14175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6282812002061665e-06,
+      "loss": 0.8984,
+      "step": 14176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6272550291577371e-06,
+      "loss": 0.8438,
+      "step": 14177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6262291529265838e-06,
+      "loss": 0.8242,
+      "step": 14178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6252035715488323e-06,
+      "loss": 0.8164,
+      "step": 14179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6241782850605925e-06,
+      "loss": 0.7891,
+      "step": 14180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6231532934979699e-06,
+      "loss": 0.8359,
+      "step": 14181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6221285968970512e-06,
+      "loss": 0.7539,
+      "step": 14182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.62110419529392e-06,
+      "loss": 0.9453,
+      "step": 14183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6200800887246483e-06,
+      "loss": 0.8516,
+      "step": 14184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.619056277225296e-06,
+      "loss": 0.8633,
+      "step": 14185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6180327608319103e-06,
+      "loss": 0.7812,
+      "step": 14186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.617009539580533e-06,
+      "loss": 0.7344,
+      "step": 14187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6159866135071933e-06,
+      "loss": 0.7969,
+      "step": 14188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6149639826479124e-06,
+      "loss": 0.7344,
+      "step": 14189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6139416470386936e-06,
+      "loss": 0.7461,
+      "step": 14190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6129196067155372e-06,
+      "loss": 0.7695,
+      "step": 14191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.611897861714431e-06,
+      "loss": 0.7383,
+      "step": 14192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.610876412071355e-06,
+      "loss": 0.793,
+      "step": 14193
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6098552578222715e-06,
+      "loss": 0.8555,
+      "step": 14194
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6088343990031375e-06,
+      "loss": 0.8828,
+      "step": 14195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6078138356499006e-06,
+      "loss": 0.8711,
+      "step": 14196
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.606793567798499e-06,
+      "loss": 0.8359,
+      "step": 14197
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6057735954848518e-06,
+      "loss": 0.7539,
+      "step": 14198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.604753918744877e-06,
+      "loss": 0.8594,
+      "step": 14199
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6037345376144796e-06,
+      "loss": 0.8398,
+      "step": 14200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6027154521295552e-06,
+      "loss": 0.8789,
+      "step": 14201
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6016966623259821e-06,
+      "loss": 0.9102,
+      "step": 14202
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6006781682396366e-06,
+      "loss": 0.8086,
+      "step": 14203
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5996599699063819e-06,
+      "loss": 0.8711,
+      "step": 14204
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5986420673620717e-06,
+      "loss": 0.8789,
+      "step": 14205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.597624460642544e-06,
+      "loss": 0.8281,
+      "step": 14206
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5966071497836322e-06,
+      "loss": 0.8984,
+      "step": 14207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5955901348211567e-06,
+      "loss": 0.8359,
+      "step": 14208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.594573415790932e-06,
+      "loss": 0.875,
+      "step": 14209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.593556992728752e-06,
+      "loss": 0.7148,
+      "step": 14210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592540865670411e-06,
+      "loss": 0.8945,
+      "step": 14211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5915250346516865e-06,
+      "loss": 0.8359,
+      "step": 14212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5905094997083515e-06,
+      "loss": 0.8906,
+      "step": 14213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5894942608761577e-06,
+      "loss": 0.8672,
+      "step": 14214
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5884793181908575e-06,
+      "loss": 0.9102,
+      "step": 14215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.9023,
+      "step": 14216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5864503214038772e-06,
+      "loss": 0.9297,
+      "step": 14217
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.58543626737364e-06,
+      "loss": 0.6953,
+      "step": 14218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5844225096331834e-06,
+      "loss": 1.1094,
+      "step": 14219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5834090482182041e-06,
+      "loss": 0.7656,
+      "step": 14220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5823958831643903e-06,
+      "loss": 0.7383,
+      "step": 14221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5813830145074115e-06,
+      "loss": 0.7852,
+      "step": 14222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5803704422829347e-06,
+      "loss": 0.8125,
+      "step": 14223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5793581665266145e-06,
+      "loss": 0.8398,
+      "step": 14224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5783461872740968e-06,
+      "loss": 0.8516,
+      "step": 14225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.577334504561011e-06,
+      "loss": 0.8945,
+      "step": 14226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.576323118422982e-06,
+      "loss": 0.8906,
+      "step": 14227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5753120288956247e-06,
+      "loss": 0.8711,
+      "step": 14228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5743012360145348e-06,
+      "loss": 0.8633,
+      "step": 14229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5732907398153086e-06,
+      "loss": 0.8086,
+      "step": 14230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.572280540333526e-06,
+      "loss": 0.9688,
+      "step": 14231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5712706376047614e-06,
+      "loss": 0.7734,
+      "step": 14232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5702610316645683e-06,
+      "loss": 0.8672,
+      "step": 14233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5692517225484993e-06,
+      "loss": 0.7969,
+      "step": 14234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5682427102920949e-06,
+      "loss": 0.9336,
+      "step": 14235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5672339949308857e-06,
+      "loss": 0.8945,
+      "step": 14236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5662255765003853e-06,
+      "loss": 0.8359,
+      "step": 14237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5652174550361033e-06,
+      "loss": 0.7656,
+      "step": 14238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.564209630573539e-06,
+      "loss": 0.8203,
+      "step": 14239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5632021031481804e-06,
+      "loss": 0.8008,
+      "step": 14240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5621948727954993e-06,
+      "loss": 0.8945,
+      "step": 14241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5611879395509644e-06,
+      "loss": 0.8242,
+      "step": 14242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5601813034500324e-06,
+      "loss": 0.8164,
+      "step": 14243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5591749645281507e-06,
+      "loss": 1.0312,
+      "step": 14244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5581689228207475e-06,
+      "loss": 0.832,
+      "step": 14245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5571631783632513e-06,
+      "loss": 0.8359,
+      "step": 14246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5561577311910747e-06,
+      "loss": 0.9844,
+      "step": 14247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.555152581339625e-06,
+      "loss": 0.7891,
+      "step": 14248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5541477288442886e-06,
+      "loss": 0.8125,
+      "step": 14249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5531431737404512e-06,
+      "loss": 0.8789,
+      "step": 14250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5521389160634848e-06,
+      "loss": 0.9219,
+      "step": 14251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5511349558487532e-06,
+      "loss": 0.7812,
+      "step": 14252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5501312931316015e-06,
+      "loss": 0.8359,
+      "step": 14253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5491279279473747e-06,
+      "loss": 0.7266,
+      "step": 14254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5481248603314015e-06,
+      "loss": 0.8945,
+      "step": 14255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5471220903190033e-06,
+      "loss": 0.8203,
+      "step": 14256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.546119617945485e-06,
+      "loss": 0.9609,
+      "step": 14257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.545117443246148e-06,
+      "loss": 0.8984,
+      "step": 14258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5441155662562811e-06,
+      "loss": 0.7812,
+      "step": 14259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5431139870111623e-06,
+      "loss": 0.7812,
+      "step": 14260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5421127055460549e-06,
+      "loss": 0.7656,
+      "step": 14261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.541111721896219e-06,
+      "loss": 0.7539,
+      "step": 14262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5401110360968985e-06,
+      "loss": 0.8477,
+      "step": 14263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539110648183334e-06,
+      "loss": 0.9297,
+      "step": 14264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538110558190744e-06,
+      "loss": 1.0,
+      "step": 14265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5371107661543471e-06,
+      "loss": 0.875,
+      "step": 14266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5361112721093486e-06,
+      "loss": 0.7891,
+      "step": 14267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5351120760909388e-06,
+      "loss": 0.8555,
+      "step": 14268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5341131781343022e-06,
+      "loss": 0.9688,
+      "step": 14269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5331145782746116e-06,
+      "loss": 0.7383,
+      "step": 14270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5321162765470322e-06,
+      "loss": 0.7773,
+      "step": 14271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5311182729867102e-06,
+      "loss": 0.8789,
+      "step": 14272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.53012056762879e-06,
+      "loss": 0.9141,
+      "step": 14273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5291231605084024e-06,
+      "loss": 0.8164,
+      "step": 14274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528126051660669e-06,
+      "loss": 0.8633,
+      "step": 14275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.527129241120696e-06,
+      "loss": 0.832,
+      "step": 14276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526132728923585e-06,
+      "loss": 0.7773,
+      "step": 14277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5251365151044229e-06,
+      "loss": 0.8594,
+      "step": 14278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5241405996982928e-06,
+      "loss": 0.793,
+      "step": 14279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5231449827402566e-06,
+      "loss": 0.8789,
+      "step": 14280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5221496642653733e-06,
+      "loss": 0.8242,
+      "step": 14281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5211546443086922e-06,
+      "loss": 0.7852,
+      "step": 14282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5201599229052455e-06,
+      "loss": 0.8828,
+      "step": 14283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5191655000900595e-06,
+      "loss": 0.8555,
+      "step": 14284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.518171375898152e-06,
+      "loss": 0.8438,
+      "step": 14285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.517177550364528e-06,
+      "loss": 0.7305,
+      "step": 14286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5161840235241787e-06,
+      "loss": 0.9219,
+      "step": 14287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5151907954120882e-06,
+      "loss": 0.7539,
+      "step": 14288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5141978660632295e-06,
+      "loss": 0.8789,
+      "step": 14289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5132052355125693e-06,
+      "loss": 0.8047,
+      "step": 14290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5122129037950529e-06,
+      "loss": 0.7812,
+      "step": 14291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5112208709456267e-06,
+      "loss": 0.7734,
+      "step": 14292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5102291369992217e-06,
+      "loss": 0.8164,
+      "step": 14293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5092377019907545e-06,
+      "loss": 0.7422,
+      "step": 14294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5082465659551381e-06,
+      "loss": 0.8008,
+      "step": 14295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5072557289272716e-06,
+      "loss": 0.7812,
+      "step": 14296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5062651909420444e-06,
+      "loss": 0.8711,
+      "step": 14297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5052749520343334e-06,
+      "loss": 0.8633,
+      "step": 14298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5042850122390062e-06,
+      "loss": 0.8594,
+      "step": 14299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5032953715909216e-06,
+      "loss": 0.9023,
+      "step": 14300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.502306030124928e-06,
+      "loss": 0.8242,
+      "step": 14301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.501316987875857e-06,
+      "loss": 0.7422,
+      "step": 14302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5003282448785373e-06,
+      "loss": 0.7852,
+      "step": 14303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.499339801167784e-06,
+      "loss": 0.8594,
+      "step": 14304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4983516567784028e-06,
+      "loss": 0.8438,
+      "step": 14305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4973638117451838e-06,
+      "loss": 0.9336,
+      "step": 14306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4963762661029135e-06,
+      "loss": 0.8828,
+      "step": 14307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4953890198863664e-06,
+      "loss": 0.8555,
+      "step": 14308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4944020731303e-06,
+      "loss": 0.8008,
+      "step": 14309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49341542586947e-06,
+      "loss": 0.8203,
+      "step": 14310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4924290781386175e-06,
+      "loss": 0.7852,
+      "step": 14311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4914430299724747e-06,
+      "loss": 0.9453,
+      "step": 14312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4904572814057573e-06,
+      "loss": 0.8203,
+      "step": 14313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4894718324731782e-06,
+      "loss": 0.8477,
+      "step": 14314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4884866832094358e-06,
+      "loss": 1.0781,
+      "step": 14315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4875018336492219e-06,
+      "loss": 0.8125,
+      "step": 14316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.486517283827209e-06,
+      "loss": 0.9258,
+      "step": 14317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4855330337780672e-06,
+      "loss": 0.8438,
+      "step": 14318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4845490835364529e-06,
+      "loss": 0.7773,
+      "step": 14319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4835654331370164e-06,
+      "loss": 0.9609,
+      "step": 14320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4825820826143877e-06,
+      "loss": 0.8555,
+      "step": 14321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4815990320031959e-06,
+      "loss": 0.7891,
+      "step": 14322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4806162813380554e-06,
+      "loss": 0.8164,
+      "step": 14323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4796338306535673e-06,
+      "loss": 0.8281,
+      "step": 14324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4786516799843286e-06,
+      "loss": 0.7695,
+      "step": 14325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4776698293649194e-06,
+      "loss": 0.8164,
+      "step": 14326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4766882788299174e-06,
+      "loss": 0.8516,
+      "step": 14327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4757070284138785e-06,
+      "loss": 0.8164,
+      "step": 14328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4747260781513574e-06,
+      "loss": 0.7852,
+      "step": 14329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4737454280768937e-06,
+      "loss": 0.875,
+      "step": 14330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.472765078225019e-06,
+      "loss": 0.8828,
+      "step": 14331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4717850286302515e-06,
+      "loss": 0.8672,
+      "step": 14332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4708052793271e-06,
+      "loss": 0.8281,
+      "step": 14333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4698258303500624e-06,
+      "loss": 0.8945,
+      "step": 14334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4688466817336322e-06,
+      "loss": 0.9297,
+      "step": 14335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4678678335122786e-06,
+      "loss": 0.9453,
+      "step": 14336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.466889285720473e-06,
+      "loss": 0.8359,
+      "step": 14337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4659110383926701e-06,
+      "loss": 0.8828,
+      "step": 14338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4649330915633186e-06,
+      "loss": 0.9102,
+      "step": 14339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4639554452668492e-06,
+      "loss": 0.8906,
+      "step": 14340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4629780995376875e-06,
+      "loss": 0.8984,
+      "step": 14341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4620010544102504e-06,
+      "loss": 0.8711,
+      "step": 14342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.461024309918937e-06,
+      "loss": 0.9023,
+      "step": 14343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.460047866098142e-06,
+      "loss": 0.8438,
+      "step": 14344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.459071722982247e-06,
+      "loss": 0.9102,
+      "step": 14345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4580958806056255e-06,
+      "loss": 0.8203,
+      "step": 14346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4571203390026355e-06,
+      "loss": 0.8281,
+      "step": 14347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4561450982076275e-06,
+      "loss": 0.875,
+      "step": 14348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4551701582549427e-06,
+      "loss": 0.8594,
+      "step": 14349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4541955191789126e-06,
+      "loss": 0.8555,
+      "step": 14350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4532211810138496e-06,
+      "loss": 0.7305,
+      "step": 14351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4522471437940655e-06,
+      "loss": 0.9219,
+      "step": 14352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.451273407553858e-06,
+      "loss": 0.8711,
+      "step": 14353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4502999723275146e-06,
+      "loss": 0.8203,
+      "step": 14354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4493268381493075e-06,
+      "loss": 0.9062,
+      "step": 14355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4483540050535038e-06,
+      "loss": 0.8398,
+      "step": 14356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4473814730743607e-06,
+      "loss": 0.8164,
+      "step": 14357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4464092422461229e-06,
+      "loss": 0.8008,
+      "step": 14358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4454373126030207e-06,
+      "loss": 0.8594,
+      "step": 14359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.444465684179278e-06,
+      "loss": 0.8125,
+      "step": 14360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4434943570091088e-06,
+      "loss": 0.8672,
+      "step": 14361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4425233311267161e-06,
+      "loss": 0.9648,
+      "step": 14362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4415526065662876e-06,
+      "loss": 0.8125,
+      "step": 14363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4405821833620071e-06,
+      "loss": 0.8281,
+      "step": 14364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4396120615480424e-06,
+      "loss": 0.9766,
+      "step": 14365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4386422411585566e-06,
+      "loss": 0.75,
+      "step": 14366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4376727222276943e-06,
+      "loss": 0.8594,
+      "step": 14367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.436703504789596e-06,
+      "loss": 1.0391,
+      "step": 14368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.435734588878388e-06,
+      "loss": 0.8359,
+      "step": 14369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4347659745281916e-06,
+      "loss": 0.9062,
+      "step": 14370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4337976617731075e-06,
+      "loss": 0.8438,
+      "step": 14371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4328296506472328e-06,
+      "loss": 0.8164,
+      "step": 14372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4318619411846546e-06,
+      "loss": 0.7227,
+      "step": 14373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4308945334194491e-06,
+      "loss": 0.8281,
+      "step": 14374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4299274273856756e-06,
+      "loss": 0.8281,
+      "step": 14375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4289606231173892e-06,
+      "loss": 0.9141,
+      "step": 14376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4279941206486326e-06,
+      "loss": 0.9531,
+      "step": 14377
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4270279200134406e-06,
+      "loss": 0.8008,
+      "step": 14378
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.42606202124583e-06,
+      "loss": 0.8125,
+      "step": 14379
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4250964243798137e-06,
+      "loss": 0.832,
+      "step": 14380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.424131129449392e-06,
+      "loss": 0.9219,
+      "step": 14381
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4231661364885573e-06,
+      "loss": 0.8867,
+      "step": 14382
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4222014455312827e-06,
+      "loss": 0.9453,
+      "step": 14383
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4212370566115385e-06,
+      "loss": 0.8125,
+      "step": 14384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4202729697632843e-06,
+      "loss": 0.832,
+      "step": 14385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4193091850204676e-06,
+      "loss": 0.9141,
+      "step": 14386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.418345702417021e-06,
+      "loss": 0.7969,
+      "step": 14387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.417382521986872e-06,
+      "loss": 0.9609,
+      "step": 14388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4164196437639355e-06,
+      "loss": 0.7578,
+      "step": 14389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.415457067782119e-06,
+      "loss": 0.9219,
+      "step": 14390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4144947940753117e-06,
+      "loss": 0.8594,
+      "step": 14391
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4135328226773993e-06,
+      "loss": 0.7969,
+      "step": 14392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.412571153622252e-06,
+      "loss": 0.8203,
+      "step": 14393
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4116097869437373e-06,
+      "loss": 0.8203,
+      "step": 14394
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4106487226756994e-06,
+      "loss": 0.9023,
+      "step": 14395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4096879608519809e-06,
+      "loss": 0.8594,
+      "step": 14396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.408727501506416e-06,
+      "loss": 0.7695,
+      "step": 14397
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.407767344672819e-06,
+      "loss": 0.8477,
+      "step": 14398
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4068074903849992e-06,
+      "loss": 0.7305,
+      "step": 14399
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4058479386767554e-06,
+      "loss": 0.9258,
+      "step": 14400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.404888689581878e-06,
+      "loss": 0.7969,
+      "step": 14401
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.403929743134138e-06,
+      "loss": 0.8203,
+      "step": 14402
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.402971099367304e-06,
+      "loss": 0.8203,
+      "step": 14403
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4020127583151322e-06,
+      "loss": 0.7344,
+      "step": 14404
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4010547200113677e-06,
+      "loss": 0.8281,
+      "step": 14405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4000969844897415e-06,
+      "loss": 0.9102,
+      "step": 14406
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3991395517839778e-06,
+      "loss": 0.9336,
+      "step": 14407
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3981824219277906e-06,
+      "loss": 0.8203,
+      "step": 14408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.397225594954884e-06,
+      "loss": 0.8555,
+      "step": 14409
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3962690708989456e-06,
+      "loss": 0.875,
+      "step": 14410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3953128497936562e-06,
+      "loss": 0.7656,
+      "step": 14411
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3943569316726868e-06,
+      "loss": 0.8398,
+      "step": 14412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.393401316569699e-06,
+      "loss": 0.7617,
+      "step": 14413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3924460045183387e-06,
+      "loss": 0.8047,
+      "step": 14414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3914909955522426e-06,
+      "loss": 0.9219,
+      "step": 14415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.390536289705041e-06,
+      "loss": 0.9102,
+      "step": 14416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3895818870103516e-06,
+      "loss": 0.8516,
+      "step": 14417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3886277875017773e-06,
+      "loss": 0.8398,
+      "step": 14418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3876739912129134e-06,
+      "loss": 0.8516,
+      "step": 14419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3867204981773463e-06,
+      "loss": 1.0703,
+      "step": 14420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3857673084286505e-06,
+      "loss": 0.7344,
+      "step": 14421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3848144220003867e-06,
+      "loss": 0.7773,
+      "step": 14422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3838618389261082e-06,
+      "loss": 0.8672,
+      "step": 14423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3829095592393581e-06,
+      "loss": 0.7422,
+      "step": 14424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3819575829736698e-06,
+      "loss": 0.8477,
+      "step": 14425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3810059101625594e-06,
+      "loss": 0.8359,
+      "step": 14426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3800545408395371e-06,
+      "loss": 0.8789,
+      "step": 14427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.379103475038105e-06,
+      "loss": 0.8984,
+      "step": 14428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3781527127917515e-06,
+      "loss": 0.8594,
+      "step": 14429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3772022541339513e-06,
+      "loss": 0.8281,
+      "step": 14430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376252099098173e-06,
+      "loss": 0.8594,
+      "step": 14431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3753022477178735e-06,
+      "loss": 0.832,
+      "step": 14432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3743527000265e-06,
+      "loss": 0.8906,
+      "step": 14433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3734034560574848e-06,
+      "loss": 0.7539,
+      "step": 14434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3724545158442526e-06,
+      "loss": 0.7969,
+      "step": 14435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3715058794202174e-06,
+      "loss": 0.8438,
+      "step": 14436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3705575468187859e-06,
+      "loss": 0.8477,
+      "step": 14437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3696095180733437e-06,
+      "loss": 0.8125,
+      "step": 14438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3686617932172764e-06,
+      "loss": 0.8789,
+      "step": 14439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3677143722839537e-06,
+      "loss": 0.875,
+      "step": 14440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3667672553067391e-06,
+      "loss": 0.9219,
+      "step": 14441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3658204423189758e-06,
+      "loss": 0.7969,
+      "step": 14442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3648739333540062e-06,
+      "loss": 0.8594,
+      "step": 14443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.363927728445159e-06,
+      "loss": 0.9062,
+      "step": 14444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3629818276257513e-06,
+      "loss": 1.0,
+      "step": 14445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.362036230929088e-06,
+      "loss": 0.7891,
+      "step": 14446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3610909383884651e-06,
+      "loss": 0.7734,
+      "step": 14447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3601459500371694e-06,
+      "loss": 0.8789,
+      "step": 14448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3592012659084764e-06,
+      "loss": 0.8711,
+      "step": 14449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.358256886035646e-06,
+      "loss": 0.8203,
+      "step": 14450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.357312810451933e-06,
+      "loss": 0.8594,
+      "step": 14451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3563690391905838e-06,
+      "loss": 0.8867,
+      "step": 14452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.355425572284823e-06,
+      "loss": 0.8047,
+      "step": 14453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3544824097678755e-06,
+      "loss": 0.9648,
+      "step": 14454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.353539551672951e-06,
+      "loss": 0.793,
+      "step": 14455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.352596998033251e-06,
+      "loss": 0.8711,
+      "step": 14456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3516547488819598e-06,
+      "loss": 0.8828,
+      "step": 14457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3507128042522588e-06,
+      "loss": 0.8711,
+      "step": 14458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3497711641773136e-06,
+      "loss": 0.8047,
+      "step": 14459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3488298286902845e-06,
+      "loss": 0.7031,
+      "step": 14460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3478887978243117e-06,
+      "loss": 0.9141,
+      "step": 14461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3469480716125327e-06,
+      "loss": 0.8906,
+      "step": 14462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3460076500880738e-06,
+      "loss": 0.8008,
+      "step": 14463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.345067533284048e-06,
+      "loss": 0.8789,
+      "step": 14464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3441277212335569e-06,
+      "loss": 0.7852,
+      "step": 14465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3431882139696916e-06,
+      "loss": 0.8438,
+      "step": 14466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3422490115255372e-06,
+      "loss": 0.9297,
+      "step": 14467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3413101139341633e-06,
+      "loss": 0.8281,
+      "step": 14468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3403715212286284e-06,
+      "loss": 0.8242,
+      "step": 14469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3394332334419814e-06,
+      "loss": 0.9766,
+      "step": 14470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3384952506072624e-06,
+      "loss": 0.8438,
+      "step": 14471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.337557572757502e-06,
+      "loss": 0.8555,
+      "step": 14472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.336620199925711e-06,
+      "loss": 0.8438,
+      "step": 14473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3356831321449004e-06,
+      "loss": 0.7461,
+      "step": 14474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3347463694480634e-06,
+      "loss": 0.8242,
+      "step": 14475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.333809911868188e-06,
+      "loss": 0.8086,
+      "step": 14476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3328737594382434e-06,
+      "loss": 0.8906,
+      "step": 14477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.331937912191197e-06,
+      "loss": 0.9141,
+      "step": 14478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3310023701599994e-06,
+      "loss": 0.8125,
+      "step": 14479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3300671333775961e-06,
+      "loss": 0.8438,
+      "step": 14480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3291322018769126e-06,
+      "loss": 0.8203,
+      "step": 14481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3281975756908717e-06,
+      "loss": 0.8945,
+      "step": 14482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3272632548523833e-06,
+      "loss": 0.7578,
+      "step": 14483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3263292393943495e-06,
+      "loss": 1.0234,
+      "step": 14484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3253955293496535e-06,
+      "loss": 0.8203,
+      "step": 14485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3244621247511736e-06,
+      "loss": 0.7188,
+      "step": 14486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3235290256317767e-06,
+      "loss": 0.9062,
+      "step": 14487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3225962320243225e-06,
+      "loss": 0.9609,
+      "step": 14488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3216637439616497e-06,
+      "loss": 0.8242,
+      "step": 14489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.320731561476597e-06,
+      "loss": 0.8398,
+      "step": 14490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3197996846019856e-06,
+      "loss": 0.7422,
+      "step": 14491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3188681133706326e-06,
+      "loss": 0.832,
+      "step": 14492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3179368478153343e-06,
+      "loss": 0.8516,
+      "step": 14493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3170058879688851e-06,
+      "loss": 0.8711,
+      "step": 14494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3160752338640647e-06,
+      "loss": 0.8438,
+      "step": 14495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3151448855336445e-06,
+      "loss": 0.832,
+      "step": 14496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3142148430103819e-06,
+      "loss": 0.8125,
+      "step": 14497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3132851063270246e-06,
+      "loss": 0.9688,
+      "step": 14498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3123556755163114e-06,
+      "loss": 0.7656,
+      "step": 14499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3114265506109702e-06,
+      "loss": 0.832,
+      "step": 14500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3104977316437128e-06,
+      "loss": 0.7734,
+      "step": 14501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3095692186472476e-06,
+      "loss": 0.7969,
+      "step": 14502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.308641011654268e-06,
+      "loss": 0.6914,
+      "step": 14503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3077131106974595e-06,
+      "loss": 0.8789,
+      "step": 14504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3067855158094912e-06,
+      "loss": 0.9258,
+      "step": 14505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.305858227023028e-06,
+      "loss": 1.0625,
+      "step": 14506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3049312443707197e-06,
+      "loss": 0.9609,
+      "step": 14507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3040045678852099e-06,
+      "loss": 0.9609,
+      "step": 14508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3030781975991226e-06,
+      "loss": 0.8281,
+      "step": 14509
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3021521335450815e-06,
+      "loss": 0.8477,
+      "step": 14510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3012263757556942e-06,
+      "loss": 0.8594,
+      "step": 14511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3003009242635556e-06,
+      "loss": 0.7539,
+      "step": 14512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299375779101253e-06,
+      "loss": 0.7695,
+      "step": 14513
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2984509403013634e-06,
+      "loss": 0.8594,
+      "step": 14514
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2975264078964544e-06,
+      "loss": 0.8281,
+      "step": 14515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2966021819190743e-06,
+      "loss": 0.9727,
+      "step": 14516
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2956782624017694e-06,
+      "loss": 0.8359,
+      "step": 14517
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2947546493770713e-06,
+      "loss": 0.8477,
+      "step": 14518
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2938313428775062e-06,
+      "loss": 0.8164,
+      "step": 14519
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2929083429355794e-06,
+      "loss": 0.9688,
+      "step": 14520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2919856495837935e-06,
+      "loss": 0.7891,
+      "step": 14521
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2910632628546382e-06,
+      "loss": 0.8711,
+      "step": 14522
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2901411827805943e-06,
+      "loss": 0.8594,
+      "step": 14523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2892194093941245e-06,
+      "loss": 0.8398,
+      "step": 14524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2882979427276887e-06,
+      "loss": 0.832,
+      "step": 14525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.287376782813734e-06,
+      "loss": 0.6836,
+      "step": 14526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2864559296846967e-06,
+      "loss": 0.8164,
+      "step": 14527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2855353833729977e-06,
+      "loss": 0.7891,
+      "step": 14528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2846151439110533e-06,
+      "loss": 0.8086,
+      "step": 14529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.283695211331265e-06,
+      "loss": 0.8516,
+      "step": 14530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2827755856660295e-06,
+      "loss": 0.9336,
+      "step": 14531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.281856266947723e-06,
+      "loss": 0.7305,
+      "step": 14532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2809372552087174e-06,
+      "loss": 0.8828,
+      "step": 14533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2800185504813735e-06,
+      "loss": 0.9141,
+      "step": 14534
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.279100152798043e-06,
+      "loss": 0.8164,
+      "step": 14535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2781820621910579e-06,
+      "loss": 0.9531,
+      "step": 14536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2772642786927491e-06,
+      "loss": 0.875,
+      "step": 14537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276346802335433e-06,
+      "loss": 0.8242,
+      "step": 14538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.275429633151417e-06,
+      "loss": 0.8516,
+      "step": 14539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2745127711729932e-06,
+      "loss": 0.8281,
+      "step": 14540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2735962164324455e-06,
+      "loss": 0.8203,
+      "step": 14541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2726799689620495e-06,
+      "loss": 0.7969,
+      "step": 14542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.271764028794069e-06,
+      "loss": 0.7969,
+      "step": 14543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.270848395960751e-06,
+      "loss": 1.0234,
+      "step": 14544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.269933070494339e-06,
+      "loss": 0.8477,
+      "step": 14545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2690180524270623e-06,
+      "loss": 0.7656,
+      "step": 14546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2681033417911427e-06,
+      "loss": 0.8281,
+      "step": 14547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.267188938618784e-06,
+      "loss": 0.7969,
+      "step": 14548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2662748429421868e-06,
+      "loss": 0.875,
+      "step": 14549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.265361054793538e-06,
+      "loss": 0.8047,
+      "step": 14550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2644475742050132e-06,
+      "loss": 0.7891,
+      "step": 14551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2635344012087758e-06,
+      "loss": 0.957,
+      "step": 14552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2626215358369809e-06,
+      "loss": 0.8242,
+      "step": 14553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.261708978121773e-06,
+      "loss": 0.9297,
+      "step": 14554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.260796728095285e-06,
+      "loss": 0.8516,
+      "step": 14555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2598847857896356e-06,
+      "loss": 0.9062,
+      "step": 14556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2589731512369374e-06,
+      "loss": 0.7773,
+      "step": 14557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.258061824469291e-06,
+      "loss": 0.8164,
+      "step": 14558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2571508055187875e-06,
+      "loss": 0.8438,
+      "step": 14559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2562400944175001e-06,
+      "loss": 0.8984,
+      "step": 14560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2553296911975e-06,
+      "loss": 0.8281,
+      "step": 14561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2544195958908434e-06,
+      "loss": 0.8203,
+      "step": 14562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2535098085295783e-06,
+      "loss": 0.9141,
+      "step": 14563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2526003291457345e-06,
+      "loss": 0.8125,
+      "step": 14564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2516911577713398e-06,
+      "loss": 0.7148,
+      "step": 14565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2507822944384085e-06,
+      "loss": 0.8125,
+      "step": 14566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2498737391789395e-06,
+      "loss": 0.7383,
+      "step": 14567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.248965492024925e-06,
+      "loss": 0.7891,
+      "step": 14568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2480575530083484e-06,
+      "loss": 0.9023,
+      "step": 14569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2471499221611806e-06,
+      "loss": 0.832,
+      "step": 14570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.246242599515376e-06,
+      "loss": 0.8672,
+      "step": 14571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2453355851028848e-06,
+      "loss": 0.8086,
+      "step": 14572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2444288789556446e-06,
+      "loss": 0.8789,
+      "step": 14573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2435224811055858e-06,
+      "loss": 0.9688,
+      "step": 14574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2426163915846169e-06,
+      "loss": 0.8906,
+      "step": 14575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.241710610424647e-06,
+      "loss": 0.875,
+      "step": 14576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2408051376575681e-06,
+      "loss": 0.8398,
+      "step": 14577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2398999733152671e-06,
+      "loss": 0.9336,
+      "step": 14578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2389951174296122e-06,
+      "loss": 0.8281,
+      "step": 14579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2380905700324652e-06,
+      "loss": 0.8008,
+      "step": 14580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2371863311556786e-06,
+      "loss": 0.8047,
+      "step": 14581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2362824008310924e-06,
+      "loss": 0.9453,
+      "step": 14582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2353787790905324e-06,
+      "loss": 0.9883,
+      "step": 14583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2344754659658175e-06,
+      "loss": 0.7695,
+      "step": 14584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2335724614887557e-06,
+      "loss": 0.7539,
+      "step": 14585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2326697656911456e-06,
+      "loss": 1.0,
+      "step": 14586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2317673786047667e-06,
+      "loss": 0.9062,
+      "step": 14587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2308653002613968e-06,
+      "loss": 0.9297,
+      "step": 14588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2299635306927992e-06,
+      "loss": 0.8672,
+      "step": 14589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2290620699307287e-06,
+      "loss": 0.7031,
+      "step": 14590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.228160918006923e-06,
+      "loss": 0.7773,
+      "step": 14591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.227260074953115e-06,
+      "loss": 0.9102,
+      "step": 14592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2263595408010253e-06,
+      "loss": 0.8867,
+      "step": 14593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2254593155823647e-06,
+      "loss": 0.9141,
+      "step": 14594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2245593993288263e-06,
+      "loss": 0.8398,
+      "step": 14595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2236597920721016e-06,
+      "loss": 0.7578,
+      "step": 14596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2227604938438654e-06,
+      "loss": 0.8359,
+      "step": 14597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2218615046757877e-06,
+      "loss": 0.8008,
+      "step": 14598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2209628245995164e-06,
+      "loss": 0.8398,
+      "step": 14599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2200644536466998e-06,
+      "loss": 0.9453,
+      "step": 14600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2191663918489693e-06,
+      "loss": 0.8203,
+      "step": 14601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2182686392379505e-06,
+      "loss": 0.8984,
+      "step": 14602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2173711958452495e-06,
+      "loss": 0.8828,
+      "step": 14603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2164740617024683e-06,
+      "loss": 0.8711,
+      "step": 14604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2155772368411978e-06,
+      "loss": 0.9336,
+      "step": 14605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2146807212930178e-06,
+      "loss": 0.8359,
+      "step": 14606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2137845150894922e-06,
+      "loss": 0.8555,
+      "step": 14607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2128886182621803e-06,
+      "loss": 0.8867,
+      "step": 14608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2119930308426264e-06,
+      "loss": 0.7812,
+      "step": 14609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2110977528623692e-06,
+      "loss": 0.8555,
+      "step": 14610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2102027843529273e-06,
+      "loss": 0.8711,
+      "step": 14611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2093081253458172e-06,
+      "loss": 0.8438,
+      "step": 14612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2084137758725433e-06,
+      "loss": 0.8281,
+      "step": 14613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2075197359645918e-06,
+      "loss": 0.8008,
+      "step": 14614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2066260056534463e-06,
+      "loss": 0.7617,
+      "step": 14615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2057325849705759e-06,
+      "loss": 0.7891,
+      "step": 14616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2048394739474424e-06,
+      "loss": 0.8789,
+      "step": 14617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.203946672615488e-06,
+      "loss": 0.7969,
+      "step": 14618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2030541810061525e-06,
+      "loss": 0.8203,
+      "step": 14619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2021619991508614e-06,
+      "loss": 0.8281,
+      "step": 14620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2012701270810322e-06,
+      "loss": 0.6758,
+      "step": 14621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2003785648280653e-06,
+      "loss": 0.7695,
+      "step": 14622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1994873124233552e-06,
+      "loss": 0.8984,
+      "step": 14623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1985963698982872e-06,
+      "loss": 0.7656,
+      "step": 14624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1977057372842282e-06,
+      "loss": 0.7734,
+      "step": 14625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1968154146125411e-06,
+      "loss": 0.8789,
+      "step": 14626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.195925401914575e-06,
+      "loss": 0.9766,
+      "step": 14627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1950356992216717e-06,
+      "loss": 0.7539,
+      "step": 14628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1941463065651537e-06,
+      "loss": 0.9336,
+      "step": 14629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1932572239763395e-06,
+      "loss": 0.7578,
+      "step": 14630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1923684514865374e-06,
+      "loss": 0.8125,
+      "step": 14631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1914799891270423e-06,
+      "loss": 0.8008,
+      "step": 14632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1905918369291358e-06,
+      "loss": 0.8555,
+      "step": 14633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1897039949240919e-06,
+      "loss": 0.8906,
+      "step": 14634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1888164631431743e-06,
+      "loss": 0.793,
+      "step": 14635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1879292416176346e-06,
+      "loss": 0.8867,
+      "step": 14636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1870423303787104e-06,
+      "loss": 0.9141,
+      "step": 14637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.186155729457632e-06,
+      "loss": 0.8867,
+      "step": 14638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1852694388856223e-06,
+      "loss": 0.8125,
+      "step": 14639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.184383458693883e-06,
+      "loss": 0.9023,
+      "step": 14640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1834977889136135e-06,
+      "loss": 0.8594,
+      "step": 14641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.182612429575999e-06,
+      "loss": 0.8242,
+      "step": 14642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1817273807122175e-06,
+      "loss": 0.9648,
+      "step": 14643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180842642353428e-06,
+      "loss": 0.8789,
+      "step": 14644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1799582145307853e-06,
+      "loss": 0.8359,
+      "step": 14645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1790740972754323e-06,
+      "loss": 0.8672,
+      "step": 14646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1781902906185028e-06,
+      "loss": 0.8633,
+      "step": 14647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.177306794591111e-06,
+      "loss": 0.8203,
+      "step": 14648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1764236092243685e-06,
+      "loss": 0.9922,
+      "step": 14649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1755407345493752e-06,
+      "loss": 0.9414,
+      "step": 14650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1746581705972205e-06,
+      "loss": 0.9141,
+      "step": 14651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1737759173989748e-06,
+      "loss": 0.8438,
+      "step": 14652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.172893974985706e-06,
+      "loss": 0.8438,
+      "step": 14653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1720123433884723e-06,
+      "loss": 0.8359,
+      "step": 14654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1711310226383122e-06,
+      "loss": 0.793,
+      "step": 14655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.170250012766261e-06,
+      "loss": 0.7969,
+      "step": 14656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1693693138033391e-06,
+      "loss": 0.9531,
+      "step": 14657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1684889257805608e-06,
+      "loss": 0.832,
+      "step": 14658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1676088487289205e-06,
+      "loss": 0.8203,
+      "step": 14659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1667290826794109e-06,
+      "loss": 0.8125,
+      "step": 14660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1658496276630083e-06,
+      "loss": 1.0234,
+      "step": 14661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1649704837106823e-06,
+      "loss": 1.0078,
+      "step": 14662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.164091650853384e-06,
+      "loss": 0.9102,
+      "step": 14663
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1632131291220628e-06,
+      "loss": 0.8047,
+      "step": 14664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1623349185476495e-06,
+      "loss": 0.8125,
+      "step": 14665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1614570191610718e-06,
+      "loss": 0.8086,
+      "step": 14666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1605794309932383e-06,
+      "loss": 0.6914,
+      "step": 14667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1597021540750497e-06,
+      "loss": 1.0078,
+      "step": 14668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1588251884374003e-06,
+      "loss": 0.918,
+      "step": 14669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157948534111164e-06,
+      "loss": 0.8945,
+      "step": 14670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157072191127212e-06,
+      "loss": 0.9258,
+      "step": 14671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.156196159516403e-06,
+      "loss": 0.9688,
+      "step": 14672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.155320439309583e-06,
+      "loss": 0.9219,
+      "step": 14673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1544450305375843e-06,
+      "loss": 0.7891,
+      "step": 14674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1535699332312334e-06,
+      "loss": 0.8594,
+      "step": 14675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1526951474213443e-06,
+      "loss": 0.9141,
+      "step": 14676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1518206731387216e-06,
+      "loss": 0.8633,
+      "step": 14677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1509465104141515e-06,
+      "loss": 0.8906,
+      "step": 14678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1500726592784183e-06,
+      "loss": 0.793,
+      "step": 14679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.149199119762292e-06,
+      "loss": 0.8516,
+      "step": 14680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1483258918965313e-06,
+      "loss": 0.8516,
+      "step": 14681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1474529757118812e-06,
+      "loss": 0.8867,
+      "step": 14682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.14658037123908e-06,
+      "loss": 0.8711,
+      "step": 14683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1457080785088548e-06,
+      "loss": 0.8281,
+      "step": 14684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.144836097551918e-06,
+      "loss": 0.9922,
+      "step": 14685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1439644283989749e-06,
+      "loss": 0.8633,
+      "step": 14686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1430930710807174e-06,
+      "loss": 0.8008,
+      "step": 14687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1422220256278293e-06,
+      "loss": 0.8477,
+      "step": 14688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1413512920709792e-06,
+      "loss": 0.8984,
+      "step": 14689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1404808704408277e-06,
+      "loss": 0.8359,
+      "step": 14690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.139610760768023e-06,
+      "loss": 0.8516,
+      "step": 14691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1387409630832068e-06,
+      "loss": 0.8164,
+      "step": 14692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1378714774169997e-06,
+      "loss": 0.7695,
+      "step": 14693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1370023038000223e-06,
+      "loss": 0.8789,
+      "step": 14694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1361334422628777e-06,
+      "loss": 0.8906,
+      "step": 14695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.135264892836162e-06,
+      "loss": 0.8203,
+      "step": 14696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1343966555504549e-06,
+      "loss": 0.8789,
+      "step": 14697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1335287304363307e-06,
+      "loss": 0.8125,
+      "step": 14698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1326611175243484e-06,
+      "loss": 0.7539,
+      "step": 14699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1317938168450627e-06,
+      "loss": 0.8086,
+      "step": 14700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1309268284290054e-06,
+      "loss": 0.7734,
+      "step": 14701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.13006015230671e-06,
+      "loss": 0.8828,
+      "step": 14702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1291937885086912e-06,
+      "loss": 0.875,
+      "step": 14703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1283277370654566e-06,
+      "loss": 0.8594,
+      "step": 14704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1274619980074996e-06,
+      "loss": 0.8047,
+      "step": 14705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1265965713653038e-06,
+      "loss": 0.9922,
+      "step": 14706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1257314571693445e-06,
+      "loss": 0.793,
+      "step": 14707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1248666554500831e-06,
+      "loss": 0.7578,
+      "step": 14708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1240021662379686e-06,
+      "loss": 0.9414,
+      "step": 14709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.123137989563442e-06,
+      "loss": 0.8555,
+      "step": 14710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1222741254569335e-06,
+      "loss": 0.6602,
+      "step": 14711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1214105739488612e-06,
+      "loss": 0.8438,
+      "step": 14712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1205473350696295e-06,
+      "loss": 0.875,
+      "step": 14713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1196844088496361e-06,
+      "loss": 0.9219,
+      "step": 14714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.118821795319266e-06,
+      "loss": 0.8906,
+      "step": 14715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1179594945088946e-06,
+      "loss": 0.7695,
+      "step": 14716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.117097506448881e-06,
+      "loss": 0.832,
+      "step": 14717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1162358311695799e-06,
+      "loss": 0.7969,
+      "step": 14718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1153744687013313e-06,
+      "loss": 0.8594,
+      "step": 14719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1145134190744677e-06,
+      "loss": 0.8555,
+      "step": 14720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1136526823193027e-06,
+      "loss": 0.9219,
+      "step": 14721
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1127922584661487e-06,
+      "loss": 0.8086,
+      "step": 14722
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1119321475453005e-06,
+      "loss": 0.9414,
+      "step": 14723
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1110723495870468e-06,
+      "loss": 0.8438,
+      "step": 14724
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.110212864621658e-06,
+      "loss": 0.8164,
+      "step": 14725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1093536926794001e-06,
+      "loss": 0.7383,
+      "step": 14726
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1084948337905255e-06,
+      "loss": 0.9609,
+      "step": 14727
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1076362879852787e-06,
+      "loss": 0.7734,
+      "step": 14728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1067780552938855e-06,
+      "loss": 0.8477,
+      "step": 14729
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1059201357465687e-06,
+      "loss": 0.9766,
+      "step": 14730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1050625293735351e-06,
+      "loss": 0.9453,
+      "step": 14731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1042052362049872e-06,
+      "loss": 0.7891,
+      "step": 14732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1033482562711052e-06,
+      "loss": 0.8594,
+      "step": 14733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1024915896020671e-06,
+      "loss": 0.8047,
+      "step": 14734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1016352362280402e-06,
+      "loss": 0.7773,
+      "step": 14735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1007791961791736e-06,
+      "loss": 0.7891,
+      "step": 14736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.099923469485612e-06,
+      "loss": 0.8633,
+      "step": 14737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0990680561774868e-06,
+      "loss": 0.7695,
+      "step": 14738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0982129562849208e-06,
+      "loss": 0.957,
+      "step": 14739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0973581698380187e-06,
+      "loss": 0.9141,
+      "step": 14740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0965036968668807e-06,
+      "loss": 0.8359,
+      "step": 14741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0956495374015942e-06,
+      "loss": 0.793,
+      "step": 14742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094795691472238e-06,
+      "loss": 0.8359,
+      "step": 14743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.093942159108874e-06,
+      "loss": 1.0156,
+      "step": 14744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0930889403415568e-06,
+      "loss": 0.8906,
+      "step": 14745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0922360352003303e-06,
+      "loss": 0.8711,
+      "step": 14746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0913834437152283e-06,
+      "loss": 0.8906,
+      "step": 14747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0905311659162676e-06,
+      "loss": 0.8633,
+      "step": 14748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0896792018334624e-06,
+      "loss": 0.8945,
+      "step": 14749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0888275514968083e-06,
+      "loss": 0.8359,
+      "step": 14750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0879762149362983e-06,
+      "loss": 0.75,
+      "step": 14751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0871251921819027e-06,
+      "loss": 0.8047,
+      "step": 14752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0862744832635907e-06,
+      "loss": 0.9219,
+      "step": 14753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0854240882113177e-06,
+      "loss": 0.8281,
+      "step": 14754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0845740070550281e-06,
+      "loss": 0.7539,
+      "step": 14755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0837242398246516e-06,
+      "loss": 0.7852,
+      "step": 14756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0828747865501111e-06,
+      "loss": 0.8047,
+      "step": 14757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0820256472613166e-06,
+      "loss": 0.8789,
+      "step": 14758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0811768219881713e-06,
+      "loss": 0.8594,
+      "step": 14759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0803283107605577e-06,
+      "loss": 0.8633,
+      "step": 14760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0794801136083577e-06,
+      "loss": 0.9023,
+      "step": 14761
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.078632230561435e-06,
+      "loss": 0.8086,
+      "step": 14762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0777846616496478e-06,
+      "loss": 0.8281,
+      "step": 14763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0769374069028371e-06,
+      "loss": 0.8633,
+      "step": 14764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0760904663508375e-06,
+      "loss": 0.7227,
+      "step": 14765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.075243840023471e-06,
+      "loss": 0.8555,
+      "step": 14766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0743975279505515e-06,
+      "loss": 0.8945,
+      "step": 14767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0735515301618726e-06,
+      "loss": 0.8125,
+      "step": 14768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0727058466872276e-06,
+      "loss": 0.9883,
+      "step": 14769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0718604775563934e-06,
+      "loss": 0.8711,
+      "step": 14770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0710154227991388e-06,
+      "loss": 0.7148,
+      "step": 14771
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0701706824452152e-06,
+      "loss": 0.8984,
+      "step": 14772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06932625652437e-06,
+      "loss": 0.9297,
+      "step": 14773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0684821450663364e-06,
+      "loss": 0.8047,
+      "step": 14774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.067638348100838e-06,
+      "loss": 0.7773,
+      "step": 14775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0667948656575822e-06,
+      "loss": 0.9453,
+      "step": 14776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.065951697766273e-06,
+      "loss": 0.8516,
+      "step": 14777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0651088444565993e-06,
+      "loss": 0.8203,
+      "step": 14778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.064266305758239e-06,
+      "loss": 0.8633,
+      "step": 14779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0634240817008578e-06,
+      "loss": 0.8633,
+      "step": 14780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0625821723141128e-06,
+      "loss": 0.9258,
+      "step": 14781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0617405776276491e-06,
+      "loss": 0.7852,
+      "step": 14782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0608992976711018e-06,
+      "loss": 0.9219,
+      "step": 14783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06005833247409e-06,
+      "loss": 0.7891,
+      "step": 14784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0592176820662282e-06,
+      "loss": 0.8125,
+      "step": 14785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0583773464771151e-06,
+      "loss": 0.9531,
+      "step": 14786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0575373257363452e-06,
+      "loss": 0.8711,
+      "step": 14787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0566976198734902e-06,
+      "loss": 0.7188,
+      "step": 14788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0558582289181208e-06,
+      "loss": 0.8555,
+      "step": 14789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0550191528997945e-06,
+      "loss": 0.7773,
+      "step": 14790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.054180391848052e-06,
+      "loss": 0.9492,
+      "step": 14791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.053341945792432e-06,
+      "loss": 0.8672,
+      "step": 14792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0525038147624545e-06,
+      "loss": 0.8008,
+      "step": 14793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0516659987876353e-06,
+      "loss": 0.8672,
+      "step": 14794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0508284978974692e-06,
+      "loss": 0.8984,
+      "step": 14795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0499913121214511e-06,
+      "loss": 0.7812,
+      "step": 14796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0491544414890563e-06,
+      "loss": 0.8828,
+      "step": 14797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0483178860297561e-06,
+      "loss": 0.8555,
+      "step": 14798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.047481645773002e-06,
+      "loss": 0.8906,
+      "step": 14799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0466457207482428e-06,
+      "loss": 0.8945,
+      "step": 14800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0458101109849107e-06,
+      "loss": 0.8867,
+      "step": 14801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.044974816512432e-06,
+      "loss": 0.8711,
+      "step": 14802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0441398373602152e-06,
+      "loss": 0.9648,
+      "step": 14803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0433051735576616e-06,
+      "loss": 1.0078,
+      "step": 14804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0424708251341632e-06,
+      "loss": 0.8203,
+      "step": 14805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0416367921190985e-06,
+      "loss": 0.875,
+      "step": 14806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040803074541832e-06,
+      "loss": 0.8008,
+      "step": 14807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0399696724317233e-06,
+      "loss": 0.8047,
+      "step": 14808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0391365858181156e-06,
+      "loss": 0.7383,
+      "step": 14809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0383038147303458e-06,
+      "loss": 0.9062,
+      "step": 14810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0374713591977348e-06,
+      "loss": 0.8359,
+      "step": 14811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0366392192495944e-06,
+      "loss": 0.8477,
+      "step": 14812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.035807394915227e-06,
+      "loss": 0.9414,
+      "step": 14813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0349758862239233e-06,
+      "loss": 0.9336,
+      "step": 14814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0341446932049593e-06,
+      "loss": 0.9062,
+      "step": 14815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0333138158876032e-06,
+      "loss": 0.8008,
+      "step": 14816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0324832543011131e-06,
+      "loss": 0.8281,
+      "step": 14817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0316530084747344e-06,
+      "loss": 0.9453,
+      "step": 14818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0308230784376994e-06,
+      "loss": 0.7734,
+      "step": 14819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0299934642192321e-06,
+      "loss": 0.7891,
+      "step": 14820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.029164165848544e-06,
+      "loss": 0.9258,
+      "step": 14821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0283351833548382e-06,
+      "loss": 0.7422,
+      "step": 14822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0275065167673016e-06,
+      "loss": 0.8203,
+      "step": 14823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0266781661151137e-06,
+      "loss": 0.8438,
+      "step": 14824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0258501314274427e-06,
+      "loss": 0.7891,
+      "step": 14825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0250224127334464e-06,
+      "loss": 0.832,
+      "step": 14826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0241950100622656e-06,
+      "loss": 0.9453,
+      "step": 14827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0233679234430383e-06,
+      "loss": 0.7969,
+      "step": 14828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0225411529048857e-06,
+      "loss": 1.0703,
+      "step": 14829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0217146984769222e-06,
+      "loss": 0.7383,
+      "step": 14830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0208885601882447e-06,
+      "loss": 0.7695,
+      "step": 14831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.020062738067945e-06,
+      "loss": 0.9258,
+      "step": 14832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0192372321451004e-06,
+      "loss": 0.8594,
+      "step": 14833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.018412042448782e-06,
+      "loss": 0.8281,
+      "step": 14834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0175871690080408e-06,
+      "loss": 0.8047,
+      "step": 14835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0167626118519237e-06,
+      "loss": 0.7891,
+      "step": 14836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.015938371009466e-06,
+      "loss": 0.793,
+      "step": 14837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0151144465096908e-06,
+      "loss": 0.8906,
+      "step": 14838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0142908383816074e-06,
+      "loss": 0.8789,
+      "step": 14839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0134675466542166e-06,
+      "loss": 0.7812,
+      "step": 14840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0126445713565092e-06,
+      "loss": 0.9375,
+      "step": 14841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0118219125174656e-06,
+      "loss": 0.9102,
+      "step": 14842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.010999570166047e-06,
+      "loss": 0.7891,
+      "step": 14843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0101775443312133e-06,
+      "loss": 0.918,
+      "step": 14844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0093558350419097e-06,
+      "loss": 1.0547,
+      "step": 14845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0085344423270671e-06,
+      "loss": 0.9258,
+      "step": 14846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0077133662156092e-06,
+      "loss": 1.0234,
+      "step": 14847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0068926067364483e-06,
+      "loss": 0.8008,
+      "step": 14848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0060721639184856e-06,
+      "loss": 0.8438,
+      "step": 14849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0052520377906071e-06,
+      "loss": 0.8672,
+      "step": 14850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.004432228381692e-06,
+      "loss": 0.793,
+      "step": 14851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.003612735720607e-06,
+      "loss": 0.7812,
+      "step": 14852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0027935598362104e-06,
+      "loss": 0.8672,
+      "step": 14853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0019747007573421e-06,
+      "loss": 0.793,
+      "step": 14854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0011561585128372e-06,
+      "loss": 0.7969,
+      "step": 14855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0003379331315178e-06,
+      "loss": 0.9219,
+      "step": 14856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.995200246421976e-07,
+      "loss": 0.8164,
+      "step": 14857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.987024330736727e-07,
+      "loss": 0.8242,
+      "step": 14858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.978851584547321e-07,
+      "loss": 0.8828,
+      "step": 14859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.970682008141552e-07,
+      "loss": 0.7656,
+      "step": 14860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.962515601807087e-07,
+      "loss": 0.8906,
+      "step": 14861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.954352365831455e-07,
+      "loss": 0.8242,
+      "step": 14862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.946192300502088e-07,
+      "loss": 0.8359,
+      "step": 14863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.93803540610635e-07,
+      "loss": 0.8086,
+      "step": 14864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.929881682931453e-07,
+      "loss": 0.8125,
+      "step": 14865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.921731131264478e-07,
+      "loss": 0.7695,
+      "step": 14866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.913583751392419e-07,
+      "loss": 0.875,
+      "step": 14867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.90543954360218e-07,
+      "loss": 0.9531,
+      "step": 14868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.89729850818053e-07,
+      "loss": 0.8008,
+      "step": 14869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.889160645414097e-07,
+      "loss": 0.8633,
+      "step": 14870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.881025955589464e-07,
+      "loss": 0.7656,
+      "step": 14871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.872894438993031e-07,
+      "loss": 0.7656,
+      "step": 14872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.864766095911171e-07,
+      "loss": 0.8945,
+      "step": 14873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.856640926630046e-07,
+      "loss": 0.8711,
+      "step": 14874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.84851893143577e-07,
+      "loss": 0.8047,
+      "step": 14875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.840400110614345e-07,
+      "loss": 0.8125,
+      "step": 14876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.832284464451648e-07,
+      "loss": 0.7461,
+      "step": 14877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.824171993233423e-07,
+      "loss": 0.9336,
+      "step": 14878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.816062697245321e-07,
+      "loss": 1.1172,
+      "step": 14879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.807956576772903e-07,
+      "loss": 0.7852,
+      "step": 14880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.799853632101608e-07,
+      "loss": 0.8945,
+      "step": 14881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.791753863516717e-07,
+      "loss": 0.8945,
+      "step": 14882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.78365727130346e-07,
+      "loss": 0.7734,
+      "step": 14883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.775563855746918e-07,
+      "loss": 0.8789,
+      "step": 14884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.767473617132094e-07,
+      "loss": 0.793,
+      "step": 14885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.759386555743821e-07,
+      "loss": 0.6719,
+      "step": 14886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.751302671866892e-07,
+      "loss": 0.8555,
+      "step": 14887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.743221965785931e-07,
+      "loss": 0.8555,
+      "step": 14888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.735144437785504e-07,
+      "loss": 0.9727,
+      "step": 14889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.727070088149982e-07,
+      "loss": 0.8438,
+      "step": 14890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.718998917163714e-07,
+      "loss": 0.7539,
+      "step": 14891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.710930925110872e-07,
+      "loss": 0.875,
+      "step": 14892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.702866112275589e-07,
+      "loss": 0.875,
+      "step": 14893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69480447894179e-07,
+      "loss": 0.8516,
+      "step": 14894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.686746025393357e-07,
+      "loss": 0.8086,
+      "step": 14895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.678690751914032e-07,
+      "loss": 0.793,
+      "step": 14896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.67063865878749e-07,
+      "loss": 0.8359,
+      "step": 14897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.6625897462972e-07,
+      "loss": 0.8438,
+      "step": 14898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.65454401472662e-07,
+      "loss": 0.832,
+      "step": 14899
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.646501464359027e-07,
+      "loss": 0.8281,
+      "step": 14900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.638462095477641e-07,
+      "loss": 0.793,
+      "step": 14901
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.630425908365503e-07,
+      "loss": 0.8359,
+      "step": 14902
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.62239290330561e-07,
+      "loss": 0.7695,
+      "step": 14903
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.614363080580813e-07,
+      "loss": 0.6992,
+      "step": 14904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.606336440473828e-07,
+      "loss": 0.9766,
+      "step": 14905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59831298326731e-07,
+      "loss": 0.7617,
+      "step": 14906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.590292709243776e-07,
+      "loss": 0.9141,
+      "step": 14907
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.582275618685644e-07,
+      "loss": 0.8008,
+      "step": 14908
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.574261711875165e-07,
+      "loss": 0.9023,
+      "step": 14909
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.56625098909455e-07,
+      "loss": 0.7891,
+      "step": 14910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55824345062587e-07,
+      "loss": 0.6992,
+      "step": 14911
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.550239096751102e-07,
+      "loss": 0.8633,
+      "step": 14912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.542237927752052e-07,
+      "loss": 0.9531,
+      "step": 14913
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.534239943910462e-07,
+      "loss": 0.9219,
+      "step": 14914
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.526245145507973e-07,
+      "loss": 0.9531,
+      "step": 14915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.518253532826094e-07,
+      "loss": 0.7773,
+      "step": 14916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.510265106146199e-07,
+      "loss": 0.7773,
+      "step": 14917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.502279865749575e-07,
+      "loss": 0.7852,
+      "step": 14918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.494297811917419e-07,
+      "loss": 0.8867,
+      "step": 14919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.486318944930784e-07,
+      "loss": 0.8672,
+      "step": 14920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.4783432650706e-07,
+      "loss": 0.8164,
+      "step": 14921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.47037077261771e-07,
+      "loss": 0.8516,
+      "step": 14922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.462401467852844e-07,
+      "loss": 0.6289,
+      "step": 14923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.454435351056635e-07,
+      "loss": 0.8633,
+      "step": 14924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.446472422509534e-07,
+      "loss": 0.9609,
+      "step": 14925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.438512682491963e-07,
+      "loss": 0.832,
+      "step": 14926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.430556131284174e-07,
+      "loss": 0.7148,
+      "step": 14927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.422602769166367e-07,
+      "loss": 0.8828,
+      "step": 14928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.41465259641855e-07,
+      "loss": 0.8906,
+      "step": 14929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.406705613320677e-07,
+      "loss": 0.8359,
+      "step": 14930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.39876182015258e-07,
+      "loss": 0.9102,
+      "step": 14931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.39082121719398e-07,
+      "loss": 0.9219,
+      "step": 14932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.382883804724452e-07,
+      "loss": 0.8359,
+      "step": 14933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.374949583023496e-07,
+      "loss": 0.8438,
+      "step": 14934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.367018552370489e-07,
+      "loss": 0.7969,
+      "step": 14935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.359090713044716e-07,
+      "loss": 0.9023,
+      "step": 14936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.351166065325279e-07,
+      "loss": 0.9062,
+      "step": 14937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.343244609491253e-07,
+      "loss": 0.7539,
+      "step": 14938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.33532634582156e-07,
+      "loss": 0.875,
+      "step": 14939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.327411274595033e-07,
+      "loss": 0.9023,
+      "step": 14940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.319499396090337e-07,
+      "loss": 0.8008,
+      "step": 14941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.311590710586071e-07,
+      "loss": 0.8633,
+      "step": 14942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.303685218360714e-07,
+      "loss": 0.8555,
+      "step": 14943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.295782919692665e-07,
+      "loss": 0.8672,
+      "step": 14944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.287883814860122e-07,
+      "loss": 0.9141,
+      "step": 14945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.279987904141252e-07,
+      "loss": 0.7695,
+      "step": 14946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.272095187814079e-07,
+      "loss": 0.918,
+      "step": 14947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.264205666156534e-07,
+      "loss": 0.8242,
+      "step": 14948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.256319339446384e-07,
+      "loss": 0.7461,
+      "step": 14949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.248436207961331e-07,
+      "loss": 0.8516,
+      "step": 14950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.240556271978973e-07,
+      "loss": 0.9062,
+      "step": 14951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.232679531776778e-07,
+      "loss": 0.8711,
+      "step": 14952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.224805987632057e-07,
+      "loss": 0.8867,
+      "step": 14953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.216935639822066e-07,
+      "loss": 0.9453,
+      "step": 14954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.209068488623952e-07,
+      "loss": 0.7852,
+      "step": 14955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.201204534314734e-07,
+      "loss": 0.7109,
+      "step": 14956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.193343777171282e-07,
+      "loss": 0.8047,
+      "step": 14957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.185486217470396e-07,
+      "loss": 0.8594,
+      "step": 14958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.177631855488778e-07,
+      "loss": 0.7891,
+      "step": 14959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.16978069150295e-07,
+      "loss": 0.7539,
+      "step": 14960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.16193272578938e-07,
+      "loss": 0.7461,
+      "step": 14961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.154087958624425e-07,
+      "loss": 0.7812,
+      "step": 14962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.146246390284308e-07,
+      "loss": 0.7969,
+      "step": 14963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.138408021045119e-07,
+      "loss": 0.7422,
+      "step": 14964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.130572851182872e-07,
+      "loss": 1.0156,
+      "step": 14965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.122740880973457e-07,
+      "loss": 0.7773,
+      "step": 14966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.114912110692676e-07,
+      "loss": 0.7812,
+      "step": 14967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.107086540616139e-07,
+      "loss": 0.9727,
+      "step": 14968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.099264171019418e-07,
+      "loss": 0.7773,
+      "step": 14969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.091445002177979e-07,
+      "loss": 0.8672,
+      "step": 14970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.083629034367114e-07,
+      "loss": 0.7734,
+      "step": 14971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.075816267862036e-07,
+      "loss": 0.7422,
+      "step": 14972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.068006702937859e-07,
+      "loss": 0.793,
+      "step": 14973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.060200339869574e-07,
+      "loss": 0.8008,
+      "step": 14974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.052397178932038e-07,
+      "loss": 0.832,
+      "step": 14975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.044597220400009e-07,
+      "loss": 0.918,
+      "step": 14976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 0.8477,
+      "step": 14977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.029006911651029e-07,
+      "loss": 0.8477,
+      "step": 14978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.021216561982993e-07,
+      "loss": 0.8477,
+      "step": 14979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.013429415818409e-07,
+      "loss": 0.8906,
+      "step": 14980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.005645473431446e-07,
+      "loss": 0.9336,
+      "step": 14981
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.997864735096218e-07,
+      "loss": 0.9961,
+      "step": 14982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.990087201086672e-07,
+      "loss": 0.8242,
+      "step": 14983
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.982312871676668e-07,
+      "loss": 0.8516,
+      "step": 14984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.974541747139975e-07,
+      "loss": 0.9141,
+      "step": 14985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.966773827750197e-07,
+      "loss": 0.7969,
+      "step": 14986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.95900911378087e-07,
+      "loss": 0.7539,
+      "step": 14987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.951247605505386e-07,
+      "loss": 0.875,
+      "step": 14988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.943489303197083e-07,
+      "loss": 0.8125,
+      "step": 14989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.935734207129087e-07,
+      "loss": 0.8906,
+      "step": 14990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.927982317574502e-07,
+      "loss": 0.9141,
+      "step": 14991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.920233634806263e-07,
+      "loss": 0.8594,
+      "step": 14992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.912488159097254e-07,
+      "loss": 0.8477,
+      "step": 14993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.904745890720146e-07,
+      "loss": 0.9297,
+      "step": 14994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.897006829947597e-07,
+      "loss": 0.7969,
+      "step": 14995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.889270977052089e-07,
+      "loss": 0.8359,
+      "step": 14996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.88153833230605e-07,
+      "loss": 0.8047,
+      "step": 14997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.873808895981717e-07,
+      "loss": 0.8047,
+      "step": 14998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.866082668351272e-07,
+      "loss": 0.9727,
+      "step": 14999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.858359649686776e-07,
+      "loss": 0.8438,
+      "step": 15000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.850639840260144e-07,
+      "loss": 0.832,
+      "step": 15001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.842923240343215e-07,
+      "loss": 0.8438,
+      "step": 15002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.835209850207716e-07,
+      "loss": 0.7305,
+      "step": 15003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.827499670125239e-07,
+      "loss": 0.957,
+      "step": 15004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.819792700367258e-07,
+      "loss": 0.8242,
+      "step": 15005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.812088941205166e-07,
+      "loss": 1.0312,
+      "step": 15006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.804388392910201e-07,
+      "loss": 0.8398,
+      "step": 15007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.796691055753559e-07,
+      "loss": 0.8125,
+      "step": 15008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.788996930006211e-07,
+      "loss": 1.0078,
+      "step": 15009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.781306015939117e-07,
+      "loss": 0.9102,
+      "step": 15010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773618313823084e-07,
+      "loss": 0.8086,
+      "step": 15011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.765933823928819e-07,
+      "loss": 0.8672,
+      "step": 15012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.758252546526868e-07,
+      "loss": 0.8047,
+      "step": 15013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750574481887719e-07,
+      "loss": 0.8281,
+      "step": 15014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.742899630281753e-07,
+      "loss": 0.7852,
+      "step": 15015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.735227991979178e-07,
+      "loss": 0.832,
+      "step": 15016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.727559567250132e-07,
+      "loss": 0.7656,
+      "step": 15017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.719894356364633e-07,
+      "loss": 0.8594,
+      "step": 15018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.712232359592621e-07,
+      "loss": 1.0391,
+      "step": 15019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.704573577203823e-07,
+      "loss": 0.7539,
+      "step": 15020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.696918009467959e-07,
+      "loss": 0.8633,
+      "step": 15021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.689265656654577e-07,
+      "loss": 0.9453,
+      "step": 15022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.681616519033165e-07,
+      "loss": 0.9062,
+      "step": 15023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.673970596873005e-07,
+      "loss": 0.8867,
+      "step": 15024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.666327890443338e-07,
+      "loss": 0.9219,
+      "step": 15025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.658688400013315e-07,
+      "loss": 0.7695,
+      "step": 15026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.651052125851888e-07,
+      "loss": 0.8789,
+      "step": 15027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.643419068227943e-07,
+      "loss": 0.8164,
+      "step": 15028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.635789227410285e-07,
+      "loss": 0.7773,
+      "step": 15029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.628162603667556e-07,
+      "loss": 0.7539,
+      "step": 15030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.620539197268296e-07,
+      "loss": 0.8906,
+      "step": 15031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.612919008480936e-07,
+      "loss": 0.8594,
+      "step": 15032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.605302037573793e-07,
+      "loss": 0.9766,
+      "step": 15033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.597688284815098e-07,
+      "loss": 0.8086,
+      "step": 15034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.590077750472903e-07,
+      "loss": 0.8125,
+      "step": 15035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.582470434815216e-07,
+      "loss": 0.7969,
+      "step": 15036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.574866338109889e-07,
+      "loss": 0.9102,
+      "step": 15037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.567265460624697e-07,
+      "loss": 0.8008,
+      "step": 15038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.559667802627236e-07,
+      "loss": 0.8125,
+      "step": 15039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.552073364385061e-07,
+      "loss": 0.8359,
+      "step": 15040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.544482146165578e-07,
+      "loss": 0.8477,
+      "step": 15041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.536894148236096e-07,
+      "loss": 0.8438,
+      "step": 15042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.529309370863781e-07,
+      "loss": 1.0859,
+      "step": 15043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.521727814315717e-07,
+      "loss": 0.7148,
+      "step": 15044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.514149478858847e-07,
+      "loss": 0.9258,
+      "step": 15045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.506574364760057e-07,
+      "loss": 0.8867,
+      "step": 15046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.499002472286022e-07,
+      "loss": 0.832,
+      "step": 15047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.491433801703386e-07,
+      "loss": 0.7305,
+      "step": 15048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.483868353278657e-07,
+      "loss": 0.7617,
+      "step": 15049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.476306127278244e-07,
+      "loss": 0.8672,
+      "step": 15050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.46874712396839e-07,
+      "loss": 0.8516,
+      "step": 15051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.461191343615271e-07,
+      "loss": 1.0156,
+      "step": 15052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.453638786484941e-07,
+      "loss": 0.8516,
+      "step": 15053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.446089452843354e-07,
+      "loss": 0.7734,
+      "step": 15054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.438543342956296e-07,
+      "loss": 0.8906,
+      "step": 15055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.431000457089489e-07,
+      "loss": 0.8711,
+      "step": 15056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.423460795508553e-07,
+      "loss": 0.9961,
+      "step": 15057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.415924358478966e-07,
+      "loss": 0.6797,
+      "step": 15058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.408391146266071e-07,
+      "loss": 0.8945,
+      "step": 15059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.400861159135143e-07,
+      "loss": 0.832,
+      "step": 15060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.393334397351327e-07,
+      "loss": 0.8945,
+      "step": 15061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.385810861179655e-07,
+      "loss": 0.8867,
+      "step": 15062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.378290550885027e-07,
+      "loss": 0.918,
+      "step": 15063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.37077346673224e-07,
+      "loss": 0.8281,
+      "step": 15064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.363259608986007e-07,
+      "loss": 1.0469,
+      "step": 15065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.355748977910894e-07,
+      "loss": 0.9375,
+      "step": 15066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.348241573771354e-07,
+      "loss": 0.8047,
+      "step": 15067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.340737396831733e-07,
+      "loss": 0.7344,
+      "step": 15068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.333236447356286e-07,
+      "loss": 0.7852,
+      "step": 15069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.3257387256091e-07,
+      "loss": 0.8008,
+      "step": 15070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.318244231854189e-07,
+      "loss": 0.8203,
+      "step": 15071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.310752966355462e-07,
+      "loss": 0.7617,
+      "step": 15072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.30326492937672e-07,
+      "loss": 0.8555,
+      "step": 15073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.295780121181563e-07,
+      "loss": 0.9258,
+      "step": 15074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.28829854203359e-07,
+      "loss": 0.8047,
+      "step": 15075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.280820192196226e-07,
+      "loss": 0.7891,
+      "step": 15076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.273345071932814e-07,
+      "loss": 0.918,
+      "step": 15077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.265873181506534e-07,
+      "loss": 0.832,
+      "step": 15078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.258404521180485e-07,
+      "loss": 0.8828,
+      "step": 15079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.250939091217669e-07,
+      "loss": 0.8633,
+      "step": 15080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.243476891880975e-07,
+      "loss": 0.7969,
+      "step": 15081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.236017923433104e-07,
+      "loss": 0.8086,
+      "step": 15082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.228562186136724e-07,
+      "loss": 0.9414,
+      "step": 15083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.221109680254358e-07,
+      "loss": 0.9102,
+      "step": 15084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.213660406048452e-07,
+      "loss": 0.832,
+      "step": 15085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.206214363781262e-07,
+      "loss": 0.8359,
+      "step": 15086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198771553715002e-07,
+      "loss": 0.8516,
+      "step": 15087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.191331976111728e-07,
+      "loss": 0.9219,
+      "step": 15088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.183895631233419e-07,
+      "loss": 0.8906,
+      "step": 15089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.176462519341899e-07,
+      "loss": 0.793,
+      "step": 15090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.169032640698904e-07,
+      "loss": 0.793,
+      "step": 15091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.161605995566058e-07,
+      "loss": 0.8711,
+      "step": 15092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.154182584204883e-07,
+      "loss": 0.8906,
+      "step": 15093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146762406876729e-07,
+      "loss": 0.9219,
+      "step": 15094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.139345463842895e-07,
+      "loss": 0.8984,
+      "step": 15095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.131931755364541e-07,
+      "loss": 0.7422,
+      "step": 15096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.124521281702724e-07,
+      "loss": 0.793,
+      "step": 15097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.117114043118357e-07,
+      "loss": 0.9062,
+      "step": 15098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.109710039872265e-07,
+      "loss": 0.9414,
+      "step": 15099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.102309272225173e-07,
+      "loss": 0.8086,
+      "step": 15100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.094911740437673e-07,
+      "loss": 0.8984,
+      "step": 15101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.08751744477021e-07,
+      "loss": 0.8203,
+      "step": 15102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.080126385483166e-07,
+      "loss": 0.9023,
+      "step": 15103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.072738562836801e-07,
+      "loss": 0.9648,
+      "step": 15104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.065353977091261e-07,
+      "loss": 0.7578,
+      "step": 15105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.05797262850655e-07,
+      "loss": 1.0312,
+      "step": 15106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.05059451734257e-07,
+      "loss": 0.8906,
+      "step": 15107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.043219643859124e-07,
+      "loss": 0.8477,
+      "step": 15108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.035848008315916e-07,
+      "loss": 0.8633,
+      "step": 15109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.028479610972473e-07,
+      "loss": 0.8008,
+      "step": 15110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.021114452088274e-07,
+      "loss": 0.8242,
+      "step": 15111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.013752531922647e-07,
+      "loss": 0.9648,
+      "step": 15112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.006393850734828e-07,
+      "loss": 0.7656,
+      "step": 15113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.99903840878391e-07,
+      "loss": 0.7812,
+      "step": 15114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.991686206328897e-07,
+      "loss": 0.8984,
+      "step": 15115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.98433724362867e-07,
+      "loss": 0.7422,
+      "step": 15116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.976991520942023e-07,
+      "loss": 0.9688,
+      "step": 15117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.969649038527572e-07,
+      "loss": 0.9102,
+      "step": 15118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.962309796643875e-07,
+      "loss": 0.9688,
+      "step": 15119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.954973795549348e-07,
+      "loss": 0.8672,
+      "step": 15120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.947641035502329e-07,
+      "loss": 0.8047,
+      "step": 15121
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.940311516760979e-07,
+      "loss": 0.7734,
+      "step": 15122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.932985239583402e-07,
+      "loss": 0.8789,
+      "step": 15123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.925662204227569e-07,
+      "loss": 0.7031,
+      "step": 15124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.918342410951351e-07,
+      "loss": 0.7656,
+      "step": 15125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.911025860012445e-07,
+      "loss": 0.8984,
+      "step": 15126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.903712551668508e-07,
+      "loss": 0.8203,
+      "step": 15127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.896402486177068e-07,
+      "loss": 0.8711,
+      "step": 15128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.889095663795487e-07,
+      "loss": 0.7891,
+      "step": 15129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.881792084781059e-07,
+      "loss": 0.8359,
+      "step": 15130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.874491749390967e-07,
+      "loss": 0.8047,
+      "step": 15131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.867194657882282e-07,
+      "loss": 0.9023,
+      "step": 15132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.859900810511912e-07,
+      "loss": 0.8086,
+      "step": 15133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.852610207536692e-07,
+      "loss": 0.793,
+      "step": 15134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.845322849213333e-07,
+      "loss": 0.8672,
+      "step": 15135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.83803873579847e-07,
+      "loss": 0.8945,
+      "step": 15136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.830757867548544e-07,
+      "loss": 0.8945,
+      "step": 15137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.823480244719939e-07,
+      "loss": 0.8477,
+      "step": 15138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.816205867568926e-07,
+      "loss": 0.9062,
+      "step": 15139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.808934736351637e-07,
+      "loss": 0.7969,
+      "step": 15140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.801666851324085e-07,
+      "loss": 0.8359,
+      "step": 15141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.794402212742203e-07,
+      "loss": 0.7266,
+      "step": 15142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.787140820861783e-07,
+      "loss": 0.8242,
+      "step": 15143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.779882675938532e-07,
+      "loss": 0.75,
+      "step": 15144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.77262777822797e-07,
+      "loss": 0.9023,
+      "step": 15145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.765376127985591e-07,
+      "loss": 0.8242,
+      "step": 15146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.758127725466724e-07,
+      "loss": 0.8438,
+      "step": 15147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.750882570926621e-07,
+      "loss": 0.8125,
+      "step": 15148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.743640664620366e-07,
+      "loss": 0.8438,
+      "step": 15149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.736402006802957e-07,
+      "loss": 0.7891,
+      "step": 15150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.729166597729287e-07,
+      "loss": 0.7852,
+      "step": 15151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.721934437654144e-07,
+      "loss": 0.793,
+      "step": 15152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.714705526832156e-07,
+      "loss": 0.8828,
+      "step": 15153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.707479865517864e-07,
+      "loss": 0.8984,
+      "step": 15154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.700257453965721e-07,
+      "loss": 0.9258,
+      "step": 15155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.693038292430022e-07,
+      "loss": 0.7969,
+      "step": 15156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.685822381164964e-07,
+      "loss": 0.6953,
+      "step": 15157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.678609720424624e-07,
+      "loss": 0.8555,
+      "step": 15158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.671400310462984e-07,
+      "loss": 0.918,
+      "step": 15159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.66419415153391e-07,
+      "loss": 0.8672,
+      "step": 15160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.65699124389111e-07,
+      "loss": 0.8047,
+      "step": 15161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.649791587788224e-07,
+      "loss": 0.7422,
+      "step": 15162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.642595183478763e-07,
+      "loss": 0.9531,
+      "step": 15163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.635402031216144e-07,
+      "loss": 0.8906,
+      "step": 15164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.62821213125361e-07,
+      "loss": 0.8945,
+      "step": 15165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.621025483844358e-07,
+      "loss": 0.8203,
+      "step": 15166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.613842089241419e-07,
+      "loss": 0.8633,
+      "step": 15167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.606661947697768e-07,
+      "loss": 0.8242,
+      "step": 15168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.599485059466183e-07,
+      "loss": 0.8281,
+      "step": 15169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.592311424799404e-07,
+      "loss": 0.7734,
+      "step": 15170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.585141043950006e-07,
+      "loss": 0.7773,
+      "step": 15171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.577973917170511e-07,
+      "loss": 0.8359,
+      "step": 15172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.570810044713228e-07,
+      "loss": 0.8555,
+      "step": 15173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.563649426830433e-07,
+      "loss": 0.8555,
+      "step": 15174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.556492063774257e-07,
+      "loss": 0.7344,
+      "step": 15175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.549337955796754e-07,
+      "loss": 0.9062,
+      "step": 15176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.54218710314979e-07,
+      "loss": 0.957,
+      "step": 15177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.535039506085162e-07,
+      "loss": 0.8477,
+      "step": 15178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.52789516485457e-07,
+      "loss": 0.8281,
+      "step": 15179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.52075407970958e-07,
+      "loss": 0.8086,
+      "step": 15180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.513616250901601e-07,
+      "loss": 0.8516,
+      "step": 15181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.506481678681999e-07,
+      "loss": 0.8398,
+      "step": 15182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.499350363302005e-07,
+      "loss": 0.8477,
+      "step": 15183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.492222305012686e-07,
+      "loss": 0.9297,
+      "step": 15184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.485097504065053e-07,
+      "loss": 0.8438,
+      "step": 15185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.477975960709982e-07,
+      "loss": 0.8086,
+      "step": 15186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.47085767519824e-07,
+      "loss": 0.9414,
+      "step": 15187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.463742647780447e-07,
+      "loss": 0.8164,
+      "step": 15188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.45663087870716e-07,
+      "loss": 0.8594,
+      "step": 15189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.449522368228779e-07,
+      "loss": 0.8516,
+      "step": 15190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.442417116595635e-07,
+      "loss": 0.8281,
+      "step": 15191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.435315124057874e-07,
+      "loss": 0.7852,
+      "step": 15192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.428216390865583e-07,
+      "loss": 0.8359,
+      "step": 15193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.421120917268732e-07,
+      "loss": 0.8086,
+      "step": 15194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.414028703517162e-07,
+      "loss": 0.918,
+      "step": 15195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.406939749860587e-07,
+      "loss": 0.8555,
+      "step": 15196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.399854056548616e-07,
+      "loss": 0.8555,
+      "step": 15197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.392771623830764e-07,
+      "loss": 0.9609,
+      "step": 15198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.385692451956427e-07,
+      "loss": 0.7734,
+      "step": 15199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.378616541174832e-07,
+      "loss": 0.8984,
+      "step": 15200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.371543891735156e-07,
+      "loss": 0.8242,
+      "step": 15201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.364474503886443e-07,
+      "loss": 0.8516,
+      "step": 15202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.357408377877629e-07,
+      "loss": 0.9141,
+      "step": 15203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.35034551395748e-07,
+      "loss": 0.9688,
+      "step": 15204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.34328591237472e-07,
+      "loss": 0.7305,
+      "step": 15205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.336229573377928e-07,
+      "loss": 0.8281,
+      "step": 15206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.329176497215584e-07,
+      "loss": 0.875,
+      "step": 15207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.322126684135999e-07,
+      "loss": 0.9453,
+      "step": 15208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.315080134387431e-07,
+      "loss": 0.9883,
+      "step": 15209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.308036848217992e-07,
+      "loss": 0.918,
+      "step": 15210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.300996825875706e-07,
+      "loss": 0.793,
+      "step": 15211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.293960067608441e-07,
+      "loss": 0.8867,
+      "step": 15212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286926573663966e-07,
+      "loss": 0.9297,
+      "step": 15213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.279896344289971e-07,
+      "loss": 0.8555,
+      "step": 15214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.272869379733993e-07,
+      "loss": 0.7578,
+      "step": 15215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.265845680243433e-07,
+      "loss": 0.7812,
+      "step": 15216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.258825246065637e-07,
+      "loss": 0.8594,
+      "step": 15217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.251808077447798e-07,
+      "loss": 0.7109,
+      "step": 15218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.244794174637004e-07,
+      "loss": 0.8359,
+      "step": 15219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.237783537880216e-07,
+      "loss": 0.8555,
+      "step": 15220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.23077616742428e-07,
+      "loss": 0.7266,
+      "step": 15221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.223772063515966e-07,
+      "loss": 0.832,
+      "step": 15222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.216771226401886e-07,
+      "loss": 0.8164,
+      "step": 15223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.209773656328544e-07,
+      "loss": 0.8203,
+      "step": 15224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.202779353542333e-07,
+      "loss": 0.832,
+      "step": 15225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.195788318289542e-07,
+      "loss": 0.8125,
+      "step": 15226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.188800550816355e-07,
+      "loss": 0.8164,
+      "step": 15227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.181816051368773e-07,
+      "loss": 0.832,
+      "step": 15228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.174834820192767e-07,
+      "loss": 0.8984,
+      "step": 15229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.167856857534139e-07,
+      "loss": 0.8359,
+      "step": 15230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.160882163638627e-07,
+      "loss": 0.8984,
+      "step": 15231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.15391073875179e-07,
+      "loss": 0.6953,
+      "step": 15232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.146942583119099e-07,
+      "loss": 0.9375,
+      "step": 15233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.139977696985922e-07,
+      "loss": 0.8281,
+      "step": 15234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.133016080597532e-07,
+      "loss": 0.8711,
+      "step": 15235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.12605773419901e-07,
+      "loss": 0.8242,
+      "step": 15236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.119102658035382e-07,
+      "loss": 0.8242,
+      "step": 15237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.112150852351585e-07,
+      "loss": 0.918,
+      "step": 15238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.105202317392356e-07,
+      "loss": 0.8672,
+      "step": 15239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.098257053402391e-07,
+      "loss": 0.8398,
+      "step": 15240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.091315060626225e-07,
+      "loss": 0.7812,
+      "step": 15241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.08437633930833e-07,
+      "loss": 0.8633,
+      "step": 15242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.077440889692977e-07,
+      "loss": 0.9648,
+      "step": 15243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.070508712024416e-07,
+      "loss": 0.8594,
+      "step": 15244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.063579806546716e-07,
+      "loss": 0.8711,
+      "step": 15245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.056654173503885e-07,
+      "loss": 0.7617,
+      "step": 15246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.049731813139748e-07,
+      "loss": 0.75,
+      "step": 15247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.042812725698067e-07,
+      "loss": 0.9141,
+      "step": 15248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.03589691142248e-07,
+      "loss": 0.8633,
+      "step": 15249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.028984370556514e-07,
+      "loss": 0.8438,
+      "step": 15250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.022075103343539e-07,
+      "loss": 0.7695,
+      "step": 15251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.015169110026853e-07,
+      "loss": 0.9336,
+      "step": 15252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.008266390849627e-07,
+      "loss": 0.875,
+      "step": 15253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.001366946054944e-07,
+      "loss": 0.8125,
+      "step": 15254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.994470775885709e-07,
+      "loss": 0.7891,
+      "step": 15255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.987577880584751e-07,
+      "loss": 1.0703,
+      "step": 15256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.980688260394785e-07,
+      "loss": 0.8008,
+      "step": 15257
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.973801915558432e-07,
+      "loss": 0.8242,
+      "step": 15258
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.966918846318115e-07,
+      "loss": 0.9102,
+      "step": 15259
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.960039052916235e-07,
+      "loss": 0.8438,
+      "step": 15260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.953162535595037e-07,
+      "loss": 0.8867,
+      "step": 15261
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.946289294596665e-07,
+      "loss": 0.7969,
+      "step": 15262
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.939419330163099e-07,
+      "loss": 0.7852,
+      "step": 15263
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.932552642536272e-07,
+      "loss": 0.8711,
+      "step": 15264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.925689231957966e-07,
+      "loss": 0.9688,
+      "step": 15265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.918829098669855e-07,
+      "loss": 0.7734,
+      "step": 15266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.911972242913468e-07,
+      "loss": 0.7734,
+      "step": 15267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.905118664930278e-07,
+      "loss": 0.7188,
+      "step": 15268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.898268364961591e-07,
+      "loss": 0.8203,
+      "step": 15269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.891421343248639e-07,
+      "loss": 0.8008,
+      "step": 15270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.884577600032494e-07,
+      "loss": 0.793,
+      "step": 15271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.87773713555413e-07,
+      "loss": 0.8828,
+      "step": 15272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.870899950054432e-07,
+      "loss": 0.8398,
+      "step": 15273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.864066043774142e-07,
+      "loss": 0.7266,
+      "step": 15274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.857235416953878e-07,
+      "loss": 0.8945,
+      "step": 15275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.850408069834169e-07,
+      "loss": 0.8086,
+      "step": 15276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.843584002655413e-07,
+      "loss": 0.7812,
+      "step": 15277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.836763215657915e-07,
+      "loss": 0.7188,
+      "step": 15278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.829945709081808e-07,
+      "loss": 0.8984,
+      "step": 15279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.823131483167167e-07,
+      "loss": 0.7344,
+      "step": 15280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.816320538153931e-07,
+      "loss": 0.832,
+      "step": 15281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.809512874281942e-07,
+      "loss": 0.7344,
+      "step": 15282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.802708491790878e-07,
+      "loss": 0.8086,
+      "step": 15283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.795907390920343e-07,
+      "loss": 0.9727,
+      "step": 15284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.789109571909813e-07,
+      "loss": 0.8594,
+      "step": 15285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.782315034998666e-07,
+      "loss": 0.8164,
+      "step": 15286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.775523780426118e-07,
+      "loss": 0.7734,
+      "step": 15287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.768735808431326e-07,
+      "loss": 0.8047,
+      "step": 15288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.761951119253285e-07,
+      "loss": 0.8047,
+      "step": 15289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.755169713130927e-07,
+      "loss": 0.918,
+      "step": 15290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.748391590302983e-07,
+      "loss": 0.7773,
+      "step": 15291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.741616751008162e-07,
+      "loss": 0.8555,
+      "step": 15292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.734845195485007e-07,
+      "loss": 0.8164,
+      "step": 15293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.728076923971971e-07,
+      "loss": 0.9531,
+      "step": 15294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.721311936707342e-07,
+      "loss": 0.8555,
+      "step": 15295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.714550233929329e-07,
+      "loss": 0.7891,
+      "step": 15296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.707791815876063e-07,
+      "loss": 0.7773,
+      "step": 15297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.701036682785467e-07,
+      "loss": 0.8594,
+      "step": 15298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.694284834895426e-07,
+      "loss": 0.8438,
+      "step": 15299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.687536272443685e-07,
+      "loss": 0.8359,
+      "step": 15300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.680790995667874e-07,
+      "loss": 0.9219,
+      "step": 15301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.674049004805483e-07,
+      "loss": 0.8008,
+      "step": 15302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.667310300093921e-07,
+      "loss": 0.8555,
+      "step": 15303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.660574881770476e-07,
+      "loss": 0.832,
+      "step": 15304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.653842750072315e-07,
+      "loss": 1.0547,
+      "step": 15305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.647113905236457e-07,
+      "loss": 0.7656,
+      "step": 15306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.64038834749986e-07,
+      "loss": 0.75,
+      "step": 15307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.633666077099332e-07,
+      "loss": 1.0234,
+      "step": 15308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.626947094271597e-07,
+      "loss": 0.8438,
+      "step": 15309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.620231399253208e-07,
+      "loss": 0.8398,
+      "step": 15310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.613518992280655e-07,
+      "loss": 0.8281,
+      "step": 15311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.606809873590269e-07,
+      "loss": 0.793,
+      "step": 15312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.600104043418343e-07,
+      "loss": 0.7812,
+      "step": 15313
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.593401502000929e-07,
+      "loss": 0.957,
+      "step": 15314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.586702249574073e-07,
+      "loss": 0.9922,
+      "step": 15315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.580006286373686e-07,
+      "loss": 0.9219,
+      "step": 15316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.57331361263549e-07,
+      "loss": 0.8438,
+      "step": 15317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.566624228595175e-07,
+      "loss": 1.0156,
+      "step": 15318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.559938134488286e-07,
+      "loss": 0.8359,
+      "step": 15319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.553255330550257e-07,
+      "loss": 0.9102,
+      "step": 15320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.546575817016377e-07,
+      "loss": 0.9375,
+      "step": 15321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.539899594121857e-07,
+      "loss": 0.8008,
+      "step": 15322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.533226662101766e-07,
+      "loss": 0.9062,
+      "step": 15323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.526557021191093e-07,
+      "loss": 0.8398,
+      "step": 15324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.519890671624662e-07,
+      "loss": 0.8477,
+      "step": 15325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.513227613637207e-07,
+      "loss": 0.9453,
+      "step": 15326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.506567847463352e-07,
+      "loss": 0.8047,
+      "step": 15327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.49991137333762e-07,
+      "loss": 0.9336,
+      "step": 15328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.493258191494345e-07,
+      "loss": 0.8125,
+      "step": 15329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.48660830216783e-07,
+      "loss": 0.9844,
+      "step": 15330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.479961705592231e-07,
+      "loss": 0.9648,
+      "step": 15331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.473318402001572e-07,
+      "loss": 0.8516,
+      "step": 15332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.466678391629766e-07,
+      "loss": 0.8906,
+      "step": 15333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.460041674710627e-07,
+      "loss": 0.8047,
+      "step": 15334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.453408251477867e-07,
+      "loss": 0.9023,
+      "step": 15335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.446778122165021e-07,
+      "loss": 0.8086,
+      "step": 15336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.440151287005558e-07,
+      "loss": 0.7812,
+      "step": 15337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.433527746232837e-07,
+      "loss": 0.9023,
+      "step": 15338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.426907500080071e-07,
+      "loss": 0.8281,
+      "step": 15339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.420290548780361e-07,
+      "loss": 0.8086,
+      "step": 15340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.413676892566711e-07,
+      "loss": 0.9023,
+      "step": 15341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.407066531671991e-07,
+      "loss": 0.9297,
+      "step": 15342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.400459466328979e-07,
+      "loss": 0.7812,
+      "step": 15343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.39385569677029e-07,
+      "loss": 0.9258,
+      "step": 15344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.387255223228472e-07,
+      "loss": 0.8008,
+      "step": 15345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.38065804593595e-07,
+      "loss": 0.7891,
+      "step": 15346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.374064165124982e-07,
+      "loss": 0.8008,
+      "step": 15347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.367473581027784e-07,
+      "loss": 0.832,
+      "step": 15348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.360886293876411e-07,
+      "loss": 0.8359,
+      "step": 15349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.354302303902815e-07,
+      "loss": 0.8867,
+      "step": 15350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.34772161133882e-07,
+      "loss": 0.8867,
+      "step": 15351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.34114421641614e-07,
+      "loss": 0.7734,
+      "step": 15352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.334570119366379e-07,
+      "loss": 0.8906,
+      "step": 15353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.327999320421041e-07,
+      "loss": 0.8828,
+      "step": 15354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.321431819811463e-07,
+      "loss": 0.8398,
+      "step": 15355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.314867617768905e-07,
+      "loss": 0.8203,
+      "step": 15356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.308306714524515e-07,
+      "loss": 0.7969,
+      "step": 15357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.301749110309319e-07,
+      "loss": 0.7773,
+      "step": 15358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.295194805354187e-07,
+      "loss": 0.7891,
+      "step": 15359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.288643799889926e-07,
+      "loss": 0.7852,
+      "step": 15360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.282096094147216e-07,
+      "loss": 0.9297,
+      "step": 15361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.275551688356596e-07,
+      "loss": 0.8164,
+      "step": 15362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.269010582748514e-07,
+      "loss": 0.8281,
+      "step": 15363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262472777553286e-07,
+      "loss": 0.832,
+      "step": 15364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.255938273001127e-07,
+      "loss": 0.9062,
+      "step": 15365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.249407069322122e-07,
+      "loss": 0.707,
+      "step": 15366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.24287916674624e-07,
+      "loss": 0.8906,
+      "step": 15367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.236354565503333e-07,
+      "loss": 0.8047,
+      "step": 15368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.229833265823182e-07,
+      "loss": 0.7734,
+      "step": 15369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.223315267935359e-07,
+      "loss": 0.7969,
+      "step": 15370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.216800572069404e-07,
+      "loss": 0.8555,
+      "step": 15371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.21028917845471e-07,
+      "loss": 0.8281,
+      "step": 15372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.203781087320526e-07,
+      "loss": 0.7578,
+      "step": 15373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.197276298896038e-07,
+      "loss": 0.7852,
+      "step": 15374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.190774813410283e-07,
+      "loss": 0.9336,
+      "step": 15375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.184276631092201e-07,
+      "loss": 0.8984,
+      "step": 15376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.177781752170576e-07,
+      "loss": 0.8203,
+      "step": 15377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.171290176874112e-07,
+      "loss": 0.7148,
+      "step": 15378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.164801905431394e-07,
+      "loss": 0.9805,
+      "step": 15379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.158316938070896e-07,
+      "loss": 0.9023,
+      "step": 15380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.151835275020946e-07,
+      "loss": 0.8789,
+      "step": 15381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.145356916509759e-07,
+      "loss": 0.8633,
+      "step": 15382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.138881862765478e-07,
+      "loss": 0.8359,
+      "step": 15383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.132410114016107e-07,
+      "loss": 0.957,
+      "step": 15384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.125941670489477e-07,
+      "loss": 0.918,
+      "step": 15385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.119476532413394e-07,
+      "loss": 0.6914,
+      "step": 15386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.113014700015496e-07,
+      "loss": 1.0781,
+      "step": 15387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.106556173523315e-07,
+      "loss": 0.8633,
+      "step": 15388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.100100953164256e-07,
+      "loss": 0.7891,
+      "step": 15389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.093649039165616e-07,
+      "loss": 0.8711,
+      "step": 15390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.087200431754592e-07,
+      "loss": 0.9023,
+      "step": 15391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.080755131158256e-07,
+      "loss": 0.8242,
+      "step": 15392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.074313137603516e-07,
+      "loss": 0.8281,
+      "step": 15393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.067874451317246e-07,
+      "loss": 0.8906,
+      "step": 15394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.061439072526143e-07,
+      "loss": 0.7617,
+      "step": 15395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.055007001456825e-07,
+      "loss": 0.8984,
+      "step": 15396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.048578238335745e-07,
+      "loss": 0.8281,
+      "step": 15397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.042152783389288e-07,
+      "loss": 0.7344,
+      "step": 15398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.035730636843695e-07,
+      "loss": 0.9609,
+      "step": 15399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.02931179892513e-07,
+      "loss": 0.8789,
+      "step": 15400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.022896269859568e-07,
+      "loss": 0.9805,
+      "step": 15401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.016484049872928e-07,
+      "loss": 0.8047,
+      "step": 15402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.010075139190997e-07,
+      "loss": 0.9375,
+      "step": 15403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.003669538039458e-07,
+      "loss": 0.8398,
+      "step": 15404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.997267246643823e-07,
+      "loss": 0.7773,
+      "step": 15405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.990868265229555e-07,
+      "loss": 0.8711,
+      "step": 15406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.984472594021973e-07,
+      "loss": 0.9141,
+      "step": 15407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.978080233246264e-07,
+      "loss": 0.8828,
+      "step": 15408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971691183127515e-07,
+      "loss": 0.8594,
+      "step": 15409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9653054438907e-07,
+      "loss": 0.9453,
+      "step": 15410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.958923015760687e-07,
+      "loss": 0.9844,
+      "step": 15411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.95254389896217e-07,
+      "loss": 0.8633,
+      "step": 15412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.946168093719795e-07,
+      "loss": 0.8633,
+      "step": 15413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.939795600258058e-07,
+      "loss": 0.8789,
+      "step": 15414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.933426418801369e-07,
+      "loss": 0.8008,
+      "step": 15415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.927060549573959e-07,
+      "loss": 0.7344,
+      "step": 15416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.920697992799984e-07,
+      "loss": 0.8281,
+      "step": 15417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.914338748703507e-07,
+      "loss": 0.8438,
+      "step": 15418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.907982817508429e-07,
+      "loss": 0.9062,
+      "step": 15419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.901630199438546e-07,
+      "loss": 0.8672,
+      "step": 15420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.895280894717559e-07,
+      "loss": 0.7656,
+      "step": 15421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.888934903569021e-07,
+      "loss": 0.9297,
+      "step": 15422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8825922262164e-07,
+      "loss": 0.7734,
+      "step": 15423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.876252862883025e-07,
+      "loss": 0.8047,
+      "step": 15424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.869916813792098e-07,
+      "loss": 0.8047,
+      "step": 15425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.86358407916674e-07,
+      "loss": 0.8828,
+      "step": 15426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.85725465922996e-07,
+      "loss": 0.8086,
+      "step": 15427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.850928554204571e-07,
+      "loss": 0.8672,
+      "step": 15428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.844605764313361e-07,
+      "loss": 0.8945,
+      "step": 15429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.83828628977895e-07,
+      "loss": 0.7734,
+      "step": 15430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.831970130823883e-07,
+      "loss": 0.8516,
+      "step": 15431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.825657287670527e-07,
+      "loss": 0.8125,
+      "step": 15432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.819347760541183e-07,
+      "loss": 1.0,
+      "step": 15433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.813041549658027e-07,
+      "loss": 0.8555,
+      "step": 15434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.806738655243105e-07,
+      "loss": 0.8438,
+      "step": 15435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.800439077518338e-07,
+      "loss": 0.7422,
+      "step": 15436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.794142816705561e-07,
+      "loss": 0.7812,
+      "step": 15437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.787849873026463e-07,
+      "loss": 0.7656,
+      "step": 15438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.781560246702644e-07,
+      "loss": 0.8047,
+      "step": 15439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.775273937955561e-07,
+      "loss": 0.832,
+      "step": 15440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.768990947006558e-07,
+      "loss": 0.75,
+      "step": 15441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.762711274076871e-07,
+      "loss": 0.9453,
+      "step": 15442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.756434919387643e-07,
+      "loss": 0.8086,
+      "step": 15443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.750161883159833e-07,
+      "loss": 0.8789,
+      "step": 15444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.74389216561435e-07,
+      "loss": 0.9141,
+      "step": 15445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.737625766971955e-07,
+      "loss": 0.8281,
+      "step": 15446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.731362687453301e-07,
+      "loss": 0.8477,
+      "step": 15447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.725102927278914e-07,
+      "loss": 0.7578,
+      "step": 15448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.718846486669216e-07,
+      "loss": 0.8906,
+      "step": 15449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.712593365844499e-07,
+      "loss": 0.9336,
+      "step": 15450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.706343565024952e-07,
+      "loss": 0.8555,
+      "step": 15451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.700097084430634e-07,
+      "loss": 0.8555,
+      "step": 15452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6938539242815e-07,
+      "loss": 0.7852,
+      "step": 15453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.687614084797366e-07,
+      "loss": 0.7422,
+      "step": 15454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.681377566197987e-07,
+      "loss": 0.7969,
+      "step": 15455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.675144368702901e-07,
+      "loss": 0.9492,
+      "step": 15456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.668914492531629e-07,
+      "loss": 0.8594,
+      "step": 15457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.662687937903521e-07,
+      "loss": 0.8359,
+      "step": 15458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.656464705037845e-07,
+      "loss": 0.8164,
+      "step": 15459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.650244794153692e-07,
+      "loss": 0.8906,
+      "step": 15460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.644028205470109e-07,
+      "loss": 0.8594,
+      "step": 15461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.637814939205966e-07,
+      "loss": 0.875,
+      "step": 15462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.631604995580065e-07,
+      "loss": 0.8633,
+      "step": 15463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.625398374811053e-07,
+      "loss": 0.8359,
+      "step": 15464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.619195077117468e-07,
+      "loss": 0.8438,
+      "step": 15465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.612995102717778e-07,
+      "loss": 0.8164,
+      "step": 15466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.606798451830231e-07,
+      "loss": 0.9805,
+      "step": 15467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.600605124673064e-07,
+      "loss": 0.9492,
+      "step": 15468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.594415121464336e-07,
+      "loss": 0.9023,
+      "step": 15469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.588228442422039e-07,
+      "loss": 0.75,
+      "step": 15470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.582045087763965e-07,
+      "loss": 0.8789,
+      "step": 15471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.575865057707864e-07,
+      "loss": 0.8711,
+      "step": 15472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.569688352471347e-07,
+      "loss": 0.9219,
+      "step": 15473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.563514972271921e-07,
+      "loss": 0.7227,
+      "step": 15474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.557344917326923e-07,
+      "loss": 0.9141,
+      "step": 15475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.551178187853634e-07,
+      "loss": 0.9219,
+      "step": 15476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.545014784069192e-07,
+      "loss": 0.9453,
+      "step": 15477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.538854706190633e-07,
+      "loss": 0.918,
+      "step": 15478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.532697954434829e-07,
+      "loss": 0.8398,
+      "step": 15479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.526544529018585e-07,
+      "loss": 0.8398,
+      "step": 15480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.520394430158571e-07,
+      "loss": 1.0,
+      "step": 15481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.51424765807137e-07,
+      "loss": 0.7148,
+      "step": 15482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.508104212973376e-07,
+      "loss": 0.8242,
+      "step": 15483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.501964095080925e-07,
+      "loss": 0.7969,
+      "step": 15484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.495827304610224e-07,
+      "loss": 0.8438,
+      "step": 15485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.489693841777366e-07,
+      "loss": 0.7539,
+      "step": 15486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.483563706798311e-07,
+      "loss": 0.6992,
+      "step": 15487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.477436899888899e-07,
+      "loss": 0.8047,
+      "step": 15488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.471313421264879e-07,
+      "loss": 0.7383,
+      "step": 15489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.465193271141888e-07,
+      "loss": 0.7656,
+      "step": 15490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.459076449735379e-07,
+      "loss": 0.8281,
+      "step": 15491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.452962957260765e-07,
+      "loss": 0.8203,
+      "step": 15492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.446852793933299e-07,
+      "loss": 0.8672,
+      "step": 15493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.440745959968152e-07,
+      "loss": 0.7734,
+      "step": 15494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.43464245558033e-07,
+      "loss": 0.8633,
+      "step": 15495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.42854228098475e-07,
+      "loss": 0.8047,
+      "step": 15496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.422445436396218e-07,
+      "loss": 0.7812,
+      "step": 15497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.416351922029417e-07,
+      "loss": 0.7461,
+      "step": 15498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.410261738098899e-07,
+      "loss": 0.832,
+      "step": 15499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.404174884819102e-07,
+      "loss": 0.8281,
+      "step": 15500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.398091362404368e-07,
+      "loss": 0.8867,
+      "step": 15501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.392011171068912e-07,
+      "loss": 0.8555,
+      "step": 15502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.385934311026808e-07,
+      "loss": 0.875,
+      "step": 15503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.379860782492042e-07,
+      "loss": 0.8672,
+      "step": 15504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.373790585678462e-07,
+      "loss": 0.7031,
+      "step": 15505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.367723720799833e-07,
+      "loss": 0.8789,
+      "step": 15506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.361660188069761e-07,
+      "loss": 0.7734,
+      "step": 15507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.355599987701743e-07,
+      "loss": 0.7969,
+      "step": 15508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.349543119909184e-07,
+      "loss": 0.9688,
+      "step": 15509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.34348958490537e-07,
+      "loss": 0.8711,
+      "step": 15510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.33743938290342e-07,
+      "loss": 0.8242,
+      "step": 15511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.331392514116384e-07,
+      "loss": 0.8789,
+      "step": 15512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.325348978757183e-07,
+      "loss": 0.8438,
+      "step": 15513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.319308777038645e-07,
+      "loss": 0.8086,
+      "step": 15514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.313271909173412e-07,
+      "loss": 0.9258,
+      "step": 15515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.30723837537408e-07,
+      "loss": 0.8047,
+      "step": 15516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.301208175853079e-07,
+      "loss": 0.8438,
+      "step": 15517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.295181310822773e-07,
+      "loss": 0.875,
+      "step": 15518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.289157780495335e-07,
+      "loss": 0.832,
+      "step": 15519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283137585082887e-07,
+      "loss": 0.8906,
+      "step": 15520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.277120724797424e-07,
+      "loss": 0.6953,
+      "step": 15521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.271107199850777e-07,
+      "loss": 0.8633,
+      "step": 15522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.265097010454712e-07,
+      "loss": 0.8438,
+      "step": 15523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.259090156820845e-07,
+      "loss": 0.8086,
+      "step": 15524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.25308663916071e-07,
+      "loss": 0.7031,
+      "step": 15525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.247086457685668e-07,
+      "loss": 0.8164,
+      "step": 15526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24108961260702e-07,
+      "loss": 0.8438,
+      "step": 15527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.235096104135895e-07,
+      "loss": 0.9023,
+      "step": 15528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.229105932483381e-07,
+      "loss": 0.7656,
+      "step": 15529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.223119097860352e-07,
+      "loss": 0.9688,
+      "step": 15530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.217135600477641e-07,
+      "loss": 0.8359,
+      "step": 15531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.211155440545923e-07,
+      "loss": 0.8398,
+      "step": 15532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.205178618275797e-07,
+      "loss": 0.9062,
+      "step": 15533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.199205133877672e-07,
+      "loss": 1.0312,
+      "step": 15534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.193234987561913e-07,
+      "loss": 0.9805,
+      "step": 15535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.187268179538729e-07,
+      "loss": 0.8086,
+      "step": 15536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.181304710018243e-07,
+      "loss": 0.7695,
+      "step": 15537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.175344579210396e-07,
+      "loss": 1.0391,
+      "step": 15538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.169387787325076e-07,
+      "loss": 0.8711,
+      "step": 15539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.163434334572026e-07,
+      "loss": 0.8008,
+      "step": 15540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.157484221160902e-07,
+      "loss": 0.7461,
+      "step": 15541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15153744730118e-07,
+      "loss": 0.9258,
+      "step": 15542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.145594013202271e-07,
+      "loss": 0.832,
+      "step": 15543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.139653919073451e-07,
+      "loss": 0.8906,
+      "step": 15544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.133717165123886e-07,
+      "loss": 0.9102,
+      "step": 15545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.12778375156261e-07,
+      "loss": 0.9141,
+      "step": 15546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.121853678598543e-07,
+      "loss": 0.7656,
+      "step": 15547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.115926946440508e-07,
+      "loss": 0.8438,
+      "step": 15548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.110003555297193e-07,
+      "loss": 0.7695,
+      "step": 15549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.104083505377144e-07,
+      "loss": 0.8125,
+      "step": 15550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.098166796888838e-07,
+      "loss": 0.957,
+      "step": 15551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.092253430040617e-07,
+      "loss": 0.8281,
+      "step": 15552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.086343405040695e-07,
+      "loss": 0.6953,
+      "step": 15553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.080436722097148e-07,
+      "loss": 0.8711,
+      "step": 15554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.07453338141799e-07,
+      "loss": 0.8359,
+      "step": 15555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.068633383211075e-07,
+      "loss": 0.7266,
+      "step": 15556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.062736727684159e-07,
+      "loss": 0.957,
+      "step": 15557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.056843415044854e-07,
+      "loss": 0.9102,
+      "step": 15558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.050953445500694e-07,
+      "loss": 0.8867,
+      "step": 15559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.045066819259048e-07,
+      "loss": 0.8125,
+      "step": 15560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.039183536527237e-07,
+      "loss": 0.8125,
+      "step": 15561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.033303597512374e-07,
+      "loss": 0.9453,
+      "step": 15562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.027427002421514e-07,
+      "loss": 0.9258,
+      "step": 15563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.021553751461594e-07,
+      "loss": 0.8359,
+      "step": 15564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.015683844839425e-07,
+      "loss": 0.8242,
+      "step": 15565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.009817282761675e-07,
+      "loss": 0.8633,
+      "step": 15566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.003954065434912e-07,
+      "loss": 0.8828,
+      "step": 15567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.998094193065605e-07,
+      "loss": 0.7031,
+      "step": 15568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9922376658601e-07,
+      "loss": 1.0,
+      "step": 15569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.986384484024575e-07,
+      "loss": 0.7734,
+      "step": 15570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.980534647765167e-07,
+      "loss": 0.918,
+      "step": 15571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.974688157287833e-07,
+      "loss": 0.8633,
+      "step": 15572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.968845012798462e-07,
+      "loss": 0.793,
+      "step": 15573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.963005214502781e-07,
+      "loss": 0.8789,
+      "step": 15574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.957168762606412e-07,
+      "loss": 0.9414,
+      "step": 15575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.951335657314893e-07,
+      "loss": 0.8203,
+      "step": 15576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.945505898833591e-07,
+      "loss": 0.9141,
+      "step": 15577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.939679487367787e-07,
+      "loss": 0.8047,
+      "step": 15578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.933856423122652e-07,
+      "loss": 0.8438,
+      "step": 15579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.928036706303208e-07,
+      "loss": 0.9258,
+      "step": 15580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.92222033711438e-07,
+      "loss": 0.7773,
+      "step": 15581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.916407315760974e-07,
+      "loss": 0.9453,
+      "step": 15582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.910597642447678e-07,
+      "loss": 0.8203,
+      "step": 15583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.904791317379065e-07,
+      "loss": 0.8633,
+      "step": 15584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898988340759558e-07,
+      "loss": 0.8359,
+      "step": 15585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.893188712793518e-07,
+      "loss": 0.707,
+      "step": 15586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.887392433685135e-07,
+      "loss": 0.8203,
+      "step": 15587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.881599503638546e-07,
+      "loss": 0.8359,
+      "step": 15588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875809922857677e-07,
+      "loss": 0.8594,
+      "step": 15589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.870023691546421e-07,
+      "loss": 0.7773,
+      "step": 15590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.864240809908505e-07,
+      "loss": 0.875,
+      "step": 15591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.858461278147563e-07,
+      "loss": 0.7734,
+      "step": 15592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.852685096467102e-07,
+      "loss": 0.9453,
+      "step": 15593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.846912265070491e-07,
+      "loss": 0.7812,
+      "step": 15594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.841142784161024e-07,
+      "loss": 0.7852,
+      "step": 15595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.835376653941859e-07,
+      "loss": 0.8281,
+      "step": 15596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829613874616001e-07,
+      "loss": 0.8125,
+      "step": 15597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.823854446386378e-07,
+      "loss": 0.9102,
+      "step": 15598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.818098369455793e-07,
+      "loss": 0.8398,
+      "step": 15599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.812345644026939e-07,
+      "loss": 0.9961,
+      "step": 15600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.806596270302355e-07,
+      "loss": 0.8789,
+      "step": 15601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.80085024848449e-07,
+      "loss": 0.7578,
+      "step": 15602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.795107578775671e-07,
+      "loss": 0.9727,
+      "step": 15603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.789368261378124e-07,
+      "loss": 0.7539,
+      "step": 15604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.783632296493923e-07,
+      "loss": 0.8633,
+      "step": 15605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.777899684325027e-07,
+      "loss": 1.1172,
+      "step": 15606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.772170425073319e-07,
+      "loss": 0.8281,
+      "step": 15607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.766444518940527e-07,
+      "loss": 0.7969,
+      "step": 15608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7607219661282566e-07,
+      "loss": 0.7734,
+      "step": 15609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7550027668380127e-07,
+      "loss": 0.8203,
+      "step": 15610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.749286921271168e-07,
+      "loss": 0.8828,
+      "step": 15611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7435744296290277e-07,
+      "loss": 0.7617,
+      "step": 15612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7378652921126867e-07,
+      "loss": 0.9258,
+      "step": 15613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7321595089231953e-07,
+      "loss": 0.8594,
+      "step": 15614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.726457080261448e-07,
+      "loss": 0.832,
+      "step": 15615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.720758006328274e-07,
+      "loss": 0.7344,
+      "step": 15616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7150622873243124e-07,
+      "loss": 0.9141,
+      "step": 15617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7093699234501246e-07,
+      "loss": 1.0156,
+      "step": 15618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7036809149061393e-07,
+      "loss": 0.7656,
+      "step": 15619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6979952618927074e-07,
+      "loss": 0.8555,
+      "step": 15620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.692312964610002e-07,
+      "loss": 0.8047,
+      "step": 15621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6866340232581077e-07,
+      "loss": 0.8516,
+      "step": 15622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6809584380369866e-07,
+      "loss": 0.8164,
+      "step": 15623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6752862091465013e-07,
+      "loss": 0.9727,
+      "step": 15624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6696173367863695e-07,
+      "loss": 0.7734,
+      "step": 15625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.663951821156187e-07,
+      "loss": 0.8633,
+      "step": 15626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6582896624554727e-07,
+      "loss": 0.8164,
+      "step": 15627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.652630860883589e-07,
+      "loss": 0.75,
+      "step": 15628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6469754166397875e-07,
+      "loss": 0.9062,
+      "step": 15629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6413233299231975e-07,
+      "loss": 0.8281,
+      "step": 15630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6356746009328603e-07,
+      "loss": 0.8984,
+      "step": 15631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6300292298676497e-07,
+      "loss": 0.8203,
+      "step": 15632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.624387216926363e-07,
+      "loss": 0.8281,
+      "step": 15633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.618748562307662e-07,
+      "loss": 0.75,
+      "step": 15634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.613113266210101e-07,
+      "loss": 0.793,
+      "step": 15635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.607481328832086e-07,
+      "loss": 1.125,
+      "step": 15636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.601852750371938e-07,
+      "loss": 0.8281,
+      "step": 15637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5962275310278524e-07,
+      "loss": 0.9062,
+      "step": 15638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.590605670997916e-07,
+      "loss": 0.8203,
+      "step": 15639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5849871704800484e-07,
+      "loss": 0.8203,
+      "step": 15640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5793720296721024e-07,
+      "loss": 0.8828,
+      "step": 15641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.573760248771797e-07,
+      "loss": 0.7852,
+      "step": 15642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5681518279767414e-07,
+      "loss": 0.9258,
+      "step": 15643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5625467674843995e-07,
+      "loss": 0.8984,
+      "step": 15644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5569450674921357e-07,
+      "loss": 0.9414,
+      "step": 15645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5513467281972035e-07,
+      "loss": 0.8594,
+      "step": 15646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.545751749796734e-07,
+      "loss": 0.9219,
+      "step": 15647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5401601324877143e-07,
+      "loss": 0.8828,
+      "step": 15648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.534571876467053e-07,
+      "loss": 0.7969,
+      "step": 15649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.528986981931505e-07,
+      "loss": 0.8984,
+      "step": 15650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.523405449077745e-07,
+      "loss": 0.7461,
+      "step": 15651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5178272781022823e-07,
+      "loss": 0.8711,
+      "step": 15652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5122524692015504e-07,
+      "loss": 0.9531,
+      "step": 15653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.506681022571846e-07,
+      "loss": 0.918,
+      "step": 15654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5011129384093466e-07,
+      "loss": 0.8203,
+      "step": 15655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.495548216910106e-07,
+      "loss": 0.75,
+      "step": 15656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4899868582700565e-07,
+      "loss": 0.7578,
+      "step": 15657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.484428862685053e-07,
+      "loss": 0.8555,
+      "step": 15658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.478874230350794e-07,
+      "loss": 0.8789,
+      "step": 15659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.473322961462845e-07,
+      "loss": 0.668,
+      "step": 15660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4677750562166833e-07,
+      "loss": 0.8438,
+      "step": 15661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.462230514807675e-07,
+      "loss": 0.8359,
+      "step": 15662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4566893374310305e-07,
+      "loss": 0.9258,
+      "step": 15663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4511515242818717e-07,
+      "loss": 0.9414,
+      "step": 15664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4456170755551973e-07,
+      "loss": 0.8398,
+      "step": 15665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.440085991445886e-07,
+      "loss": 0.8281,
+      "step": 15666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4345582721486926e-07,
+      "loss": 0.9766,
+      "step": 15667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4290339178582385e-07,
+      "loss": 0.9531,
+      "step": 15668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.423512928769069e-07,
+      "loss": 0.8242,
+      "step": 15669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4179953050755955e-07,
+      "loss": 0.8516,
+      "step": 15670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4124810469720723e-07,
+      "loss": 0.832,
+      "step": 15671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.406970154652679e-07,
+      "loss": 0.8711,
+      "step": 15672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.401462628311448e-07,
+      "loss": 0.6875,
+      "step": 15673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3959584681423473e-07,
+      "loss": 0.8516,
+      "step": 15674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.390457674339144e-07,
+      "loss": 0.7656,
+      "step": 15675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.384960247095549e-07,
+      "loss": 0.8359,
+      "step": 15676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.379466186605152e-07,
+      "loss": 0.8828,
+      "step": 15677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3739754930613665e-07,
+      "loss": 0.8398,
+      "step": 15678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3684881666575587e-07,
+      "loss": 0.8047,
+      "step": 15679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3630042075869297e-07,
+      "loss": 0.9336,
+      "step": 15680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.357523616042592e-07,
+      "loss": 0.875,
+      "step": 15681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.352046392217524e-07,
+      "loss": 0.8555,
+      "step": 15682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.346572536304572e-07,
+      "loss": 0.9609,
+      "step": 15683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.341102048496482e-07,
+      "loss": 0.8125,
+      "step": 15684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3356349289859104e-07,
+      "loss": 0.8711,
+      "step": 15685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3301711779653254e-07,
+      "loss": 0.9219,
+      "step": 15686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3247107956271184e-07,
+      "loss": 0.7969,
+      "step": 15687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3192537821635794e-07,
+      "loss": 0.9453,
+      "step": 15688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3138001377668436e-07,
+      "loss": 0.8594,
+      "step": 15689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3083498626289466e-07,
+      "loss": 0.7734,
+      "step": 15690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.30290295694179e-07,
+      "loss": 0.8828,
+      "step": 15691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.297459420897188e-07,
+      "loss": 0.8047,
+      "step": 15692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.292019254686786e-07,
+      "loss": 0.8281,
+      "step": 15693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2865824585021754e-07,
+      "loss": 1.0391,
+      "step": 15694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.281149032534759e-07,
+      "loss": 0.7461,
+      "step": 15695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2757189769758956e-07,
+      "loss": 0.7891,
+      "step": 15696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2702922920167536e-07,
+      "loss": 0.9453,
+      "step": 15697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2648689778484243e-07,
+      "loss": 0.7344,
+      "step": 15698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.259449034661878e-07,
+      "loss": 0.875,
+      "step": 15699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.254032462647961e-07,
+      "loss": 0.8672,
+      "step": 15700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.248619261997389e-07,
+      "loss": 0.9297,
+      "step": 15701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2432094329007634e-07,
+      "loss": 0.8125,
+      "step": 15702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2378029755485774e-07,
+      "loss": 0.8164,
+      "step": 15703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2323998901312333e-07,
+      "loss": 0.9844,
+      "step": 15704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2270001768389244e-07,
+      "loss": 0.7656,
+      "step": 15705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2216038358618203e-07,
+      "loss": 0.8516,
+      "step": 15706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.216210867389936e-07,
+      "loss": 0.8398,
+      "step": 15707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2108212716131525e-07,
+      "loss": 0.8242,
+      "step": 15708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.20543504872124e-07,
+      "loss": 0.8047,
+      "step": 15709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.20005219890387e-07,
+      "loss": 0.9414,
+      "step": 15710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.19467272235059e-07,
+      "loss": 0.8125,
+      "step": 15711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1892966192507935e-07,
+      "loss": 0.8281,
+      "step": 15712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.183923889793795e-07,
+      "loss": 0.8047,
+      "step": 15713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1785545341687773e-07,
+      "loss": 0.7891,
+      "step": 15714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1731885525648217e-07,
+      "loss": 0.7812,
+      "step": 15715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1678259451708226e-07,
+      "loss": 0.8516,
+      "step": 15716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.16246671217565e-07,
+      "loss": 0.7578,
+      "step": 15717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1571108537680094e-07,
+      "loss": 0.8594,
+      "step": 15718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1517583701364605e-07,
+      "loss": 0.832,
+      "step": 15719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.146409261469486e-07,
+      "loss": 0.8516,
+      "step": 15720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1410635279554467e-07,
+      "loss": 1.0156,
+      "step": 15721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1357211697825695e-07,
+      "loss": 0.9141,
+      "step": 15722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1303821871389593e-07,
+      "loss": 0.8828,
+      "step": 15723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1250465802126103e-07,
+      "loss": 0.793,
+      "step": 15724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.119714349191406e-07,
+      "loss": 0.8398,
+      "step": 15725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.114385494263107e-07,
+      "loss": 0.8945,
+      "step": 15726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1090600156153294e-07,
+      "loss": 0.793,
+      "step": 15727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1037379134356014e-07,
+      "loss": 0.7695,
+      "step": 15728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098419187911329e-07,
+      "loss": 0.9258,
+      "step": 15729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0931038392298064e-07,
+      "loss": 0.832,
+      "step": 15730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.08779186757815e-07,
+      "loss": 0.8594,
+      "step": 15731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.082483273143445e-07,
+      "loss": 0.7695,
+      "step": 15732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.077178056112596e-07,
+      "loss": 0.9258,
+      "step": 15733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0718762166724214e-07,
+      "loss": 0.8516,
+      "step": 15734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.066577755009582e-07,
+      "loss": 0.8828,
+      "step": 15735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0612826713106624e-07,
+      "loss": 0.8516,
+      "step": 15736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0559909657621135e-07,
+      "loss": 0.7773,
+      "step": 15737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 1.0156,
+      "step": 15738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0454176898613196e-07,
+      "loss": 0.8516,
+      "step": 15739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.040136119881366e-07,
+      "loss": 0.8398,
+      "step": 15740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0348579287963874e-07,
+      "loss": 0.8047,
+      "step": 15741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029583116792235e-07,
+      "loss": 0.8516,
+      "step": 15742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.024311684054627e-07,
+      "loss": 0.6836,
+      "step": 15743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.019043630769193e-07,
+      "loss": 0.918,
+      "step": 15744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.013778957121439e-07,
+      "loss": 0.8164,
+      "step": 15745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0085176632967183e-07,
+      "loss": 0.8203,
+      "step": 15746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.003259749480304e-07,
+      "loss": 0.9727,
+      "step": 15747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9980052158573255e-07,
+      "loss": 0.7812,
+      "step": 15748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.992754062612825e-07,
+      "loss": 0.7812,
+      "step": 15749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.987506289931686e-07,
+      "loss": 0.8906,
+      "step": 15750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.982261897998685e-07,
+      "loss": 0.8164,
+      "step": 15751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.977020886998484e-07,
+      "loss": 0.8242,
+      "step": 15752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.971783257115669e-07,
+      "loss": 0.8242,
+      "step": 15753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.966549008534604e-07,
+      "loss": 0.7461,
+      "step": 15754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9613181414396294e-07,
+      "loss": 0.8359,
+      "step": 15755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9560906560149327e-07,
+      "loss": 0.8008,
+      "step": 15756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9508665524445764e-07,
+      "loss": 0.8633,
+      "step": 15757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9456458309125034e-07,
+      "loss": 0.8086,
+      "step": 15758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9404284916025436e-07,
+      "loss": 0.9531,
+      "step": 15759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9352145346984173e-07,
+      "loss": 0.8477,
+      "step": 15760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.930003960383721e-07,
+      "loss": 0.8477,
+      "step": 15761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9247967688418987e-07,
+      "loss": 0.9531,
+      "step": 15762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9195929602563356e-07,
+      "loss": 0.793,
+      "step": 15763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.914392534810241e-07,
+      "loss": 0.8984,
+      "step": 15764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909195492686757e-07,
+      "loss": 0.875,
+      "step": 15765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.904001834068849e-07,
+      "loss": 0.7344,
+      "step": 15766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.898811559139404e-07,
+      "loss": 0.8086,
+      "step": 15767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.893624668081197e-07,
+      "loss": 0.8125,
+      "step": 15768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8884411610768503e-07,
+      "loss": 0.8047,
+      "step": 15769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.883261038308883e-07,
+      "loss": 0.7617,
+      "step": 15770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8780842999597056e-07,
+      "loss": 0.8516,
+      "step": 15771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8729109462115834e-07,
+      "loss": 0.7578,
+      "step": 15772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8677409772466925e-07,
+      "loss": 0.8359,
+      "step": 15773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.862574393247065e-07,
+      "loss": 0.7969,
+      "step": 15774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8574111943946335e-07,
+      "loss": 0.9102,
+      "step": 15775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8522513808711856e-07,
+      "loss": 0.8984,
+      "step": 15776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.847094952858443e-07,
+      "loss": 0.7812,
+      "step": 15777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8419419105379274e-07,
+      "loss": 0.918,
+      "step": 15778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.836792254091104e-07,
+      "loss": 0.9688,
+      "step": 15779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.831645983699306e-07,
+      "loss": 0.8555,
+      "step": 15780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8265030995437437e-07,
+      "loss": 0.9375,
+      "step": 15781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8213636018054834e-07,
+      "loss": 0.9531,
+      "step": 15782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.816227490665514e-07,
+      "loss": 0.918,
+      "step": 15783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.81109476630469e-07,
+      "loss": 0.875,
+      "step": 15784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.805965428903735e-07,
+      "loss": 0.7266,
+      "step": 15785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.800839478643259e-07,
+      "loss": 0.8008,
+      "step": 15786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.79571691570374e-07,
+      "loss": 0.793,
+      "step": 15787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.790597740265578e-07,
+      "loss": 0.8359,
+      "step": 15788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7854819525090294e-07,
+      "loss": 0.7383,
+      "step": 15789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.780369552614205e-07,
+      "loss": 0.8281,
+      "step": 15790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7752605407611274e-07,
+      "loss": 0.957,
+      "step": 15791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.770154917129698e-07,
+      "loss": 0.8125,
+      "step": 15792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7650526818997055e-07,
+      "loss": 0.9727,
+      "step": 15793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.759953835250785e-07,
+      "loss": 0.875,
+      "step": 15794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7548583773624805e-07,
+      "loss": 0.875,
+      "step": 15795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7497663084142156e-07,
+      "loss": 0.8281,
+      "step": 15796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.744677628585303e-07,
+      "loss": 0.7852,
+      "step": 15797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.739592338054887e-07,
+      "loss": 0.8672,
+      "step": 15798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.734510437002059e-07,
+      "loss": 0.875,
+      "step": 15799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7294319256057645e-07,
+      "loss": 0.8242,
+      "step": 15800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7243568040448043e-07,
+      "loss": 0.8594,
+      "step": 15801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.71928507249788e-07,
+      "loss": 0.8398,
+      "step": 15802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.714216731143594e-07,
+      "loss": 0.8477,
+      "step": 15803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.709151780160414e-07,
+      "loss": 0.832,
+      "step": 15804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.704090219726653e-07,
+      "loss": 0.7461,
+      "step": 15805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.699032050020568e-07,
+      "loss": 0.7852,
+      "step": 15806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.693977271220239e-07,
+      "loss": 0.8828,
+      "step": 15807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.68892588350368e-07,
+      "loss": 0.9336,
+      "step": 15808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.683877887048748e-07,
+      "loss": 0.8086,
+      "step": 15809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6788332820331674e-07,
+      "loss": 0.8594,
+      "step": 15810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6737920686345965e-07,
+      "loss": 0.7734,
+      "step": 15811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.668754247030548e-07,
+      "loss": 0.7109,
+      "step": 15812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.663719817398381e-07,
+      "loss": 0.9766,
+      "step": 15813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.658688779915387e-07,
+      "loss": 0.8203,
+      "step": 15814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6536611347587013e-07,
+      "loss": 0.8906,
+      "step": 15815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.648636882105394e-07,
+      "loss": 0.8906,
+      "step": 15816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6436160221323234e-07,
+      "loss": 0.7656,
+      "step": 15817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6385985550163036e-07,
+      "loss": 0.8672,
+      "step": 15818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.633584480934016e-07,
+      "loss": 0.7578,
+      "step": 15819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6285738000620296e-07,
+      "loss": 0.7773,
+      "step": 15820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.623566512576737e-07,
+      "loss": 0.7734,
+      "step": 15821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.618562618654464e-07,
+      "loss": 0.8828,
+      "step": 15822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6135621184714255e-07,
+      "loss": 0.8164,
+      "step": 15823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.608565012203691e-07,
+      "loss": 0.8164,
+      "step": 15824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6035713000271977e-07,
+      "loss": 0.8398,
+      "step": 15825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5985809821177944e-07,
+      "loss": 0.8047,
+      "step": 15826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.593594058651206e-07,
+      "loss": 0.8516,
+      "step": 15827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.588610529803027e-07,
+      "loss": 0.8906,
+      "step": 15828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.583630395748716e-07,
+      "loss": 0.8945,
+      "step": 15829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.578653656663655e-07,
+      "loss": 0.8203,
+      "step": 15830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5736803127230604e-07,
+      "loss": 0.8477,
+      "step": 15831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5687103641020794e-07,
+      "loss": 0.8125,
+      "step": 15832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.563743810975695e-07,
+      "loss": 0.8203,
+      "step": 15833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5587806535187787e-07,
+      "loss": 0.7891,
+      "step": 15834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.553820891906101e-07,
+      "loss": 0.7891,
+      "step": 15835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.548864526312312e-07,
+      "loss": 0.832,
+      "step": 15836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5439115569119164e-07,
+      "loss": 0.9102,
+      "step": 15837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5389619838793186e-07,
+      "loss": 0.8594,
+      "step": 15838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5340158073888133e-07,
+      "loss": 0.75,
+      "step": 15839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.529073027614549e-07,
+      "loss": 0.8906,
+      "step": 15840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.524133644730576e-07,
+      "loss": 0.9375,
+      "step": 15841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5191976589108113e-07,
+      "loss": 0.9297,
+      "step": 15842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.51426507032907e-07,
+      "loss": 0.8594,
+      "step": 15843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.509335879159037e-07,
+      "loss": 0.8672,
+      "step": 15844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5044100855742503e-07,
+      "loss": 0.8789,
+      "step": 15845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.499487689748182e-07,
+      "loss": 0.7383,
+      "step": 15846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.494568691854139e-07,
+      "loss": 0.8477,
+      "step": 15847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.489653092065359e-07,
+      "loss": 0.7734,
+      "step": 15848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.484740890554883e-07,
+      "loss": 0.8438,
+      "step": 15849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.479832087495705e-07,
+      "loss": 0.8047,
+      "step": 15850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4749266830606756e-07,
+      "loss": 0.8047,
+      "step": 15851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4700246774225124e-07,
+      "loss": 0.7578,
+      "step": 15852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4651260707538104e-07,
+      "loss": 0.8008,
+      "step": 15853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.460230863227065e-07,
+      "loss": 0.7891,
+      "step": 15854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4553390550146614e-07,
+      "loss": 0.7852,
+      "step": 15855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4504506462888387e-07,
+      "loss": 0.8359,
+      "step": 15856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.445565637221704e-07,
+      "loss": 0.8828,
+      "step": 15857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.440684027985286e-07,
+      "loss": 0.75,
+      "step": 15858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.435805818751481e-07,
+      "loss": 1.0547,
+      "step": 15859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.43093100969204e-07,
+      "loss": 0.8086,
+      "step": 15860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.426059600978626e-07,
+      "loss": 0.9453,
+      "step": 15861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4211915927827575e-07,
+      "loss": 0.7891,
+      "step": 15862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.416326985275853e-07,
+      "loss": 0.7227,
+      "step": 15863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.411465778629197e-07,
+      "loss": 0.8984,
+      "step": 15864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4066079730139647e-07,
+      "loss": 0.7188,
+      "step": 15865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.401753568601207e-07,
+      "loss": 0.7812,
+      "step": 15866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.396902565561866e-07,
+      "loss": 0.9102,
+      "step": 15867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3920549640667153e-07,
+      "loss": 0.8828,
+      "step": 15868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.387210764286486e-07,
+      "loss": 0.7266,
+      "step": 15869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3823699663917297e-07,
+      "loss": 0.8555,
+      "step": 15870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3775325705529105e-07,
+      "loss": 0.9023,
+      "step": 15871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.372698576940359e-07,
+      "loss": 0.8086,
+      "step": 15872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.367867985724271e-07,
+      "loss": 0.8477,
+      "step": 15873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3630407970747457e-07,
+      "loss": 0.8438,
+      "step": 15874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3582170111617796e-07,
+      "loss": 0.8477,
+      "step": 15875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.353396628155192e-07,
+      "loss": 0.8438,
+      "step": 15876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.348579648224737e-07,
+      "loss": 0.7852,
+      "step": 15877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.343766071540022e-07,
+      "loss": 0.8047,
+      "step": 15878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.338955898270546e-07,
+      "loss": 0.957,
+      "step": 15879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.334149128585673e-07,
+      "loss": 0.7617,
+      "step": 15880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3293457626546566e-07,
+      "loss": 0.9922,
+      "step": 15881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.324545800646628e-07,
+      "loss": 0.8242,
+      "step": 15882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.31974924273063e-07,
+      "loss": 0.8516,
+      "step": 15883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.314956089075516e-07,
+      "loss": 0.8086,
+      "step": 15884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.310166339850085e-07,
+      "loss": 0.8945,
+      "step": 15885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3053799952229903e-07,
+      "loss": 0.7461,
+      "step": 15886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3005970553627643e-07,
+      "loss": 0.8672,
+      "step": 15887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.295817520437805e-07,
+      "loss": 0.8555,
+      "step": 15888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.291041390616434e-07,
+      "loss": 0.7227,
+      "step": 15889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2862686660668053e-07,
+      "loss": 0.9336,
+      "step": 15890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2814993469569956e-07,
+      "loss": 0.7891,
+      "step": 15891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.276733433454915e-07,
+      "loss": 0.7148,
+      "step": 15892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.271970925728396e-07,
+      "loss": 0.7695,
+      "step": 15893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.267211823945127e-07,
+      "loss": 0.9336,
+      "step": 15894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.262456128272695e-07,
+      "loss": 0.7148,
+      "step": 15895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.257703838878523e-07,
+      "loss": 0.7969,
+      "step": 15896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2529549559299876e-07,
+      "loss": 0.7852,
+      "step": 15897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2482094795942666e-07,
+      "loss": 0.8945,
+      "step": 15898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2434674100385035e-07,
+      "loss": 0.8359,
+      "step": 15899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2387287474296203e-07,
+      "loss": 0.9844,
+      "step": 15900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2339934919344953e-07,
+      "loss": 0.8828,
+      "step": 15901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.229261643719872e-07,
+      "loss": 0.8008,
+      "step": 15902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2245332029523736e-07,
+      "loss": 0.875,
+      "step": 15903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.219808169798466e-07,
+      "loss": 0.832,
+      "step": 15904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.215086544424539e-07,
+      "loss": 1.0547,
+      "step": 15905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2103683269968597e-07,
+      "loss": 0.8438,
+      "step": 15906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2056535176815617e-07,
+      "loss": 0.8125,
+      "step": 15907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.200942116644634e-07,
+      "loss": 0.9062,
+      "step": 15908
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.196234124052e-07,
+      "loss": 0.9805,
+      "step": 15909
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1915295400694267e-07,
+      "loss": 0.8594,
+      "step": 15910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1868283648625817e-07,
+      "loss": 0.8359,
+      "step": 15911
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1821305985969775e-07,
+      "loss": 0.8477,
+      "step": 15912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1774362414380364e-07,
+      "loss": 0.8516,
+      "step": 15913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.172745293551072e-07,
+      "loss": 0.9922,
+      "step": 15914
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.168057755101239e-07,
+      "loss": 0.8555,
+      "step": 15915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1633736262535966e-07,
+      "loss": 0.5938,
+      "step": 15916
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.158692907173089e-07,
+      "loss": 0.8438,
+      "step": 15917
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.154015598024529e-07,
+      "loss": 0.8516,
+      "step": 15918
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.149341698972608e-07,
+      "loss": 0.8984,
+      "step": 15919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.144671210181904e-07,
+      "loss": 0.9609,
+      "step": 15920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.140004131816865e-07,
+      "loss": 0.7148,
+      "step": 15921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1353404640418474e-07,
+      "loss": 0.8125,
+      "step": 15922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1306802070210307e-07,
+      "loss": 0.9102,
+      "step": 15923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.12602336091854e-07,
+      "loss": 0.9102,
+      "step": 15924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1213699258983323e-07,
+      "loss": 0.7969,
+      "step": 15925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1167199021242877e-07,
+      "loss": 0.957,
+      "step": 15926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.112073289760098e-07,
+      "loss": 0.8008,
+      "step": 15927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.107430088969421e-07,
+      "loss": 0.8281,
+      "step": 15928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1027902999157146e-07,
+      "loss": 0.7969,
+      "step": 15929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.098153922762392e-07,
+      "loss": 0.8438,
+      "step": 15930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.09352095767268e-07,
+      "loss": 0.8945,
+      "step": 15931
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.088891404809713e-07,
+      "loss": 0.7969,
+      "step": 15932
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0842652643365055e-07,
+      "loss": 0.8516,
+      "step": 15933
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.079642536415972e-07,
+      "loss": 1.0625,
+      "step": 15934
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0750232212108487e-07,
+      "loss": 0.8359,
+      "step": 15935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.070407318883817e-07,
+      "loss": 0.8125,
+      "step": 15936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.065794829597413e-07,
+      "loss": 0.8516,
+      "step": 15937
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.061185753514029e-07,
+      "loss": 0.8438,
+      "step": 15938
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.056580090795969e-07,
+      "loss": 0.8125,
+      "step": 15939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0519778416054023e-07,
+      "loss": 0.8438,
+      "step": 15940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0473790061043893e-07,
+      "loss": 0.8203,
+      "step": 15941
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.042783584454856e-07,
+      "loss": 0.8008,
+      "step": 15942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0381915768186167e-07,
+      "loss": 0.7969,
+      "step": 15943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.033602983357342e-07,
+      "loss": 0.918,
+      "step": 15944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.029017804232637e-07,
+      "loss": 0.8438,
+      "step": 15945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.024436039605949e-07,
+      "loss": 0.8984,
+      "step": 15946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0198576896385835e-07,
+      "loss": 0.7539,
+      "step": 15947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.015282754491755e-07,
+      "loss": 0.7109,
+      "step": 15948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.01071123432658e-07,
+      "loss": 0.9375,
+      "step": 15949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0061431293040176e-07,
+      "loss": 0.8359,
+      "step": 15950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.001578439584907e-07,
+      "loss": 0.8398,
+      "step": 15951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.997017165329974e-07,
+      "loss": 0.8633,
+      "step": 15952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9924593066998463e-07,
+      "loss": 0.8984,
+      "step": 15953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.987904863855018e-07,
+      "loss": 0.9414,
+      "step": 15954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.983353836955838e-07,
+      "loss": 0.8398,
+      "step": 15955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.978806226162556e-07,
+      "loss": 0.9023,
+      "step": 15956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.974262031635311e-07,
+      "loss": 0.7109,
+      "step": 15957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.969721253534119e-07,
+      "loss": 0.8906,
+      "step": 15958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9651838920188415e-07,
+      "loss": 0.8906,
+      "step": 15959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9606499472492614e-07,
+      "loss": 0.8047,
+      "step": 15960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9561194193850174e-07,
+      "loss": 0.8867,
+      "step": 15961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.951592308585671e-07,
+      "loss": 0.8281,
+      "step": 15962
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9470686150105844e-07,
+      "loss": 0.7656,
+      "step": 15963
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.942548338819051e-07,
+      "loss": 0.918,
+      "step": 15964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.938031480170256e-07,
+      "loss": 0.8281,
+      "step": 15965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9335180392232485e-07,
+      "loss": 0.8711,
+      "step": 15966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.929008016136925e-07,
+      "loss": 0.832,
+      "step": 15967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.924501411070113e-07,
+      "loss": 0.918,
+      "step": 15968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9199982241814974e-07,
+      "loss": 0.875,
+      "step": 15969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.915498455629628e-07,
+      "loss": 0.8047,
+      "step": 15970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9110021055729466e-07,
+      "loss": 0.8125,
+      "step": 15971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9065091741697914e-07,
+      "loss": 0.9336,
+      "step": 15972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9020196615783705e-07,
+      "loss": 0.8516,
+      "step": 15973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.897533567956756e-07,
+      "loss": 0.8438,
+      "step": 15974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.893050893462901e-07,
+      "loss": 0.8281,
+      "step": 15975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8885716382546557e-07,
+      "loss": 0.8594,
+      "step": 15976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8840958024897505e-07,
+      "loss": 0.8867,
+      "step": 15977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8796233863257697e-07,
+      "loss": 0.8516,
+      "step": 15978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8751543899202094e-07,
+      "loss": 0.8359,
+      "step": 15979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.870688813430422e-07,
+      "loss": 0.8711,
+      "step": 15980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8662266570136487e-07,
+      "loss": 0.793,
+      "step": 15981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8617679208270075e-07,
+      "loss": 0.8828,
+      "step": 15982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.857312605027507e-07,
+      "loss": 0.9453,
+      "step": 15983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8528607097719986e-07,
+      "loss": 0.8086,
+      "step": 15984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.848412235217279e-07,
+      "loss": 0.9219,
+      "step": 15985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8439671815199576e-07,
+      "loss": 0.7422,
+      "step": 15986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839525548836564e-07,
+      "loss": 0.75,
+      "step": 15987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.835087337323483e-07,
+      "loss": 0.8164,
+      "step": 15988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.830652547137014e-07,
+      "loss": 0.8516,
+      "step": 15989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.826221178433286e-07,
+      "loss": 0.7734,
+      "step": 15990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.821793231368353e-07,
+      "loss": 0.8438,
+      "step": 15991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8173687060981116e-07,
+      "loss": 0.9062,
+      "step": 15992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.812947602778382e-07,
+      "loss": 0.8516,
+      "step": 15993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8085299215648174e-07,
+      "loss": 0.8711,
+      "step": 15994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.804115662612972e-07,
+      "loss": 1.0469,
+      "step": 15995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7997048260782866e-07,
+      "loss": 0.7305,
+      "step": 15996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7952974121160824e-07,
+      "loss": 0.7969,
+      "step": 15997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.790893420881535e-07,
+      "loss": 0.832,
+      "step": 15998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7864928525297097e-07,
+      "loss": 0.8203,
+      "step": 15999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7820957072155706e-07,
+      "loss": 0.8086,
+      "step": 16000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7777019850939615e-07,
+      "loss": 0.918,
+      "step": 16001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7733116863195684e-07,
+      "loss": 0.8086,
+      "step": 16002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.76892481104698e-07,
+      "loss": 0.8242,
+      "step": 16003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7645413594306725e-07,
+      "loss": 0.9023,
+      "step": 16004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.760161331625011e-07,
+      "loss": 0.8672,
+      "step": 16005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.755784727784194e-07,
+      "loss": 0.8633,
+      "step": 16006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7514115480623435e-07,
+      "loss": 0.8867,
+      "step": 16007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7470417926134583e-07,
+      "loss": 0.832,
+      "step": 16008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.742675461591371e-07,
+      "loss": 0.8711,
+      "step": 16009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7383125551498466e-07,
+      "loss": 0.8281,
+      "step": 16010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.733953073442508e-07,
+      "loss": 0.8398,
+      "step": 16011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.729597016622876e-07,
+      "loss": 0.793,
+      "step": 16012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7252443848443054e-07,
+      "loss": 0.7812,
+      "step": 16013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7208951782600746e-07,
+      "loss": 0.8516,
+      "step": 16014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.716549397023316e-07,
+      "loss": 0.7188,
+      "step": 16015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7122070412870737e-07,
+      "loss": 0.7539,
+      "step": 16016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.707868111204226e-07,
+      "loss": 1.0703,
+      "step": 16017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.70353260692755e-07,
+      "loss": 0.8906,
+      "step": 16018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6992005286097246e-07,
+      "loss": 0.8398,
+      "step": 16019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6948718764032935e-07,
+      "loss": 0.9141,
+      "step": 16020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.690546650460646e-07,
+      "loss": 0.7578,
+      "step": 16021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6862248509341047e-07,
+      "loss": 0.8477,
+      "step": 16022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6819064779758373e-07,
+      "loss": 0.8281,
+      "step": 16023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6775915317379e-07,
+      "loss": 0.8047,
+      "step": 16024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.673280012372226e-07,
+      "loss": 0.8203,
+      "step": 16025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6689719200306384e-07,
+      "loss": 0.7422,
+      "step": 16026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.664667254864828e-07,
+      "loss": 0.9258,
+      "step": 16027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6603660170263725e-07,
+      "loss": 0.8477,
+      "step": 16028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6560682066667175e-07,
+      "loss": 0.8047,
+      "step": 16029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6517738239371984e-07,
+      "loss": 0.7891,
+      "step": 16030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.647482868989037e-07,
+      "loss": 0.8086,
+      "step": 16031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.643195341973304e-07,
+      "loss": 0.8555,
+      "step": 16032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.638911243040987e-07,
+      "loss": 0.9141,
+      "step": 16033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.634630572342922e-07,
+      "loss": 0.6875,
+      "step": 16034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6303533300298665e-07,
+      "loss": 0.9102,
+      "step": 16035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.626079516252389e-07,
+      "loss": 0.8242,
+      "step": 16036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6218091311610017e-07,
+      "loss": 0.8516,
+      "step": 16037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.617542174906074e-07,
+      "loss": 0.8828,
+      "step": 16038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.61327864763784e-07,
+      "loss": 0.8008,
+      "step": 16039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6090185495064256e-07,
+      "loss": 0.8516,
+      "step": 16040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6047618806618324e-07,
+      "loss": 0.8438,
+      "step": 16041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6005086412539736e-07,
+      "loss": 0.8125,
+      "step": 16042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.596258831432574e-07,
+      "loss": 0.9688,
+      "step": 16043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5920124513472923e-07,
+      "loss": 0.832,
+      "step": 16044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5877695011476523e-07,
+      "loss": 0.7461,
+      "step": 16045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.583529980983057e-07,
+      "loss": 1.0625,
+      "step": 16046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.579293891002776e-07,
+      "loss": 0.7578,
+      "step": 16047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5750612313559685e-07,
+      "loss": 0.8125,
+      "step": 16048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5708320021916923e-07,
+      "loss": 0.8945,
+      "step": 16049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5666062036588504e-07,
+      "loss": 0.8086,
+      "step": 16050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5623838359062346e-07,
+      "loss": 0.8555,
+      "step": 16051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5581648990825267e-07,
+      "loss": 0.793,
+      "step": 16052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.553949393336286e-07,
+      "loss": 0.9219,
+      "step": 16053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.549737318815937e-07,
+      "loss": 0.7188,
+      "step": 16054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.545528675669806e-07,
+      "loss": 0.8906,
+      "step": 16055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5413234640460747e-07,
+      "loss": 0.8398,
+      "step": 16056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.537121684092836e-07,
+      "loss": 0.8945,
+      "step": 16057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.532923335958004e-07,
+      "loss": 0.8984,
+      "step": 16058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.528728419789439e-07,
+      "loss": 0.8477,
+      "step": 16059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5245369357348336e-07,
+      "loss": 0.8086,
+      "step": 16060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5203488839418035e-07,
+      "loss": 1.0234,
+      "step": 16061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5161642645577856e-07,
+      "loss": 0.8164,
+      "step": 16062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.511983077730129e-07,
+      "loss": 0.8398,
+      "step": 16063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.507805323606094e-07,
+      "loss": 0.8477,
+      "step": 16064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5036310023327403e-07,
+      "loss": 0.7539,
+      "step": 16065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.499460114057073e-07,
+      "loss": 0.8086,
+      "step": 16066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.495292658925963e-07,
+      "loss": 0.8672,
+      "step": 16067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.491128637086149e-07,
+      "loss": 1.0,
+      "step": 16068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.486968048684246e-07,
+      "loss": 0.8906,
+      "step": 16069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4828108938667496e-07,
+      "loss": 0.7656,
+      "step": 16070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4786571727800525e-07,
+      "loss": 0.7539,
+      "step": 16071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.474506885570416e-07,
+      "loss": 0.7578,
+      "step": 16072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.470360032383967e-07,
+      "loss": 0.8047,
+      "step": 16073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.466216613366723e-07,
+      "loss": 0.7578,
+      "step": 16074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.462076628664578e-07,
+      "loss": 0.7852,
+      "step": 16075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4579400784233266e-07,
+      "loss": 0.875,
+      "step": 16076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4538069627886074e-07,
+      "loss": 0.793,
+      "step": 16077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4496772819059487e-07,
+      "loss": 0.7422,
+      "step": 16078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.445551035920768e-07,
+      "loss": 0.7461,
+      "step": 16079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4414282249783703e-07,
+      "loss": 0.8359,
+      "step": 16080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.437308849223907e-07,
+      "loss": 0.7852,
+      "step": 16081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.43319290880244e-07,
+      "loss": 0.7969,
+      "step": 16082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4290804038588966e-07,
+      "loss": 0.9258,
+      "step": 16083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.424971334538073e-07,
+      "loss": 0.8438,
+      "step": 16084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.420865700984665e-07,
+      "loss": 0.8828,
+      "step": 16085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4167635033432335e-07,
+      "loss": 0.8008,
+      "step": 16086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4126647417582414e-07,
+      "loss": 0.9805,
+      "step": 16087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4085694163739846e-07,
+      "loss": 0.8281,
+      "step": 16088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.404477527334692e-07,
+      "loss": 0.7969,
+      "step": 16089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.400389074784415e-07,
+      "loss": 0.8789,
+      "step": 16090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3963040588671493e-07,
+      "loss": 0.8984,
+      "step": 16091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.392222479726713e-07,
+      "loss": 0.7188,
+      "step": 16092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3881443375068257e-07,
+      "loss": 0.7812,
+      "step": 16093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3840696323510826e-07,
+      "loss": 0.7812,
+      "step": 16094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.37999836440298e-07,
+      "loss": 0.8828,
+      "step": 16095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3759305338058591e-07,
+      "loss": 0.7852,
+      "step": 16096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3718661407029497e-07,
+      "loss": 0.8164,
+      "step": 16097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3678051852373596e-07,
+      "loss": 0.832,
+      "step": 16098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3637476675521187e-07,
+      "loss": 0.7578,
+      "step": 16099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3596935877900573e-07,
+      "loss": 0.8242,
+      "step": 16100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.355642946093939e-07,
+      "loss": 0.9297,
+      "step": 16101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.351595742606394e-07,
+      "loss": 0.8516,
+      "step": 16102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.347551977469942e-07,
+      "loss": 0.8398,
+      "step": 16103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3435116508269463e-07,
+      "loss": 0.7695,
+      "step": 16104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3394747628196933e-07,
+      "loss": 0.8047,
+      "step": 16105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3354413135903142e-07,
+      "loss": 0.8086,
+      "step": 16106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3314113032808617e-07,
+      "loss": 0.9609,
+      "step": 16107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3273847320331999e-07,
+      "loss": 0.8477,
+      "step": 16108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3233615999891267e-07,
+      "loss": 0.8438,
+      "step": 16109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3193419072903066e-07,
+      "loss": 0.9453,
+      "step": 16110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.315325654078282e-07,
+      "loss": 0.8828,
+      "step": 16111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.311312840494462e-07,
+      "loss": 0.9922,
+      "step": 16112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3073034666801443e-07,
+      "loss": 0.8242,
+      "step": 16113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.303297532776505e-07,
+      "loss": 0.8047,
+      "step": 16114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2992950389246204e-07,
+      "loss": 0.9453,
+      "step": 16115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2952959852653888e-07,
+      "loss": 0.8164,
+      "step": 16116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2913003719396532e-07,
+      "loss": 0.8906,
+      "step": 16117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2873081990880784e-07,
+      "loss": 0.8164,
+      "step": 16118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2833194668512747e-07,
+      "loss": 0.8281,
+      "step": 16119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.279334175369652e-07,
+      "loss": 0.8672,
+      "step": 16120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2753523247835418e-07,
+      "loss": 0.8672,
+      "step": 16121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.271373915233177e-07,
+      "loss": 0.8008,
+      "step": 16122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2673989468586343e-07,
+      "loss": 0.7656,
+      "step": 16123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2634274197998574e-07,
+      "loss": 0.9102,
+      "step": 16124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.259459334196723e-07,
+      "loss": 0.7383,
+      "step": 16125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2554946901889197e-07,
+      "loss": 0.8125,
+      "step": 16126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.25153348791608e-07,
+      "loss": 0.8594,
+      "step": 16127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.247575727517659e-07,
+      "loss": 0.7461,
+      "step": 16128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2436214091330344e-07,
+      "loss": 0.8164,
+      "step": 16129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.239670532901428e-07,
+      "loss": 0.7031,
+      "step": 16130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2357230989619728e-07,
+      "loss": 0.7852,
+      "step": 16131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2317791074536465e-07,
+      "loss": 0.9102,
+      "step": 16132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2278385585153273e-07,
+      "loss": 0.9141,
+      "step": 16133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2239014522857816e-07,
+      "loss": 0.7812,
+      "step": 16134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.219967788903632e-07,
+      "loss": 0.8672,
+      "step": 16135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2160375685073897e-07,
+      "loss": 0.8555,
+      "step": 16136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2121107912354333e-07,
+      "loss": 0.8047,
+      "step": 16137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.208187457226052e-07,
+      "loss": 1.0078,
+      "step": 16138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2042675666173685e-07,
+      "loss": 0.8398,
+      "step": 16139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.200351119547428e-07,
+      "loss": 0.9609,
+      "step": 16140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1964381161541203e-07,
+      "loss": 0.9492,
+      "step": 16141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1925285565752464e-07,
+      "loss": 0.8867,
+      "step": 16142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1886224409484514e-07,
+      "loss": 0.832,
+      "step": 16143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.18471976941127e-07,
+      "loss": 0.7852,
+      "step": 16144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1808205421011363e-07,
+      "loss": 0.832,
+      "step": 16145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1769247591553522e-07,
+      "loss": 0.8633,
+      "step": 16146
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1730324207110852e-07,
+      "loss": 0.8789,
+      "step": 16147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1691435269053817e-07,
+      "loss": 0.8281,
+      "step": 16148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1652580778751875e-07,
+      "loss": 0.8828,
+      "step": 16149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1613760737573265e-07,
+      "loss": 0.875,
+      "step": 16150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1574975146884558e-07,
+      "loss": 0.8359,
+      "step": 16151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1536224008051775e-07,
+      "loss": 0.7969,
+      "step": 16152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1497507322439272e-07,
+      "loss": 0.8945,
+      "step": 16153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1458825091410395e-07,
+      "loss": 0.6445,
+      "step": 16154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1420177316327063e-07,
+      "loss": 0.8867,
+      "step": 16155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1381563998550182e-07,
+      "loss": 0.8359,
+      "step": 16156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1342985139439555e-07,
+      "loss": 0.7383,
+      "step": 16157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1304440740353426e-07,
+      "loss": 0.7383,
+      "step": 16158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1265930802648937e-07,
+      "loss": 0.8008,
+      "step": 16159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1227455327682223e-07,
+      "loss": 0.7656,
+      "step": 16160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1189014316808087e-07,
+      "loss": 0.793,
+      "step": 16161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1150607771380004e-07,
+      "loss": 0.8555,
+      "step": 16162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1112235692750338e-07,
+      "loss": 0.9531,
+      "step": 16163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1073898082270338e-07,
+      "loss": 0.8125,
+      "step": 16164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1035594941289704e-07,
+      "loss": 0.8281,
+      "step": 16165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.099732627115747e-07,
+      "loss": 0.9062,
+      "step": 16166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0959092073220777e-07,
+      "loss": 0.8008,
+      "step": 16167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0920892348826215e-07,
+      "loss": 0.8398,
+      "step": 16168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0882727099318602e-07,
+      "loss": 0.8203,
+      "step": 16169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0844596326042077e-07,
+      "loss": 0.8086,
+      "step": 16170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0806500030339016e-07,
+      "loss": 0.75,
+      "step": 16171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0768438213550902e-07,
+      "loss": 1.0078,
+      "step": 16172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0730410877018102e-07,
+      "loss": 0.7578,
+      "step": 16173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0692418022079552e-07,
+      "loss": 0.7773,
+      "step": 16174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0654459650072955e-07,
+      "loss": 0.7734,
+      "step": 16175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0616535762335023e-07,
+      "loss": 0.8828,
+      "step": 16176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0578646360200905e-07,
+      "loss": 0.7148,
+      "step": 16177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0540791445004982e-07,
+      "loss": 0.8711,
+      "step": 16178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0502971018080075e-07,
+      "loss": 0.8359,
+      "step": 16179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0465185080757787e-07,
+      "loss": 0.8203,
+      "step": 16180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.042743363436872e-07,
+      "loss": 0.7461,
+      "step": 16181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0389716680242368e-07,
+      "loss": 0.7617,
+      "step": 16182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.035203421970644e-07,
+      "loss": 0.7852,
+      "step": 16183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0314386254087991e-07,
+      "loss": 0.8438,
+      "step": 16184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0276772784712627e-07,
+      "loss": 0.8867,
+      "step": 16185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0239193812904845e-07,
+      "loss": 0.8984,
+      "step": 16186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0201649339987694e-07,
+      "loss": 0.9492,
+      "step": 16187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0164139367283342e-07,
+      "loss": 0.8594,
+      "step": 16188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0126663896112398e-07,
+      "loss": 0.8789,
+      "step": 16189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0089222927794693e-07,
+      "loss": 0.875,
+      "step": 16190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.00518164636484e-07,
+      "loss": 0.875,
+      "step": 16191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0014444504990572e-07,
+      "loss": 0.8711,
+      "step": 16192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.997710705313738e-07,
+      "loss": 0.7891,
+      "step": 16193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9939804109403327e-07,
+      "loss": 0.9688,
+      "step": 16194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.990253567510203e-07,
+      "loss": 0.793,
+      "step": 16195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9865301751545661e-07,
+      "loss": 0.793,
+      "step": 16196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9828102340045397e-07,
+      "loss": 0.875,
+      "step": 16197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9790937441910963e-07,
+      "loss": 0.8086,
+      "step": 16198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.975380705845109e-07,
+      "loss": 1.0469,
+      "step": 16199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9716711190973182e-07,
+      "loss": 0.8125,
+      "step": 16200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.967964984078352e-07,
+      "loss": 0.875,
+      "step": 16201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9642623009186845e-07,
+      "loss": 0.832,
+      "step": 16202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.960563069748711e-07,
+      "loss": 0.8086,
+      "step": 16203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.956867290698683e-07,
+      "loss": 0.7539,
+      "step": 16204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9531749638987518e-07,
+      "loss": 0.8516,
+      "step": 16205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9494860894788915e-07,
+      "loss": 0.6797,
+      "step": 16206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.945800667569031e-07,
+      "loss": 0.9492,
+      "step": 16207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9421186982989116e-07,
+      "loss": 0.7969,
+      "step": 16208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9384401817982069e-07,
+      "loss": 0.8906,
+      "step": 16209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9347651181964245e-07,
+      "loss": 0.9297,
+      "step": 16210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9310935076229719e-07,
+      "loss": 0.7969,
+      "step": 16211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.927425350207135e-07,
+      "loss": 0.7812,
+      "step": 16212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.923760646078088e-07,
+      "loss": 0.7617,
+      "step": 16213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9200993953648494e-07,
+      "loss": 0.7383,
+      "step": 16214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.916441598196339e-07,
+      "loss": 0.8125,
+      "step": 16215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.912787254701365e-07,
+      "loss": 0.7812,
+      "step": 16216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9091363650086015e-07,
+      "loss": 0.8672,
+      "step": 16217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9054889292466016e-07,
+      "loss": 0.8672,
+      "step": 16218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.901844947543785e-07,
+      "loss": 0.8594,
+      "step": 16219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.898204420028471e-07,
+      "loss": 0.8672,
+      "step": 16220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8945673468288573e-07,
+      "loss": 0.7695,
+      "step": 16221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8909337280729967e-07,
+      "loss": 0.8477,
+      "step": 16222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8873035638888314e-07,
+      "loss": 0.7148,
+      "step": 16223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.883676854404204e-07,
+      "loss": 0.8164,
+      "step": 16224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8800535997468005e-07,
+      "loss": 0.8359,
+      "step": 16225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8764338000442083e-07,
+      "loss": 0.8633,
+      "step": 16226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872817455423881e-07,
+      "loss": 0.8711,
+      "step": 16227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8692045660131607e-07,
+      "loss": 0.8945,
+      "step": 16228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8655951319392685e-07,
+      "loss": 0.8711,
+      "step": 16229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.861989153329291e-07,
+      "loss": 0.8672,
+      "step": 16230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8583866303101937e-07,
+      "loss": 0.8984,
+      "step": 16231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8547875630088307e-07,
+      "loss": 0.9961,
+      "step": 16232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8511919515519449e-07,
+      "loss": 0.8672,
+      "step": 16233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.847599796066124e-07,
+      "loss": 0.7969,
+      "step": 16234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8440110966778557e-07,
+      "loss": 0.7969,
+      "step": 16235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8404258535135054e-07,
+      "loss": 0.8789,
+      "step": 16236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8368440666993393e-07,
+      "loss": 0.8594,
+      "step": 16237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8332657363614448e-07,
+      "loss": 0.7188,
+      "step": 16238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8296908626258325e-07,
+      "loss": 0.7148,
+      "step": 16239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8261194456183796e-07,
+      "loss": 0.8984,
+      "step": 16240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.822551485464852e-07,
+      "loss": 0.7109,
+      "step": 16241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.81898698229086e-07,
+      "loss": 0.7969,
+      "step": 16242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.815425936221926e-07,
+      "loss": 0.8945,
+      "step": 16243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8118683473834388e-07,
+      "loss": 0.8633,
+      "step": 16244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8083142159006862e-07,
+      "loss": 0.8633,
+      "step": 16245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8047635418987797e-07,
+      "loss": 0.8438,
+      "step": 16246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.801216325502775e-07,
+      "loss": 0.9297,
+      "step": 16247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7976725668375493e-07,
+      "loss": 0.7266,
+      "step": 16248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.794132266027915e-07,
+      "loss": 0.8438,
+      "step": 16249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7905954231984935e-07,
+      "loss": 0.9258,
+      "step": 16250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7870620384738524e-07,
+      "loss": 0.7383,
+      "step": 16251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7835321119784033e-07,
+      "loss": 0.75,
+      "step": 16252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7800056438364244e-07,
+      "loss": 0.8359,
+      "step": 16253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7764826341720942e-07,
+      "loss": 0.8594,
+      "step": 16254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7729630831094691e-07,
+      "loss": 0.7695,
+      "step": 16255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7694469907724831e-07,
+      "loss": 0.7578,
+      "step": 16256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7659343572849375e-07,
+      "loss": 0.7695,
+      "step": 16257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7624251827705108e-07,
+      "loss": 0.8047,
+      "step": 16258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.758919467352771e-07,
+      "loss": 0.7305,
+      "step": 16259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7554172111551636e-07,
+      "loss": 0.8203,
+      "step": 16260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7519184143010015e-07,
+      "loss": 0.918,
+      "step": 16261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7484230769134858e-07,
+      "loss": 0.8594,
+      "step": 16262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7449311991156959e-07,
+      "loss": 0.8477,
+      "step": 16263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.741442781030589e-07,
+      "loss": 0.8125,
+      "step": 16264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.737957822780989e-07,
+      "loss": 0.7031,
+      "step": 16265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7344763244896091e-07,
+      "loss": 0.7734,
+      "step": 16266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7309982862790398e-07,
+      "loss": 0.7266,
+      "step": 16267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.72752370827175e-07,
+      "loss": 0.8633,
+      "step": 16268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7240525905900863e-07,
+      "loss": 0.9414,
+      "step": 16269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7205849333562509e-07,
+      "loss": 0.8789,
+      "step": 16270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7171207366923792e-07,
+      "loss": 0.8711,
+      "step": 16271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7136600007204407e-07,
+      "loss": 0.957,
+      "step": 16272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.710202725562271e-07,
+      "loss": 0.8906,
+      "step": 16273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7067489113396285e-07,
+      "loss": 0.875,
+      "step": 16274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7032985581741156e-07,
+      "loss": 0.7656,
+      "step": 16275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6998516661872466e-07,
+      "loss": 0.8047,
+      "step": 16276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6964082355003574e-07,
+      "loss": 0.8672,
+      "step": 16277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.692968266234718e-07,
+      "loss": 0.8516,
+      "step": 16278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6895317585114645e-07,
+      "loss": 0.8281,
+      "step": 16279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.686098712451578e-07,
+      "loss": 0.832,
+      "step": 16280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6826691281759512e-07,
+      "loss": 0.8125,
+      "step": 16281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6792430058053533e-07,
+      "loss": 0.9414,
+      "step": 16282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6758203454604105e-07,
+      "loss": 0.8633,
+      "step": 16283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6724011472616486e-07,
+      "loss": 0.7266,
+      "step": 16284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6689854113294602e-07,
+      "loss": 0.8867,
+      "step": 16285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6655731377841044e-07,
+      "loss": 0.7891,
+      "step": 16286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6621643267457631e-07,
+      "loss": 0.9141,
+      "step": 16287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6587589783344516e-07,
+      "loss": 0.8203,
+      "step": 16288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6553570926700624e-07,
+      "loss": 0.7656,
+      "step": 16289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6519586698724e-07,
+      "loss": 0.9219,
+      "step": 16290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.648563710061113e-07,
+      "loss": 0.9492,
+      "step": 16291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6451722133557613e-07,
+      "loss": 0.7773,
+      "step": 16292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6417841798757494e-07,
+      "loss": 0.7656,
+      "step": 16293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6383996097403932e-07,
+      "loss": 0.8984,
+      "step": 16294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6350185030688416e-07,
+      "loss": 0.7305,
+      "step": 16295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.631640859980177e-07,
+      "loss": 0.7617,
+      "step": 16296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6282666805933044e-07,
+      "loss": 0.8008,
+      "step": 16297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6248959650270623e-07,
+      "loss": 0.8633,
+      "step": 16298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.621528713400111e-07,
+      "loss": 1.0156,
+      "step": 16299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.618164925831045e-07,
+      "loss": 0.8203,
+      "step": 16300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6148046024382803e-07,
+      "loss": 0.9023,
+      "step": 16301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6114477433401443e-07,
+      "loss": 0.8906,
+      "step": 16302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6080943486548538e-07,
+      "loss": 0.8398,
+      "step": 16303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6047444185004923e-07,
+      "loss": 0.7617,
+      "step": 16304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6013979529949875e-07,
+      "loss": 0.7578,
+      "step": 16305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5980549522561894e-07,
+      "loss": 0.8555,
+      "step": 16306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5947154164018154e-07,
+      "loss": 0.8281,
+      "step": 16307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5913793455494375e-07,
+      "loss": 0.8633,
+      "step": 16308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.58804673981654e-07,
+      "loss": 0.7617,
+      "step": 16309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5847175993204622e-07,
+      "loss": 0.9258,
+      "step": 16310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5813919241784325e-07,
+      "loss": 0.8789,
+      "step": 16311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5780697145075462e-07,
+      "loss": 0.8047,
+      "step": 16312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5747509704247987e-07,
+      "loss": 0.8398,
+      "step": 16313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5714356920470297e-07,
+      "loss": 0.9492,
+      "step": 16314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5681238794909903e-07,
+      "loss": 0.7773,
+      "step": 16315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.564815532873276e-07,
+      "loss": 0.9609,
+      "step": 16316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5615106523104052e-07,
+      "loss": 1.0312,
+      "step": 16317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5582092379187174e-07,
+      "loss": 0.875,
+      "step": 16318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.554911289814487e-07,
+      "loss": 1.0,
+      "step": 16319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5516168081138316e-07,
+      "loss": 0.793,
+      "step": 16320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5483257929327478e-07,
+      "loss": 0.9102,
+      "step": 16321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5450382443871094e-07,
+      "loss": 0.9023,
+      "step": 16322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5417541625927124e-07,
+      "loss": 0.8008,
+      "step": 16323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5384735476651535e-07,
+      "loss": 0.957,
+      "step": 16324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5351963997199625e-07,
+      "loss": 0.7812,
+      "step": 16325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5319227188725472e-07,
+      "loss": 0.7656,
+      "step": 16326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5286525052381708e-07,
+      "loss": 0.7539,
+      "step": 16327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5253857589319633e-07,
+      "loss": 1.0078,
+      "step": 16328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5221224800689772e-07,
+      "loss": 0.9609,
+      "step": 16329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5188626687640985e-07,
+      "loss": 1.0,
+      "step": 16330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.515606325132135e-07,
+      "loss": 0.9062,
+      "step": 16331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.512353449287729e-07,
+      "loss": 0.8242,
+      "step": 16332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5091040413454105e-07,
+      "loss": 0.8281,
+      "step": 16333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5058581014196105e-07,
+      "loss": 0.9727,
+      "step": 16334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5026156296246374e-07,
+      "loss": 0.875,
+      "step": 16335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4993766260746333e-07,
+      "loss": 0.8984,
+      "step": 16336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4961410908836627e-07,
+      "loss": 0.9141,
+      "step": 16337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4929090241656564e-07,
+      "loss": 0.7812,
+      "step": 16338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4896804260344232e-07,
+      "loss": 0.7734,
+      "step": 16339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4864552966036394e-07,
+      "loss": 0.7891,
+      "step": 16340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4832336359868693e-07,
+      "loss": 0.957,
+      "step": 16341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4800154442975444e-07,
+      "loss": 0.8672,
+      "step": 16342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4768007216490078e-07,
+      "loss": 0.793,
+      "step": 16343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4735894681544238e-07,
+      "loss": 0.7617,
+      "step": 16344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4703816839268693e-07,
+      "loss": 1.0312,
+      "step": 16345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.46717736907932e-07,
+      "loss": 0.9102,
+      "step": 16346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4639765237245862e-07,
+      "loss": 0.8203,
+      "step": 16347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4607791479753774e-07,
+      "loss": 0.8828,
+      "step": 16348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4575852419442815e-07,
+      "loss": 0.7852,
+      "step": 16349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.454394805743753e-07,
+      "loss": 0.7891,
+      "step": 16350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4512078394861462e-07,
+      "loss": 0.8516,
+      "step": 16351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4480243432836716e-07,
+      "loss": 0.8008,
+      "step": 16352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4448443172484062e-07,
+      "loss": 0.7227,
+      "step": 16353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.44166776149236e-07,
+      "loss": 0.8789,
+      "step": 16354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4384946761273554e-07,
+      "loss": 0.8008,
+      "step": 16355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4353250612651358e-07,
+      "loss": 0.8477,
+      "step": 16356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4321589170173011e-07,
+      "loss": 0.9844,
+      "step": 16357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4289962434953508e-07,
+      "loss": 0.875,
+      "step": 16358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.425837040810618e-07,
+      "loss": 0.7422,
+      "step": 16359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4226813090743697e-07,
+      "loss": 0.8125,
+      "step": 16360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4195290483977165e-07,
+      "loss": 0.875,
+      "step": 16361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4163802588916587e-07,
+      "loss": 0.8398,
+      "step": 16362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4132349406670521e-07,
+      "loss": 0.7891,
+      "step": 16363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4100930938346635e-07,
+      "loss": 0.7969,
+      "step": 16364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4069547185051268e-07,
+      "loss": 0.8125,
+      "step": 16365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4038198147889425e-07,
+      "loss": 0.875,
+      "step": 16366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.400688382796478e-07,
+      "loss": 0.957,
+      "step": 16367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3975604226380224e-07,
+      "loss": 0.7344,
+      "step": 16368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3944359344237214e-07,
+      "loss": 0.8906,
+      "step": 16369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3913149182635643e-07,
+      "loss": 0.8008,
+      "step": 16370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3881973742674525e-07,
+      "loss": 0.7188,
+      "step": 16371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.385083302545176e-07,
+      "loss": 0.7109,
+      "step": 16372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.38197270320638e-07,
+      "loss": 0.8008,
+      "step": 16373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3788655763605885e-07,
+      "loss": 0.8867,
+      "step": 16374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3757619221172025e-07,
+      "loss": 0.9141,
+      "step": 16375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3726617405855237e-07,
+      "loss": 0.8555,
+      "step": 16376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.369565031874709e-07,
+      "loss": 0.875,
+      "step": 16377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3664717960937822e-07,
+      "loss": 0.8633,
+      "step": 16378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.363382033351679e-07,
+      "loss": 0.8594,
+      "step": 16379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3602957437571896e-07,
+      "loss": 0.8789,
+      "step": 16380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3572129274189938e-07,
+      "loss": 0.7578,
+      "step": 16381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3541335844456273e-07,
+      "loss": 0.8164,
+      "step": 16382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.351057714945525e-07,
+      "loss": 0.7812,
+      "step": 16383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.347985319027001e-07,
+      "loss": 0.8203,
+      "step": 16384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3449163967982237e-07,
+      "loss": 0.8594,
+      "step": 16385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3418509483672625e-07,
+      "loss": 0.8047,
+      "step": 16386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3387889738420644e-07,
+      "loss": 0.8867,
+      "step": 16387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3357304733304431e-07,
+      "loss": 0.9062,
+      "step": 16388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3326754469400793e-07,
+      "loss": 0.8516,
+      "step": 16389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3296238947785534e-07,
+      "loss": 0.6914,
+      "step": 16390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3265758169533238e-07,
+      "loss": 0.8984,
+      "step": 16391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3235312135717048e-07,
+      "loss": 0.8359,
+      "step": 16392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3204900847409109e-07,
+      "loss": 0.8242,
+      "step": 16393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3174524305680224e-07,
+      "loss": 0.8906,
+      "step": 16394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.314418251159999e-07,
+      "loss": 0.8281,
+      "step": 16395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.311387546623688e-07,
+      "loss": 0.9062,
+      "step": 16396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.308360317065782e-07,
+      "loss": 0.8125,
+      "step": 16397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.305336562592896e-07,
+      "loss": 0.7852,
+      "step": 16398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3023162833114888e-07,
+      "loss": 0.8516,
+      "step": 16399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.29929947932792e-07,
+      "loss": 0.7773,
+      "step": 16400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2962861507484048e-07,
+      "loss": 0.7891,
+      "step": 16401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.293276297679058e-07,
+      "loss": 0.8047,
+      "step": 16402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2902699202258505e-07,
+      "loss": 0.9141,
+      "step": 16403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2872670184946535e-07,
+      "loss": 0.7344,
+      "step": 16404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.284267592591193e-07,
+      "loss": 0.8516,
+      "step": 16405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2812716426210736e-07,
+      "loss": 0.7969,
+      "step": 16406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2782791686898223e-07,
+      "loss": 0.7344,
+      "step": 16407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2752901709027766e-07,
+      "loss": 0.7969,
+      "step": 16408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.272304649365197e-07,
+      "loss": 0.8789,
+      "step": 16409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2693226041822104e-07,
+      "loss": 0.793,
+      "step": 16410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2663440354588108e-07,
+      "loss": 0.8125,
+      "step": 16411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.263368943299881e-07,
+      "loss": 0.8945,
+      "step": 16412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.26039732781017e-07,
+      "loss": 0.8906,
+      "step": 16413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.257429189094339e-07,
+      "loss": 0.8281,
+      "step": 16414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2544645272568822e-07,
+      "loss": 0.7734,
+      "step": 16415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2515033424021828e-07,
+      "loss": 0.9492,
+      "step": 16416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2485456346345237e-07,
+      "loss": 0.8086,
+      "step": 16417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.245591404058044e-07,
+      "loss": 0.9062,
+      "step": 16418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.242640650776772e-07,
+      "loss": 0.7383,
+      "step": 16419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2396933748945904e-07,
+      "loss": 0.7734,
+      "step": 16420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.236749576515295e-07,
+      "loss": 0.8359,
+      "step": 16421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2338092557425462e-07,
+      "loss": 0.8906,
+      "step": 16422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.230872412679862e-07,
+      "loss": 0.8086,
+      "step": 16423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2279390474306596e-07,
+      "loss": 0.8086,
+      "step": 16424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.225009160098234e-07,
+      "loss": 0.8516,
+      "step": 16425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2220827507857358e-07,
+      "loss": 0.875,
+      "step": 16426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2191598195962163e-07,
+      "loss": 0.7852,
+      "step": 16427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.216240366632604e-07,
+      "loss": 0.8828,
+      "step": 16428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2133243919976835e-07,
+      "loss": 0.7969,
+      "step": 16429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2104118957941392e-07,
+      "loss": 0.8867,
+      "step": 16430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2075028781245225e-07,
+      "loss": 0.8008,
+      "step": 16431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2045973390912625e-07,
+      "loss": 0.8125,
+      "step": 16432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2016952787966773e-07,
+      "loss": 0.9375,
+      "step": 16433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.198796697342941e-07,
+      "loss": 0.8359,
+      "step": 16434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1959015948321162e-07,
+      "loss": 0.8711,
+      "step": 16435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1930099713661546e-07,
+      "loss": 0.8633,
+      "step": 16436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.190121827046875e-07,
+      "loss": 0.8906,
+      "step": 16437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1872371619759626e-07,
+      "loss": 0.8047,
+      "step": 16438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1843559762549917e-07,
+      "loss": 0.7852,
+      "step": 16439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1814782699854144e-07,
+      "loss": 0.9414,
+      "step": 16440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.178604043268583e-07,
+      "loss": 0.8789,
+      "step": 16441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1757332962056611e-07,
+      "loss": 0.7539,
+      "step": 16442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1728660288977678e-07,
+      "loss": 0.7227,
+      "step": 16443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1700022414458446e-07,
+      "loss": 0.8984,
+      "step": 16444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1671419339507329e-07,
+      "loss": 0.918,
+      "step": 16445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1642851065131632e-07,
+      "loss": 0.8477,
+      "step": 16446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1614317592336999e-07,
+      "loss": 0.7773,
+      "step": 16447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.15858189221284e-07,
+      "loss": 0.75,
+      "step": 16448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1557355055509256e-07,
+      "loss": 0.8438,
+      "step": 16449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1528925993481765e-07,
+      "loss": 0.7695,
+      "step": 16450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1500531737046905e-07,
+      "loss": 0.8203,
+      "step": 16451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1472172287204653e-07,
+      "loss": 0.8359,
+      "step": 16452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1443847644953543e-07,
+      "loss": 0.7969,
+      "step": 16453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1415557811290778e-07,
+      "loss": 0.6914,
+      "step": 16454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1387302787212673e-07,
+      "loss": 0.7695,
+      "step": 16455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1359082573713986e-07,
+      "loss": 0.7891,
+      "step": 16456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.133089717178848e-07,
+      "loss": 0.9219,
+      "step": 16457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1302746582428581e-07,
+      "loss": 0.7617,
+      "step": 16458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1274630806625497e-07,
+      "loss": 0.793,
+      "step": 16459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1246549845369325e-07,
+      "loss": 0.9531,
+      "step": 16460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1218503699648719e-07,
+      "loss": 0.8359,
+      "step": 16461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1190492370451333e-07,
+      "loss": 0.7812,
+      "step": 16462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1162515858763268e-07,
+      "loss": 0.8672,
+      "step": 16463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1134574165569956e-07,
+      "loss": 0.8008,
+      "step": 16464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1106667291855056e-07,
+      "loss": 0.8281,
+      "step": 16465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1078795238601115e-07,
+      "loss": 0.7422,
+      "step": 16466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1050958006789791e-07,
+      "loss": 0.8008,
+      "step": 16467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1023155597401191e-07,
+      "loss": 0.9141,
+      "step": 16468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0995388011414199e-07,
+      "loss": 0.8281,
+      "step": 16469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0967655249806697e-07,
+      "loss": 0.8984,
+      "step": 16470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0939957313555017e-07,
+      "loss": 0.7383,
+      "step": 16471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0912294203634599e-07,
+      "loss": 0.8086,
+      "step": 16472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0884665921019444e-07,
+      "loss": 0.8008,
+      "step": 16473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0857072466682328e-07,
+      "loss": 0.9336,
+      "step": 16474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.082951384159503e-07,
+      "loss": 0.8125,
+      "step": 16475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0801990046727772e-07,
+      "loss": 0.7617,
+      "step": 16476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0774501083049782e-07,
+      "loss": 0.7578,
+      "step": 16477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.074704695152895e-07,
+      "loss": 0.8789,
+      "step": 16478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0719627653131948e-07,
+      "loss": 0.8789,
+      "step": 16479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0692243188824447e-07,
+      "loss": 0.8203,
+      "step": 16480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0664893559570455e-07,
+      "loss": 0.8359,
+      "step": 16481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.063757876633309e-07,
+      "loss": 0.9297,
+      "step": 16482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0610298810074138e-07,
+      "loss": 0.875,
+      "step": 16483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0583053691754164e-07,
+      "loss": 0.8125,
+      "step": 16484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0555843412332622e-07,
+      "loss": 0.7734,
+      "step": 16485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0528667972767414e-07,
+      "loss": 0.832,
+      "step": 16486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0501527374015552e-07,
+      "loss": 0.9141,
+      "step": 16487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0474421617032826e-07,
+      "loss": 0.8594,
+      "step": 16488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0447350702773362e-07,
+      "loss": 0.8477,
+      "step": 16489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.042031463219062e-07,
+      "loss": 0.8203,
+      "step": 16490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0393313406236504e-07,
+      "loss": 0.8008,
+      "step": 16491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0366347025861812e-07,
+      "loss": 0.9297,
+      "step": 16492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0339415492016002e-07,
+      "loss": 0.9531,
+      "step": 16493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.031251880564743e-07,
+      "loss": 0.9609,
+      "step": 16494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0285656967703118e-07,
+      "loss": 0.8125,
+      "step": 16495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.025882997912897e-07,
+      "loss": 0.7812,
+      "step": 16496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0232037840869569e-07,
+      "loss": 0.8125,
+      "step": 16497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0205280553868269e-07,
+      "loss": 0.9219,
+      "step": 16498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 0.8008,
+      "step": 16499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0151870537407627e-07,
+      "loss": 0.8789,
+      "step": 16500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0125217809828781e-07,
+      "loss": 0.8672,
+      "step": 16501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0098599937269471e-07,
+      "loss": 0.8594,
+      "step": 16502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0072016920666838e-07,
+      "loss": 0.8359,
+      "step": 16503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0045468760957022e-07,
+      "loss": 0.8086,
+      "step": 16504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0018955459074719e-07,
+      "loss": 0.918,
+      "step": 16505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.992477015953405e-08,
+      "loss": 0.8203,
+      "step": 16506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.966033432525557e-08,
+      "loss": 1.0078,
+      "step": 16507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.939624709722429e-08,
+      "loss": 0.875,
+      "step": 16508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.91325084847361e-08,
+      "loss": 0.9531,
+      "step": 16509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.886911849708025e-08,
+      "loss": 0.8008,
+      "step": 16510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.860607714352933e-08,
+      "loss": 0.7969,
+      "step": 16511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.8343384433347e-08,
+      "loss": 0.9023,
+      "step": 16512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808104037578147e-08,
+      "loss": 0.8945,
+      "step": 16513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.781904498007089e-08,
+      "loss": 0.7734,
+      "step": 16514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.755739825544009e-08,
+      "loss": 0.7812,
+      "step": 16515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.729610021110392e-08,
+      "loss": 0.8086,
+      "step": 16516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.70351508562617e-08,
+      "loss": 0.8398,
+      "step": 16517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.677455020010162e-08,
+      "loss": 0.8359,
+      "step": 16518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.65142982517997e-08,
+      "loss": 0.8672,
+      "step": 16519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.62543950205208e-08,
+      "loss": 0.8398,
+      "step": 16520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.599484051541653e-08,
+      "loss": 0.6992,
+      "step": 16521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.573563474562508e-08,
+      "loss": 0.9062,
+      "step": 16522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.547677772027363e-08,
+      "loss": 0.8359,
+      "step": 16523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.521826944847823e-08,
+      "loss": 0.918,
+      "step": 16524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.496010993933935e-08,
+      "loss": 0.8555,
+      "step": 16525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.470229920194974e-08,
+      "loss": 0.7695,
+      "step": 16526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.444483724538433e-08,
+      "loss": 0.832,
+      "step": 16527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.418772407871258e-08,
+      "loss": 0.8008,
+      "step": 16528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.393095971098387e-08,
+      "loss": 0.7344,
+      "step": 16529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.3674544151241e-08,
+      "loss": 0.793,
+      "step": 16530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.34184774085134e-08,
+      "loss": 0.8672,
+      "step": 16531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.316275949181718e-08,
+      "loss": 0.9258,
+      "step": 16532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.290739041015518e-08,
+      "loss": 0.8125,
+      "step": 16533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.265237017252127e-08,
+      "loss": 0.8164,
+      "step": 16534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.239769878789495e-08,
+      "loss": 0.8359,
+      "step": 16535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.214337626524128e-08,
+      "loss": 0.9141,
+      "step": 16536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.188940261351864e-08,
+      "loss": 0.8516,
+      "step": 16537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.163577784166656e-08,
+      "loss": 0.8828,
+      "step": 16538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.138250195861897e-08,
+      "loss": 0.7148,
+      "step": 16539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.1129574973291e-08,
+      "loss": 0.875,
+      "step": 16540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.087699689459107e-08,
+      "loss": 0.7344,
+      "step": 16541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.062476773141093e-08,
+      "loss": 0.8672,
+      "step": 16542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.037288749263351e-08,
+      "loss": 0.8984,
+      "step": 16543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.012135618712615e-08,
+      "loss": 0.8398,
+      "step": 16544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.98701738237473e-08,
+      "loss": 0.7695,
+      "step": 16545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.96193404113399e-08,
+      "loss": 0.9844,
+      "step": 16546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.936885595873801e-08,
+      "loss": 0.8203,
+      "step": 16547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.911872047476122e-08,
+      "loss": 0.8125,
+      "step": 16548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.886893396821583e-08,
+      "loss": 0.8555,
+      "step": 16549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.861949644789814e-08,
+      "loss": 0.8125,
+      "step": 16550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.83704079225911e-08,
+      "loss": 0.8438,
+      "step": 16551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.812166840106551e-08,
+      "loss": 0.8438,
+      "step": 16552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.787327789207989e-08,
+      "loss": 0.7617,
+      "step": 16553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.762523640437947e-08,
+      "loss": 0.7812,
+      "step": 16554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.737754394670061e-08,
+      "loss": 0.8867,
+      "step": 16555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.7130200527763e-08,
+      "loss": 0.8867,
+      "step": 16556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.688320615627633e-08,
+      "loss": 0.8789,
+      "step": 16557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.663656084093808e-08,
+      "loss": 0.7852,
+      "step": 16558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.639026459043243e-08,
+      "loss": 0.8711,
+      "step": 16559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.614431741343243e-08,
+      "loss": 0.832,
+      "step": 16560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.589871931859783e-08,
+      "loss": 0.8516,
+      "step": 16561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.565347031457616e-08,
+      "loss": 0.8711,
+      "step": 16562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.540857041000494e-08,
+      "loss": 0.9336,
+      "step": 16563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.516401961350506e-08,
+      "loss": 0.7969,
+      "step": 16564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.491981793368742e-08,
+      "loss": 0.8516,
+      "step": 16565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.46759653791529e-08,
+      "loss": 0.8984,
+      "step": 16566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.443246195848798e-08,
+      "loss": 0.9023,
+      "step": 16567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.418930768026357e-08,
+      "loss": 0.8789,
+      "step": 16568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.394650255304504e-08,
+      "loss": 0.9609,
+      "step": 16569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.370404658538001e-08,
+      "loss": 0.8086,
+      "step": 16570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.346193978580608e-08,
+      "loss": 0.7891,
+      "step": 16571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.322018216284866e-08,
+      "loss": 0.9609,
+      "step": 16572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.297877372501984e-08,
+      "loss": 0.8438,
+      "step": 16573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.273771448082057e-08,
+      "loss": 0.7578,
+      "step": 16574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.249700443873853e-08,
+      "loss": 0.9414,
+      "step": 16575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.225664360724916e-08,
+      "loss": 0.793,
+      "step": 16576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.20166319948168e-08,
+      "loss": 0.9961,
+      "step": 16577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.177696960989134e-08,
+      "loss": 0.9297,
+      "step": 16578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.153765646091383e-08,
+      "loss": 0.7852,
+      "step": 16579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.129869255630862e-08,
+      "loss": 0.8047,
+      "step": 16580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.106007790449122e-08,
+      "loss": 0.9883,
+      "step": 16581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.082181251386268e-08,
+      "loss": 0.9688,
+      "step": 16582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.058389639281405e-08,
+      "loss": 0.8242,
+      "step": 16583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.034632954972199e-08,
+      "loss": 0.8516,
+      "step": 16584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.01091119929509e-08,
+      "loss": 0.8086,
+      "step": 16585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.987224373085523e-08,
+      "loss": 0.8086,
+      "step": 16586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.963572477177272e-08,
+      "loss": 0.918,
+      "step": 16587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.939955512403453e-08,
+      "loss": 0.8203,
+      "step": 16588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.916373479595507e-08,
+      "loss": 0.9414,
+      "step": 16589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.892826379583885e-08,
+      "loss": 0.9609,
+      "step": 16590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.869314213197587e-08,
+      "loss": 0.75,
+      "step": 16591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.84583698126462e-08,
+      "loss": 0.8711,
+      "step": 16592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.822394684611657e-08,
+      "loss": 0.9336,
+      "step": 16593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.798987324064034e-08,
+      "loss": 0.8633,
+      "step": 16594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.775614900445982e-08,
+      "loss": 0.8672,
+      "step": 16595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.752277414580623e-08,
+      "loss": 0.9727,
+      "step": 16596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.728974867289519e-08,
+      "loss": 0.8047,
+      "step": 16597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.705707259393346e-08,
+      "loss": 0.8477,
+      "step": 16598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.682474591711231e-08,
+      "loss": 0.8125,
+      "step": 16599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.659276865061405e-08,
+      "loss": 0.8516,
+      "step": 16600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.636114080260548e-08,
+      "loss": 0.8867,
+      "step": 16601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.612986238124342e-08,
+      "loss": 0.7891,
+      "step": 16602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.589893339467135e-08,
+      "loss": 0.9219,
+      "step": 16603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.566835385102056e-08,
+      "loss": 0.8438,
+      "step": 16604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.543812375841009e-08,
+      "loss": 1.0391,
+      "step": 16605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.520824312494679e-08,
+      "loss": 0.8281,
+      "step": 16606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.49787119587253e-08,
+      "loss": 0.9492,
+      "step": 16607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.474953026782694e-08,
+      "loss": 0.7578,
+      "step": 16608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.452069806032303e-08,
+      "loss": 0.9609,
+      "step": 16609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.429221534426934e-08,
+      "loss": 0.8438,
+      "step": 16610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.406408212771277e-08,
+      "loss": 0.7891,
+      "step": 16611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.383629841868468e-08,
+      "loss": 0.8242,
+      "step": 16612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.360886422520752e-08,
+      "loss": 0.9023,
+      "step": 16613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.338177955528825e-08,
+      "loss": 0.8828,
+      "step": 16614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.315504441692268e-08,
+      "loss": 0.793,
+      "step": 16615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.292865881809552e-08,
+      "loss": 0.9453,
+      "step": 16616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.270262276677708e-08,
+      "loss": 0.7812,
+      "step": 16617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.247693627092767e-08,
+      "loss": 0.7266,
+      "step": 16618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.225159933849312e-08,
+      "loss": 0.8125,
+      "step": 16619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.202661197740824e-08,
+      "loss": 1.0547,
+      "step": 16620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.180197419559442e-08,
+      "loss": 0.7891,
+      "step": 16621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.157768600096315e-08,
+      "loss": 0.9258,
+      "step": 16622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.135374740141032e-08,
+      "loss": 0.8516,
+      "step": 16623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.113015840482184e-08,
+      "loss": 0.9062,
+      "step": 16624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.090691901907143e-08,
+      "loss": 0.9258,
+      "step": 16625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.068402925201833e-08,
+      "loss": 0.8984,
+      "step": 16626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.046148911151073e-08,
+      "loss": 0.8281,
+      "step": 16627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.02392986053857e-08,
+      "loss": 0.8008,
+      "step": 16628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.001745774146695e-08,
+      "loss": 0.7891,
+      "step": 16629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.979596652756603e-08,
+      "loss": 0.8281,
+      "step": 16630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.957482497148004e-08,
+      "loss": 0.7148,
+      "step": 16631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.93540330809972e-08,
+      "loss": 0.9062,
+      "step": 16632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.913359086389127e-08,
+      "loss": 0.9961,
+      "step": 16633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.891349832792605e-08,
+      "loss": 0.9883,
+      "step": 16634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.869375548084978e-08,
+      "loss": 0.8594,
+      "step": 16635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.847436233039961e-08,
+      "loss": 0.8203,
+      "step": 16636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.825531888430159e-08,
+      "loss": 0.8203,
+      "step": 16637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.803662515026843e-08,
+      "loss": 0.7031,
+      "step": 16638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.781828113600063e-08,
+      "loss": 0.7734,
+      "step": 16639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.760028684918652e-08,
+      "loss": 0.8281,
+      "step": 16640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.738264229750214e-08,
+      "loss": 0.8945,
+      "step": 16641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.716534748861136e-08,
+      "loss": 0.8086,
+      "step": 16642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.694840243016364e-08,
+      "loss": 0.9102,
+      "step": 16643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.673180712979955e-08,
+      "loss": 0.8555,
+      "step": 16644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.651556159514627e-08,
+      "loss": 0.793,
+      "step": 16645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.629966583381665e-08,
+      "loss": 0.9141,
+      "step": 16646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.608411985341345e-08,
+      "loss": 0.793,
+      "step": 16647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.58689236615262e-08,
+      "loss": 0.8867,
+      "step": 16648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.565407726573326e-08,
+      "loss": 0.9258,
+      "step": 16649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.543958067359745e-08,
+      "loss": 0.8281,
+      "step": 16650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.522543389267389e-08,
+      "loss": 0.8164,
+      "step": 16651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.501163693050094e-08,
+      "loss": 0.8633,
+      "step": 16652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.479818979460928e-08,
+      "loss": 0.7656,
+      "step": 16653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.458509249251287e-08,
+      "loss": 0.9375,
+      "step": 16654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.437234503171463e-08,
+      "loss": 0.8047,
+      "step": 16655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.415994741970744e-08,
+      "loss": 0.875,
+      "step": 16656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.394789966396975e-08,
+      "loss": 0.7734,
+      "step": 16657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.373620177196783e-08,
+      "loss": 0.8906,
+      "step": 16658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.352485375115569e-08,
+      "loss": 0.7617,
+      "step": 16659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.331385560897407e-08,
+      "loss": 0.9141,
+      "step": 16660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.310320735285592e-08,
+      "loss": 0.8359,
+      "step": 16661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.289290899021417e-08,
+      "loss": 0.9414,
+      "step": 16662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.268296052845735e-08,
+      "loss": 0.8906,
+      "step": 16663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.247336197497622e-08,
+      "loss": 0.8086,
+      "step": 16664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.226411333715155e-08,
+      "loss": 0.7891,
+      "step": 16665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.205521462235186e-08,
+      "loss": 0.8398,
+      "step": 16666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.184666583793131e-08,
+      "loss": 0.8594,
+      "step": 16667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.163846699123399e-08,
+      "loss": 0.9062,
+      "step": 16668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.143061808959183e-08,
+      "loss": 0.9219,
+      "step": 16669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.122311914032231e-08,
+      "loss": 0.7383,
+      "step": 16670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.101597015073291e-08,
+      "loss": 0.7344,
+      "step": 16671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.080917112811557e-08,
+      "loss": 0.8359,
+      "step": 16672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.060272207975448e-08,
+      "loss": 0.7852,
+      "step": 16673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.039662301291826e-08,
+      "loss": 0.7969,
+      "step": 16674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.019087393486223e-08,
+      "loss": 0.918,
+      "step": 16675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.998547485283391e-08,
+      "loss": 0.8008,
+      "step": 16676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.978042577406417e-08,
+      "loss": 0.7969,
+      "step": 16677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.957572670577283e-08,
+      "loss": 0.9023,
+      "step": 16678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9371377655167425e-08,
+      "loss": 0.7695,
+      "step": 16679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.916737862944444e-08,
+      "loss": 0.8672,
+      "step": 16680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.896372963578811e-08,
+      "loss": 0.9062,
+      "step": 16681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.876043068136606e-08,
+      "loss": 0.832,
+      "step": 16682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.85574817733392e-08,
+      "loss": 0.7695,
+      "step": 16683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.835488291885294e-08,
+      "loss": 0.8125,
+      "step": 16684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.815263412504157e-08,
+      "loss": 0.8867,
+      "step": 16685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.795073539902607e-08,
+      "loss": 0.8438,
+      "step": 16686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7749186747915186e-08,
+      "loss": 0.7891,
+      "step": 16687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.754798817880547e-08,
+      "loss": 0.8086,
+      "step": 16688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.734713969878347e-08,
+      "loss": 0.9219,
+      "step": 16689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.71466413149202e-08,
+      "loss": 0.8594,
+      "step": 16690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.694649303427446e-08,
+      "loss": 0.7969,
+      "step": 16691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.674669486389506e-08,
+      "loss": 0.918,
+      "step": 16692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.654724681081747e-08,
+      "loss": 0.832,
+      "step": 16693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6348148882062746e-08,
+      "loss": 0.6406,
+      "step": 16694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6149401084644175e-08,
+      "loss": 0.8203,
+      "step": 16695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.595100342555726e-08,
+      "loss": 0.9141,
+      "step": 16696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.575295591178975e-08,
+      "loss": 0.8945,
+      "step": 16697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.555525855031385e-08,
+      "loss": 0.8438,
+      "step": 16698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.535791134809176e-08,
+      "loss": 0.7617,
+      "step": 16699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.516091431207238e-08,
+      "loss": 0.7188,
+      "step": 16700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.496426744919236e-08,
+      "loss": 0.8438,
+      "step": 16701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.476797076637508e-08,
+      "loss": 0.7734,
+      "step": 16702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.457202427053276e-08,
+      "loss": 0.8477,
+      "step": 16703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4376427968565457e-08,
+      "loss": 0.9062,
+      "step": 16704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.418118186735988e-08,
+      "loss": 0.8398,
+      "step": 16705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.398628597379163e-08,
+      "loss": 1.0,
+      "step": 16706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.37917402947219e-08,
+      "loss": 0.9062,
+      "step": 16707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.359754483700297e-08,
+      "loss": 0.8359,
+      "step": 16708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.34036996074716e-08,
+      "loss": 0.7617,
+      "step": 16709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.321020461295345e-08,
+      "loss": 0.875,
+      "step": 16710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.301705986026084e-08,
+      "loss": 0.8438,
+      "step": 16711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2824265356197225e-08,
+      "loss": 0.832,
+      "step": 16712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.263182110754827e-08,
+      "loss": 0.8203,
+      "step": 16713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.243972712109191e-08,
+      "loss": 0.8945,
+      "step": 16714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2247983403592714e-08,
+      "loss": 0.7461,
+      "step": 16715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.205658996179974e-08,
+      "loss": 0.8359,
+      "step": 16716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1865546802454255e-08,
+      "loss": 0.832,
+      "step": 16717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.167485393228311e-08,
+      "loss": 0.9297,
+      "step": 16718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1484511358000924e-08,
+      "loss": 0.8164,
+      "step": 16719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1294519086309004e-08,
+      "loss": 0.8984,
+      "step": 16720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.110487712389867e-08,
+      "loss": 0.7617,
+      "step": 16721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0915585477445686e-08,
+      "loss": 0.793,
+      "step": 16722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.072664415361694e-08,
+      "loss": 0.7539,
+      "step": 16723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0538053159063794e-08,
+      "loss": 0.8633,
+      "step": 16724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0349812500427585e-08,
+      "loss": 0.9766,
+      "step": 16725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.016192218433746e-08,
+      "loss": 0.8672,
+      "step": 16726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.997438221740925e-08,
+      "loss": 0.9414,
+      "step": 16727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.978719260624432e-08,
+      "loss": 0.9453,
+      "step": 16728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9600353357436294e-08,
+      "loss": 0.7812,
+      "step": 16729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.941386447756324e-08,
+      "loss": 0.8555,
+      "step": 16730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.922772597319214e-08,
+      "loss": 0.8047,
+      "step": 16731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.904193785087552e-08,
+      "loss": 0.8047,
+      "step": 16732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8856500117157036e-08,
+      "loss": 0.8828,
+      "step": 16733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.867141277856591e-08,
+      "loss": 0.75,
+      "step": 16734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8486675841619144e-08,
+      "loss": 0.8867,
+      "step": 16735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.830228931282155e-08,
+      "loss": 0.875,
+      "step": 16736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.81182531986657e-08,
+      "loss": 0.8203,
+      "step": 16737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7934567505631967e-08,
+      "loss": 0.8438,
+      "step": 16738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7751232240187405e-08,
+      "loss": 0.7969,
+      "step": 16739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.756824740878907e-08,
+      "loss": 0.8398,
+      "step": 16740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7385613017878474e-08,
+      "loss": 0.8242,
+      "step": 16741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.720332907388825e-08,
+      "loss": 0.7617,
+      "step": 16742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7021395583235486e-08,
+      "loss": 0.8203,
+      "step": 16743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.683981255232617e-08,
+      "loss": 0.7422,
+      "step": 16744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6658579987555186e-08,
+      "loss": 0.8867,
+      "step": 16745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6477697895302986e-08,
+      "loss": 0.8359,
+      "step": 16746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.629716628194003e-08,
+      "loss": 0.9961,
+      "step": 16747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6116985153821237e-08,
+      "loss": 0.8594,
+      "step": 16748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.593715451729375e-08,
+      "loss": 0.7656,
+      "step": 16749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5757674378686944e-08,
+      "loss": 0.707,
+      "step": 16750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.557854474432244e-08,
+      "loss": 0.8164,
+      "step": 16751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.53997656205063e-08,
+      "loss": 0.7617,
+      "step": 16752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.52213370135346e-08,
+      "loss": 0.875,
+      "step": 16753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.504325892968897e-08,
+      "loss": 0.8125,
+      "step": 16754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.486553137524108e-08,
+      "loss": 0.7305,
+      "step": 16755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4688154356449244e-08,
+      "loss": 0.7344,
+      "step": 16756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.451112787955847e-08,
+      "loss": 0.8633,
+      "step": 16757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.433445195080155e-08,
+      "loss": 0.8945,
+      "step": 16758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.415812657639906e-08,
+      "loss": 0.8594,
+      "step": 16759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.398215176256271e-08,
+      "loss": 0.8828,
+      "step": 16760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.380652751548531e-08,
+      "loss": 0.8672,
+      "step": 16761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.363125384135303e-08,
+      "loss": 0.7617,
+      "step": 16762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.345633074633648e-08,
+      "loss": 0.7344,
+      "step": 16763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3281758236596306e-08,
+      "loss": 0.8555,
+      "step": 16764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.310753631827869e-08,
+      "loss": 0.8438,
+      "step": 16765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.29336649975165e-08,
+      "loss": 0.8398,
+      "step": 16766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.276014428043485e-08,
+      "loss": 0.875,
+      "step": 16767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.258697417314217e-08,
+      "loss": 0.9375,
+      "step": 16768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.241415468173693e-08,
+      "loss": 0.8359,
+      "step": 16769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2241685812303145e-08,
+      "loss": 0.8828,
+      "step": 16770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.206956757091485e-08,
+      "loss": 0.9141,
+      "step": 16771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.189779996363163e-08,
+      "loss": 0.793,
+      "step": 16772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1726382996502004e-08,
+      "loss": 0.8281,
+      "step": 16773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.155531667556112e-08,
+      "loss": 0.8828,
+      "step": 16774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1384601006835276e-08,
+      "loss": 0.8633,
+      "step": 16775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1214235996331894e-08,
+      "loss": 0.8086,
+      "step": 16776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.104422165005284e-08,
+      "loss": 0.8945,
+      "step": 16777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.087455797398332e-08,
+      "loss": 0.8125,
+      "step": 16778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.070524497409745e-08,
+      "loss": 0.8594,
+      "step": 16779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.053628265635601e-08,
+      "loss": 0.6719,
+      "step": 16780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.03676710267098e-08,
+      "loss": 0.7031,
+      "step": 16781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.019941009109629e-08,
+      "loss": 0.8008,
+      "step": 16782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0031499855438525e-08,
+      "loss": 0.8164,
+      "step": 16783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9863940325650664e-08,
+      "loss": 0.9453,
+      "step": 16784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.969673150763132e-08,
+      "loss": 0.8633,
+      "step": 16785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9529873407269104e-08,
+      "loss": 0.875,
+      "step": 16786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.936336603043822e-08,
+      "loss": 0.8164,
+      "step": 16787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.919720938300287e-08,
+      "loss": 0.8477,
+      "step": 16788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.90314034708128e-08,
+      "loss": 0.9023,
+      "step": 16789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.88659482997078e-08,
+      "loss": 0.8281,
+      "step": 16790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.87008438755121e-08,
+      "loss": 0.8867,
+      "step": 16791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8536090204039924e-08,
+      "loss": 0.8359,
+      "step": 16792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8371687291092194e-08,
+      "loss": 0.8281,
+      "step": 16793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.820763514245762e-08,
+      "loss": 0.8164,
+      "step": 16794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.804393376391491e-08,
+      "loss": 0.9883,
+      "step": 16795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7880583161225e-08,
+      "loss": 0.9297,
+      "step": 16796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.77175833401422e-08,
+      "loss": 0.8555,
+      "step": 16797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.755493430640411e-08,
+      "loss": 0.7812,
+      "step": 16798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.739263606573951e-08,
+      "loss": 0.8789,
+      "step": 16799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.723068862386159e-08,
+      "loss": 0.8398,
+      "step": 16800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7069091986473574e-08,
+      "loss": 0.8438,
+      "step": 16801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.690784615926646e-08,
+      "loss": 0.8164,
+      "step": 16802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.674695114791682e-08,
+      "loss": 0.8164,
+      "step": 16803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.658640695809013e-08,
+      "loss": 0.9609,
+      "step": 16804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.642621359543963e-08,
+      "loss": 0.9609,
+      "step": 16805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.626637106560527e-08,
+      "loss": 0.8203,
+      "step": 16806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.610687937421698e-08,
+      "loss": 0.9062,
+      "step": 16807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5947738526888045e-08,
+      "loss": 0.793,
+      "step": 16808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.57889485292251e-08,
+      "loss": 0.7617,
+      "step": 16809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5630509386817e-08,
+      "loss": 0.8281,
+      "step": 16810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.547242110524485e-08,
+      "loss": 0.9141,
+      "step": 16811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.531468369007307e-08,
+      "loss": 0.9805,
+      "step": 16812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.515729714685723e-08,
+      "loss": 0.7422,
+      "step": 16813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.500026148113844e-08,
+      "loss": 0.7695,
+      "step": 16814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4843576698446733e-08,
+      "loss": 1.0469,
+      "step": 16815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4687242804297696e-08,
+      "loss": 0.918,
+      "step": 16816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.453125980419803e-08,
+      "loss": 0.7812,
+      "step": 16817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4375627703640003e-08,
+      "loss": 0.7734,
+      "step": 16818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4220346508101464e-08,
+      "loss": 0.7891,
+      "step": 16819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.406541622305248e-08,
+      "loss": 0.9102,
+      "step": 16820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391083685394758e-08,
+      "loss": 0.8594,
+      "step": 16821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.375660840623018e-08,
+      "loss": 0.875,
+      "step": 16822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3602730885329285e-08,
+      "loss": 0.9844,
+      "step": 16823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.344920429666498e-08,
+      "loss": 0.8438,
+      "step": 16824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.329602864564296e-08,
+      "loss": 0.8125,
+      "step": 16825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3143203937655576e-08,
+      "loss": 0.8203,
+      "step": 16826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2990730178085184e-08,
+      "loss": 0.8438,
+      "step": 16827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.283860737229971e-08,
+      "loss": 0.8359,
+      "step": 16828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.268683552565599e-08,
+      "loss": 0.707,
+      "step": 16829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2535414643499744e-08,
+      "loss": 0.7539,
+      "step": 16830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2384344731160036e-08,
+      "loss": 0.8203,
+      "step": 16831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2233625793957054e-08,
+      "loss": 0.8242,
+      "step": 16832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2083257837199897e-08,
+      "loss": 0.7578,
+      "step": 16833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.193324086618099e-08,
+      "loss": 0.8281,
+      "step": 16834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1783574886182775e-08,
+      "loss": 0.9219,
+      "step": 16835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1634259902475486e-08,
+      "loss": 0.8633,
+      "step": 16836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1485295920318236e-08,
+      "loss": 0.7891,
+      "step": 16837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1336682944954624e-08,
+      "loss": 0.9258,
+      "step": 16838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1188420981617115e-08,
+      "loss": 0.9297,
+      "step": 16839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.104051003552822e-08,
+      "loss": 0.9297,
+      "step": 16840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.089295011189486e-08,
+      "loss": 0.8359,
+      "step": 16841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0745741215912897e-08,
+      "loss": 0.7266,
+      "step": 16842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.059888335276484e-08,
+      "loss": 0.9023,
+      "step": 16843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.045237652762434e-08,
+      "loss": 0.8125,
+      "step": 16844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0306220745647263e-08,
+      "loss": 1.0312,
+      "step": 16845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.016041601198283e-08,
+      "loss": 0.8438,
+      "step": 16846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0014962331762485e-08,
+      "loss": 0.8789,
+      "step": 16847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.986985971010881e-08,
+      "loss": 0.8438,
+      "step": 16848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.972510815213214e-08,
+      "loss": 0.8398,
+      "step": 16849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.958070766292731e-08,
+      "loss": 0.8086,
+      "step": 16850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9436658247581352e-08,
+      "loss": 0.8438,
+      "step": 16851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9292959911163542e-08,
+      "loss": 0.8086,
+      "step": 16852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9149612658736502e-08,
+      "loss": 0.9062,
+      "step": 16853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9006616495346197e-08,
+      "loss": 0.8438,
+      "step": 16854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.886397142602859e-08,
+      "loss": 0.8516,
+      "step": 16855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.872167745580523e-08,
+      "loss": 0.9219,
+      "step": 16856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8579734589687658e-08,
+      "loss": 0.8555,
+      "step": 16857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.84381428326741e-08,
+      "loss": 0.832,
+      "step": 16858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8296902189748344e-08,
+      "loss": 0.8555,
+      "step": 16859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.81560126658853e-08,
+      "loss": 0.957,
+      "step": 16860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.801547426604545e-08,
+      "loss": 0.8242,
+      "step": 16861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7875286995178164e-08,
+      "loss": 0.8047,
+      "step": 16862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7735450858218382e-08,
+      "loss": 0.8711,
+      "step": 16863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7595965860089947e-08,
+      "loss": 0.875,
+      "step": 16864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7456832005705592e-08,
+      "loss": 0.8359,
+      "step": 16865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7318049299964733e-08,
+      "loss": 0.6992,
+      "step": 16866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7179617747751242e-08,
+      "loss": 0.8398,
+      "step": 16867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7041537353942325e-08,
+      "loss": 0.7695,
+      "step": 16868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6903808123399656e-08,
+      "loss": 0.8516,
+      "step": 16869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.676643006097157e-08,
+      "loss": 0.8828,
+      "step": 16870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6629403171496427e-08,
+      "loss": 0.8672,
+      "step": 16871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.649272745979814e-08,
+      "loss": 0.7734,
+      "step": 16872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.635640293069064e-08,
+      "loss": 0.7383,
+      "step": 16873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6220429588973418e-08,
+      "loss": 0.832,
+      "step": 16874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6084807439433758e-08,
+      "loss": 0.8281,
+      "step": 16875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5949536486848947e-08,
+      "loss": 0.9102,
+      "step": 16876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5814616735979625e-08,
+      "loss": 0.9414,
+      "step": 16877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5680048191578654e-08,
+      "loss": 0.7891,
+      "step": 16878
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.554583085838336e-08,
+      "loss": 0.8125,
+      "step": 16879
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5411964741118844e-08,
+      "loss": 0.7891,
+      "step": 16880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5278449844501342e-08,
+      "loss": 0.8008,
+      "step": 16881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5145286173229312e-08,
+      "loss": 1.0156,
+      "step": 16882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5012473731993447e-08,
+      "loss": 0.8555,
+      "step": 16883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.48800125254689e-08,
+      "loss": 0.8125,
+      "step": 16884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4747902558321935e-08,
+      "loss": 0.8477,
+      "step": 16885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4616143835202166e-08,
+      "loss": 0.9219,
+      "step": 16886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.448473636075033e-08,
+      "loss": 0.7812,
+      "step": 16887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4353680139592715e-08,
+      "loss": 0.918,
+      "step": 16888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4222975176345642e-08,
+      "loss": 0.8359,
+      "step": 16889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4092621475608756e-08,
+      "loss": 0.8398,
+      "step": 16890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3962619041973945e-08,
+      "loss": 0.8281,
+      "step": 16891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3832967880017544e-08,
+      "loss": 0.793,
+      "step": 16892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.370366799430701e-08,
+      "loss": 0.7891,
+      "step": 16893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.357471938939204e-08,
+      "loss": 0.8164,
+      "step": 16894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3446122069815668e-08,
+      "loss": 0.7773,
+      "step": 16895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.331787604010538e-08,
+      "loss": 0.7969,
+      "step": 16896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3189981304776455e-08,
+      "loss": 0.7891,
+      "step": 16897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3062437868331953e-08,
+      "loss": 0.9414,
+      "step": 16898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2935245735263845e-08,
+      "loss": 0.8867,
+      "step": 16899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2808404910049652e-08,
+      "loss": 0.9062,
+      "step": 16900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2681915397156918e-08,
+      "loss": 0.875,
+      "step": 16901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2555777201038742e-08,
+      "loss": 0.7305,
+      "step": 16902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.242999032613713e-08,
+      "loss": 0.8242,
+      "step": 16903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.230455477687965e-08,
+      "loss": 0.9961,
+      "step": 16904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.21794705576861e-08,
+      "loss": 0.8477,
+      "step": 16905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2054737672957403e-08,
+      "loss": 0.8984,
+      "step": 16906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1930356127087827e-08,
+      "loss": 0.8867,
+      "step": 16907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1806325924457196e-08,
+      "loss": 0.8242,
+      "step": 16908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1682647069432018e-08,
+      "loss": 0.75,
+      "step": 16909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.155931956636659e-08,
+      "loss": 0.8438,
+      "step": 16910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.14363434196041e-08,
+      "loss": 0.9727,
+      "step": 16911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1313718633475534e-08,
+      "loss": 0.9922,
+      "step": 16912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1191445212297433e-08,
+      "loss": 0.7461,
+      "step": 16913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.106952316037636e-08,
+      "loss": 0.7812,
+      "step": 16914
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.094795248200443e-08,
+      "loss": 0.8086,
+      "step": 16915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0826733181462667e-08,
+      "loss": 0.9297,
+      "step": 16916
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0705865263019877e-08,
+      "loss": 0.918,
+      "step": 16917
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.058534873093154e-08,
+      "loss": 0.7852,
+      "step": 16918
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.046518358944094e-08,
+      "loss": 0.8633,
+      "step": 16919
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0345369842779128e-08,
+      "loss": 0.7734,
+      "step": 16920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0225907495166063e-08,
+      "loss": 0.8008,
+      "step": 16921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0106796550807272e-08,
+      "loss": 0.8125,
+      "step": 16922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9988037013897176e-08,
+      "loss": 0.8984,
+      "step": 16923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9869628888617986e-08,
+      "loss": 0.8281,
+      "step": 16924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.975157217913748e-08,
+      "loss": 0.7266,
+      "step": 16925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.963386688961344e-08,
+      "loss": 0.8477,
+      "step": 16926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9516513024191442e-08,
+      "loss": 0.8789,
+      "step": 16927
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9399510587002625e-08,
+      "loss": 0.8516,
+      "step": 16928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9282859582167025e-08,
+      "loss": 0.8125,
+      "step": 16929
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9166560013791358e-08,
+      "loss": 0.7656,
+      "step": 16930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9050611885972347e-08,
+      "loss": 0.8125,
+      "step": 16931
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8935015202791174e-08,
+      "loss": 1.0234,
+      "step": 16932
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8819769968320133e-08,
+      "loss": 0.9531,
+      "step": 16933
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8704876186614872e-08,
+      "loss": 0.7891,
+      "step": 16934
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8590333861722153e-08,
+      "loss": 0.8633,
+      "step": 16935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8476142997675416e-08,
+      "loss": 0.8281,
+      "step": 16936
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.836230359849478e-08,
+      "loss": 0.8359,
+      "step": 16937
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.824881566818815e-08,
+      "loss": 0.8086,
+      "step": 16938
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8135679210753432e-08,
+      "loss": 0.7773,
+      "step": 16939
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8022894230174114e-08,
+      "loss": 0.8945,
+      "step": 16940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7910460730419242e-08,
+      "loss": 0.7812,
+      "step": 16941
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7798378715450094e-08,
+      "loss": 0.8242,
+      "step": 16942
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7686648189212395e-08,
+      "loss": 0.9766,
+      "step": 16943
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7575269155641894e-08,
+      "loss": 0.8906,
+      "step": 16944
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7464241618657675e-08,
+      "loss": 0.8672,
+      "step": 16945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.735356558216994e-08,
+      "loss": 1.0312,
+      "step": 16946
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7243241050076685e-08,
+      "loss": 0.7734,
+      "step": 16947
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7133268026262585e-08,
+      "loss": 0.9922,
+      "step": 16948
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.702364651460009e-08,
+      "loss": 0.9648,
+      "step": 16949
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.691437651894834e-08,
+      "loss": 0.7539,
+      "step": 16950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6805458043154255e-08,
+      "loss": 0.832,
+      "step": 16951
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.669689109105588e-08,
+      "loss": 0.9062,
+      "step": 16952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6588675666472376e-08,
+      "loss": 0.8945,
+      "step": 16953
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6480811773216256e-08,
+      "loss": 0.7773,
+      "step": 16954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6373299415085587e-08,
+      "loss": 0.918,
+      "step": 16955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6266138595866234e-08,
+      "loss": 0.8203,
+      "step": 16956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.615932931933073e-08,
+      "loss": 0.7891,
+      "step": 16957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6052871589240515e-08,
+      "loss": 0.7969,
+      "step": 16958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.594676540934259e-08,
+      "loss": 0.8984,
+      "step": 16959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5841010783376187e-08,
+      "loss": 0.8164,
+      "step": 16960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5735607715061664e-08,
+      "loss": 0.8594,
+      "step": 16961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5630556208112713e-08,
+      "loss": 0.7539,
+      "step": 16962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.552585626622749e-08,
+      "loss": 0.8008,
+      "step": 16963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5421507893093046e-08,
+      "loss": 0.8711,
+      "step": 16964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5317511092384217e-08,
+      "loss": 0.7891,
+      "step": 16965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.521386586776141e-08,
+      "loss": 0.7656,
+      "step": 16966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5110572222875043e-08,
+      "loss": 0.9102,
+      "step": 16967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5007630161361086e-08,
+      "loss": 0.7969,
+      "step": 16968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4905039686846645e-08,
+      "loss": 0.8906,
+      "step": 16969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4802800802942164e-08,
+      "loss": 0.7773,
+      "step": 16970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4700913513248094e-08,
+      "loss": 0.9922,
+      "step": 16971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4599377821351568e-08,
+      "loss": 0.7461,
+      "step": 16972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4498193730829724e-08,
+      "loss": 0.6484,
+      "step": 16973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4397361245243046e-08,
+      "loss": 0.9297,
+      "step": 16974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.429688036814314e-08,
+      "loss": 0.9805,
+      "step": 16975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4196751103068284e-08,
+      "loss": 0.9023,
+      "step": 16976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.409697345354344e-08,
+      "loss": 0.8633,
+      "step": 16977
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3997547423082459e-08,
+      "loss": 0.8164,
+      "step": 16978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3898473015185876e-08,
+      "loss": 0.8867,
+      "step": 16979
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3799750233343124e-08,
+      "loss": 0.9453,
+      "step": 16980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3701379081030308e-08,
+      "loss": 0.7773,
+      "step": 16981
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3603359561709106e-08,
+      "loss": 0.9023,
+      "step": 16982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3505691678834532e-08,
+      "loss": 0.8203,
+      "step": 16983
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3408375435842724e-08,
+      "loss": 0.9414,
+      "step": 16984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3311410836160943e-08,
+      "loss": 0.8945,
+      "step": 16985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3214797883204233e-08,
+      "loss": 0.8711,
+      "step": 16986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3118536580373209e-08,
+      "loss": 0.8906,
+      "step": 16987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3022626931059602e-08,
+      "loss": 0.7344,
+      "step": 16988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2927068938638488e-08,
+      "loss": 0.8984,
+      "step": 16989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2831862606474954e-08,
+      "loss": 0.9219,
+      "step": 16990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2737007937921875e-08,
+      "loss": 1.0391,
+      "step": 16991
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.26425049363188e-08,
+      "loss": 0.6758,
+      "step": 16992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2548353604994179e-08,
+      "loss": 0.8398,
+      "step": 16993
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2454553947262026e-08,
+      "loss": 0.8867,
+      "step": 16994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2361105966426368e-08,
+      "loss": 0.8594,
+      "step": 16995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2268009665776793e-08,
+      "loss": 0.7891,
+      "step": 16996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2175265048591789e-08,
+      "loss": 0.9023,
+      "step": 16997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2082872118136524e-08,
+      "loss": 0.9844,
+      "step": 16998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.199083087766506e-08,
+      "loss": 0.7383,
+      "step": 16999
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.189914133041703e-08,
+      "loss": 0.8438,
+      "step": 17000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.180780347962318e-08,
+      "loss": 0.9219,
+      "step": 17001
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1716817328496499e-08,
+      "loss": 0.8516,
+      "step": 17002
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1626182880244419e-08,
+      "loss": 1.0234,
+      "step": 17003
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.15359001380555e-08,
+      "loss": 0.8906,
+      "step": 17004
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1445969105109422e-08,
+      "loss": 0.875,
+      "step": 17005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1356389784573652e-08,
+      "loss": 0.8906,
+      "step": 17006
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1267162179601221e-08,
+      "loss": 0.7773,
+      "step": 17007
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1178286293335172e-08,
+      "loss": 0.8633,
+      "step": 17008
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1089762128904113e-08,
+      "loss": 0.75,
+      "step": 17009
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1001589689425551e-08,
+      "loss": 0.8281,
+      "step": 17010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.091376897800367e-08,
+      "loss": 0.9258,
+      "step": 17011
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0826299997731548e-08,
+      "loss": 0.7969,
+      "step": 17012
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0739182751687838e-08,
+      "loss": 0.9453,
+      "step": 17013
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0652417242941193e-08,
+      "loss": 0.8789,
+      "step": 17014
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0566003474546949e-08,
+      "loss": 0.9023,
+      "step": 17015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0479941449547116e-08,
+      "loss": 0.8594,
+      "step": 17016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0394231170972602e-08,
+      "loss": 0.7812,
+      "step": 17017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0308872641839885e-08,
+      "loss": 0.8594,
+      "step": 17018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0223865865157667e-08,
+      "loss": 0.9219,
+      "step": 17019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0139210843915781e-08,
+      "loss": 0.9453,
+      "step": 17020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0054907581097394e-08,
+      "loss": 0.9883,
+      "step": 17021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.970956079670135e-09,
+      "loss": 0.7148,
+      "step": 17022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.887356342589416e-09,
+      "loss": 0.7812,
+      "step": 17023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.804108372799548e-09,
+      "loss": 0.9141,
+      "step": 17024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.72121217323263e-09,
+      "loss": 0.7578,
+      "step": 17025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.638667746806329e-09,
+      "loss": 0.7812,
+      "step": 17026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.556475096428319e-09,
+      "loss": 1.0312,
+      "step": 17027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.47463422499073e-09,
+      "loss": 0.8008,
+      "step": 17028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.393145135377924e-09,
+      "loss": 0.8203,
+      "step": 17029
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.312007830456492e-09,
+      "loss": 0.7891,
+      "step": 17030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.231222313086374e-09,
+      "loss": 0.7578,
+      "step": 17031
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.150788586109737e-09,
+      "loss": 0.7539,
+      "step": 17032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.070706652359872e-09,
+      "loss": 0.7773,
+      "step": 17033
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.990976514656745e-09,
+      "loss": 0.8633,
+      "step": 17034
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.911598175807002e-09,
+      "loss": 0.832,
+      "step": 17035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.83257163860618e-09,
+      "loss": 0.9336,
+      "step": 17036
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.7538969058365e-09,
+      "loss": 0.8398,
+      "step": 17037
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.675573980269081e-09,
+      "loss": 1.0234,
+      "step": 17038
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.597602864660604e-09,
+      "loss": 0.8516,
+      "step": 17039
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.51998356175776e-09,
+      "loss": 0.9023,
+      "step": 17040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.442716074292812e-09,
+      "loss": 0.918,
+      "step": 17041
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.365800404986913e-09,
+      "loss": 0.8594,
+      "step": 17042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.289236556547898e-09,
+      "loss": 0.7812,
+      "step": 17043
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.213024531672498e-09,
+      "loss": 0.9219,
+      "step": 17044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.137164333044123e-09,
+      "loss": 0.8281,
+      "step": 17045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.06165596333286e-09,
+      "loss": 0.8125,
+      "step": 17046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.986499425198802e-09,
+      "loss": 0.875,
+      "step": 17047
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.9116947212865e-09,
+      "loss": 0.793,
+      "step": 17048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.837241854232736e-09,
+      "loss": 0.8086,
+      "step": 17049
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.763140826656524e-09,
+      "loss": 0.8008,
+      "step": 17050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.689391641169108e-09,
+      "loss": 0.8477,
+      "step": 17051
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.615994300365081e-09,
+      "loss": 0.8828,
+      "step": 17052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.542948806830153e-09,
+      "loss": 0.9219,
+      "step": 17053
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.470255163137818e-09,
+      "loss": 0.8203,
+      "step": 17054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.397913371844923e-09,
+      "loss": 0.957,
+      "step": 17055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.325923435500537e-09,
+      "loss": 0.9688,
+      "step": 17056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.254285356638191e-09,
+      "loss": 0.8477,
+      "step": 17057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.182999137782531e-09,
+      "loss": 0.957,
+      "step": 17058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.112064781441552e-09,
+      "loss": 0.8516,
+      "step": 17059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.041482290113255e-09,
+      "loss": 0.832,
+      "step": 17060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.97125166628454e-09,
+      "loss": 1.0391,
+      "step": 17061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.901372912426762e-09,
+      "loss": 0.8906,
+      "step": 17062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.831846031000178e-09,
+      "loss": 0.9414,
+      "step": 17063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.762671024455047e-09,
+      "loss": 0.7578,
+      "step": 17064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.6938478952249805e-09,
+      "loss": 0.8008,
+      "step": 17065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.625376645734704e-09,
+      "loss": 0.7656,
+      "step": 17066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.557257278394514e-09,
+      "loss": 0.8945,
+      "step": 17067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.489489795602488e-09,
+      "loss": 0.8906,
+      "step": 17068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.422074199745609e-09,
+      "loss": 0.7539,
+      "step": 17069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.355010493198644e-09,
+      "loss": 0.9375,
+      "step": 17070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2882986783208145e-09,
+      "loss": 0.8164,
+      "step": 17071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.221938757462464e-09,
+      "loss": 0.8125,
+      "step": 17072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.155930732960613e-09,
+      "loss": 0.8672,
+      "step": 17073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0902746071378475e-09,
+      "loss": 0.8438,
+      "step": 17074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.024970382307871e-09,
+      "loss": 0.9609,
+      "step": 17075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.960018060768846e-09,
+      "loss": 0.8672,
+      "step": 17076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.895417644808943e-09,
+      "loss": 0.6992,
+      "step": 17077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8311691367018974e-09,
+      "loss": 0.7969,
+      "step": 17078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.767272538709234e-09,
+      "loss": 0.7617,
+      "step": 17079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.703727853083596e-09,
+      "loss": 0.7812,
+      "step": 17080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.640535082059862e-09,
+      "loss": 0.8711,
+      "step": 17081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.577694227865138e-09,
+      "loss": 0.7734,
+      "step": 17082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.51520529270988e-09,
+      "loss": 0.8789,
+      "step": 17083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.453068278796769e-09,
+      "loss": 0.7617,
+      "step": 17084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.391283188312946e-09,
+      "loss": 0.9023,
+      "step": 17085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.329850023432226e-09,
+      "loss": 0.8203,
+      "step": 17086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.268768786320655e-09,
+      "loss": 0.832,
+      "step": 17087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2080394791265145e-09,
+      "loss": 0.875,
+      "step": 17088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.147662103989204e-09,
+      "loss": 0.9336,
+      "step": 17089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0876366630359105e-09,
+      "loss": 0.7031,
+      "step": 17090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0279631583782794e-09,
+      "loss": 0.7969,
+      "step": 17091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.968641592117962e-09,
+      "loss": 1.0156,
+      "step": 17092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9096719663443985e-09,
+      "loss": 0.7578,
+      "step": 17093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.851054283133705e-09,
+      "loss": 0.7656,
+      "step": 17094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.792788544549787e-09,
+      "loss": 0.9453,
+      "step": 17095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7348747526443364e-09,
+      "loss": 0.8711,
+      "step": 17096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6773129094568325e-09,
+      "loss": 0.9023,
+      "step": 17097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.620103017013433e-09,
+      "loss": 0.8242,
+      "step": 17098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5632450773291926e-09,
+      "loss": 0.8867,
+      "step": 17099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.506739092405843e-09,
+      "loss": 0.8359,
+      "step": 17100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4505850642329044e-09,
+      "loss": 0.7852,
+      "step": 17101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.394782994788793e-09,
+      "loss": 0.9102,
+      "step": 17102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.339332886037495e-09,
+      "loss": 0.8008,
+      "step": 17103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.284234739930782e-09,
+      "loss": 0.7969,
+      "step": 17104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.229488558410433e-09,
+      "loss": 0.8789,
+      "step": 17105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.175094343402686e-09,
+      "loss": 0.8672,
+      "step": 17106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1210520968237855e-09,
+      "loss": 0.8477,
+      "step": 17107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.067361820575544e-09,
+      "loss": 0.9922,
+      "step": 17108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.014023516549781e-09,
+      "loss": 0.8125,
+      "step": 17109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.961037186623884e-09,
+      "loss": 0.8086,
+      "step": 17110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.908402832663028e-09,
+      "loss": 0.8867,
+      "step": 17111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.856120456522394e-09,
+      "loss": 0.7695,
+      "step": 17112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.804190060040513e-09,
+      "loss": 0.8711,
+      "step": 17113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.752611645048143e-09,
+      "loss": 0.8516,
+      "step": 17114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.701385213360498e-09,
+      "loss": 1.0156,
+      "step": 17115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6505107667805795e-09,
+      "loss": 0.8242,
+      "step": 17116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5999883071013985e-09,
+      "loss": 0.8242,
+      "step": 17117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.549817836100422e-09,
+      "loss": 0.9453,
+      "step": 17118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.499999355545125e-09,
+      "loss": 0.8945,
+      "step": 17119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.45053286718966e-09,
+      "loss": 0.8047,
+      "step": 17120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4014183727748564e-09,
+      "loss": 0.7891,
+      "step": 17121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3526558740315516e-09,
+      "loss": 0.8125,
+      "step": 17122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.304245372675041e-09,
+      "loss": 0.8789,
+      "step": 17123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.256186870411737e-09,
+      "loss": 0.918,
+      "step": 17124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2084803689325096e-09,
+      "loss": 0.8945,
+      "step": 17125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1611258699182355e-09,
+      "loss": 0.8164,
+      "step": 17126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1141233750353606e-09,
+      "loss": 0.8164,
+      "step": 17127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0674728859392267e-09,
+      "loss": 0.7812,
+      "step": 17128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.021174404272964e-09,
+      "loss": 0.875,
+      "step": 17129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.97522793166638e-09,
+      "loss": 0.7578,
+      "step": 17130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9296334697381803e-09,
+      "loss": 0.8594,
+      "step": 17131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.884391020091526e-09,
+      "loss": 1.0547,
+      "step": 17132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8395005843218083e-09,
+      "loss": 0.8945,
+      "step": 17133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7949621640088743e-09,
+      "loss": 0.7617,
+      "step": 17134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.750775760720359e-09,
+      "loss": 0.8125,
+      "step": 17135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7069413760139052e-09,
+      "loss": 0.8984,
+      "step": 17136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6634590114305027e-09,
+      "loss": 0.8828,
+      "step": 17137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.620328668503369e-09,
+      "loss": 0.8711,
+      "step": 17138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5775503487501795e-09,
+      "loss": 1.0078,
+      "step": 17139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5351240536775067e-09,
+      "loss": 0.8281,
+      "step": 17140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4930497847786005e-09,
+      "loss": 0.8359,
+      "step": 17141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.451327543536719e-09,
+      "loss": 0.7852,
+      "step": 17142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4099573314184664e-09,
+      "loss": 0.7734,
+      "step": 17143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3689391498826765e-09,
+      "loss": 0.7617,
+      "step": 17144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3282730003726382e-09,
+      "loss": 0.8789,
+      "step": 17145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.28795888432054e-09,
+      "loss": 0.8359,
+      "step": 17146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.247996803145247e-09,
+      "loss": 0.7266,
+      "step": 17147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2083867582556317e-09,
+      "loss": 0.7773,
+      "step": 17148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.169128751043914e-09,
+      "loss": 0.832,
+      "step": 17149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.130222782894542e-09,
+      "loss": 0.8086,
+      "step": 17150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0916688551764207e-09,
+      "loss": 0.9492,
+      "step": 17151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0534669692484633e-09,
+      "loss": 0.9062,
+      "step": 17152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.015617126452929e-09,
+      "loss": 0.8125,
+      "step": 17153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9781193281254163e-09,
+      "loss": 0.9922,
+      "step": 17154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.94097357558487e-09,
+      "loss": 0.9141,
+      "step": 17155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9041798701402435e-09,
+      "loss": 0.7734,
+      "step": 17156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8677382130860567e-09,
+      "loss": 0.9023,
+      "step": 17157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.831648605705727e-09,
+      "loss": 0.7617,
+      "step": 17158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7959110492704602e-09,
+      "loss": 0.7773,
+      "step": 17159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7605255450381386e-09,
+      "loss": 0.7969,
+      "step": 17160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7254920942555432e-09,
+      "loss": 0.9492,
+      "step": 17161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6908106981550209e-09,
+      "loss": 0.8164,
+      "step": 17162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6564813579578175e-09,
+      "loss": 0.8203,
+      "step": 17163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6225040748740762e-09,
+      "loss": 0.7969,
+      "step": 17164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5888788500995068e-09,
+      "loss": 0.8555,
+      "step": 17165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5556056848176072e-09,
+      "loss": 0.8359,
+      "step": 17166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5226845802007728e-09,
+      "loss": 0.7695,
+      "step": 17167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4901155374080767e-09,
+      "loss": 0.8594,
+      "step": 17168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4578985575852688e-09,
+      "loss": 0.8242,
+      "step": 17169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4260336418681076e-09,
+      "loss": 0.7617,
+      "step": 17170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.394520791377918e-09,
+      "loss": 0.8281,
+      "step": 17171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3633600072249232e-09,
+      "loss": 0.8086,
+      "step": 17172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3325512905049132e-09,
+      "loss": 0.8477,
+      "step": 17173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3020946423047965e-09,
+      "loss": 0.8047,
+      "step": 17174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.271990063694828e-09,
+      "loss": 0.9219,
+      "step": 17175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2422375557363808e-09,
+      "loss": 0.8281,
+      "step": 17176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2128371194775057e-09,
+      "loss": 0.8594,
+      "step": 17177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1837887559518202e-09,
+      "loss": 0.8828,
+      "step": 17178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1550924661840601e-09,
+      "loss": 0.8711,
+      "step": 17179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.126748251182308e-09,
+      "loss": 0.7461,
+      "step": 17180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.098756111946875e-09,
+      "loss": 1.0078,
+      "step": 17181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0711160494625283e-09,
+      "loss": 0.9609,
+      "step": 17182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.043828064701824e-09,
+      "loss": 0.9219,
+      "step": 17183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.016892158627325e-09,
+      "loss": 0.8594,
+      "step": 17184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.90308332186052e-10,
+      "loss": 0.8633,
+      "step": 17185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.640765863139223e-10,
+      "loss": 0.8047,
+      "step": 17186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.381969219357522e-10,
+      "loss": 0.9102,
+      "step": 17187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.126693399619246e-10,
+      "loss": 0.8828,
+      "step": 17188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.874938412906097e-10,
+      "loss": 0.75,
+      "step": 17189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.626704268110963e-10,
+      "loss": 0.7109,
+      "step": 17190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.381990973937992e-10,
+      "loss": 0.6914,
+      "step": 17191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.140798539035821e-10,
+      "loss": 0.7656,
+      "step": 17192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.903126971875452e-10,
+      "loss": 0.918,
+      "step": 17193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.668976280839069e-10,
+      "loss": 0.7578,
+      "step": 17194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.438346474153424e-10,
+      "loss": 0.8984,
+      "step": 17195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.211237559967554e-10,
+      "loss": 0.8398,
+      "step": 17196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.987649546252862e-10,
+      "loss": 0.9141,
+      "step": 17197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.767582440891929e-10,
+      "loss": 0.9453,
+      "step": 17198
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.551036251634113e-10,
+      "loss": 0.7734,
+      "step": 17199
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.338010986106646e-10,
+      "loss": 0.7539,
+      "step": 17200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.128506651814636e-10,
+      "loss": 0.8281,
+      "step": 17201
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.922523256118862e-10,
+      "loss": 0.8594,
+      "step": 17202
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.720060806280181e-10,
+      "loss": 0.7305,
+      "step": 17203
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.521119309426226e-10,
+      "loss": 0.7812,
+      "step": 17204
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.325698772573606e-10,
+      "loss": 0.8906,
+      "step": 17205
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.133799202594603e-10,
+      "loss": 0.875,
+      "step": 17206
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.945420606239371e-10,
+      "loss": 0.75,
+      "step": 17207
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.760562990147044e-10,
+      "loss": 0.832,
+      "step": 17208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.579226360834632e-10,
+      "loss": 0.8672,
+      "step": 17209
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.401410724674815e-10,
+      "loss": 0.8438,
+      "step": 17210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2271160879403526e-10,
+      "loss": 0.8789,
+      "step": 17211
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0563424567596767e-10,
+      "loss": 0.8516,
+      "step": 17212
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.889089837139093e-10,
+      "loss": 0.8359,
+      "step": 17213
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.725358234984988e-10,
+      "loss": 0.8594,
+      "step": 17214
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5651476560594203e-10,
+      "loss": 0.8359,
+      "step": 17215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.408458105991219e-10,
+      "loss": 1.0078,
+      "step": 17216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.255289590309296e-10,
+      "loss": 0.8594,
+      "step": 17217
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1056421144093353e-10,
+      "loss": 0.8203,
+      "step": 17218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9595156835426907e-10,
+      "loss": 0.8242,
+      "step": 17219
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8169103028719005e-10,
+      "loss": 0.8047,
+      "step": 17220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6778259774151717e-10,
+      "loss": 0.8164,
+      "step": 17221
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5422627120685883e-10,
+      "loss": 0.8203,
+      "step": 17222
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4102205115950073e-10,
+      "loss": 0.9258,
+      "step": 17223
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2816993806684674e-10,
+      "loss": 0.8398,
+      "step": 17224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1566993237853716e-10,
+      "loss": 0.9102,
+      "step": 17225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0352203453644083e-10,
+      "loss": 0.7383,
+      "step": 17226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9172624496799353e-10,
+      "loss": 0.8125,
+      "step": 17227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8028256408952894e-10,
+      "loss": 1.0234,
+      "step": 17228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6919099230183755e-10,
+      "loss": 0.8203,
+      "step": 17229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5845152999682812e-10,
+      "loss": 0.832,
+      "step": 17230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4806417755308667e-10,
+      "loss": 0.8477,
+      "step": 17231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3802893533476635e-10,
+      "loss": 0.793,
+      "step": 17232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2834580369713856e-10,
+      "loss": 0.8125,
+      "step": 17233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1901478297993153e-10,
+      "loss": 0.7031,
+      "step": 17234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1003587351177125e-10,
+      "loss": 0.9023,
+      "step": 17235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0140907560907131e-10,
+      "loss": 0.8398,
+      "step": 17236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.31343895749226e-11,
+      "loss": 0.8516,
+      "step": 17237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.521181570242398e-11,
+      "loss": 0.7383,
+      "step": 17238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.764135426913122e-11,
+      "loss": 0.8086,
+      "step": 17239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.042300554149784e-11,
+      "loss": 0.793,
+      "step": 17240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.355676977487513e-11,
+      "loss": 0.9141,
+      "step": 17241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.704264721018149e-11,
+      "loss": 0.8516,
+      "step": 17242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.088063807612287e-11,
+      "loss": 0.8711,
+      "step": 17243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5070742591413195e-11,
+      "loss": 0.9453,
+      "step": 17244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.961296095811307e-11,
+      "loss": 0.8203,
+      "step": 17245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.450729337162173e-11,
+      "loss": 0.8633,
+      "step": 17246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.975374000957487e-11,
+      "loss": 0.957,
+      "step": 17247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5352301039616166e-11,
+      "loss": 0.8359,
+      "step": 17248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1302976616066616e-11,
+      "loss": 0.7656,
+      "step": 17249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7605766883255215e-11,
+      "loss": 0.9609,
+      "step": 17250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4260671971078054e-11,
+      "loss": 0.8203,
+      "step": 17251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1267691996108554e-11,
+      "loss": 0.8828,
+      "step": 17252
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.6268270638179e-12,
+      "loss": 0.7266,
+      "step": 17253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.338077268575049e-12,
+      "loss": 0.9453,
+      "step": 17254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.401442689205837e-12,
+      "loss": 0.7617,
+      "step": 17255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.816923395654314e-12,
+      "loss": 0.8555,
+      "step": 17256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5845194423214083e-12,
+      "loss": 0.8086,
+      "step": 17257
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.042308736160408e-13,
+      "loss": 0.7773,
+      "step": 17258
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7605771951423323e-13,
+      "loss": 0.9219,
+      "step": 17259
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8438,
+      "step": 17260
+    },
+    {
+      "epoch": 1.0,
+      "step": 17260,
+      "total_flos": 8419693899546624.0,
+      "train_loss": 0.8945484501738122,
+      "train_runtime": 102586.6031,
+      "train_samples_per_second": 21.536,
+      "train_steps_per_second": 0.168
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 17260,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 8419693899546624.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}